{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9836065573770494,
  "eval_steps": 500,
  "global_step": 426,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00702576112412178,
      "grad_norm": 0.18053510785102844,
      "learning_rate": 2.3255813953488374e-07,
      "loss": 0.7704,
      "step": 1
    },
    {
      "epoch": 0.01405152224824356,
      "grad_norm": 0.18344072997570038,
      "learning_rate": 4.651162790697675e-07,
      "loss": 0.7172,
      "step": 2
    },
    {
      "epoch": 0.02107728337236534,
      "grad_norm": 0.17655028402805328,
      "learning_rate": 6.976744186046513e-07,
      "loss": 0.7562,
      "step": 3
    },
    {
      "epoch": 0.02810304449648712,
      "grad_norm": 0.1985616534948349,
      "learning_rate": 9.30232558139535e-07,
      "loss": 0.8166,
      "step": 4
    },
    {
      "epoch": 0.0351288056206089,
      "grad_norm": 0.14133895933628082,
      "learning_rate": 1.1627906976744188e-06,
      "loss": 0.668,
      "step": 5
    },
    {
      "epoch": 0.04215456674473068,
      "grad_norm": 0.17750921845436096,
      "learning_rate": 1.3953488372093025e-06,
      "loss": 0.7862,
      "step": 6
    },
    {
      "epoch": 0.04918032786885246,
      "grad_norm": 0.1687806397676468,
      "learning_rate": 1.6279069767441862e-06,
      "loss": 0.7697,
      "step": 7
    },
    {
      "epoch": 0.05620608899297424,
      "grad_norm": 0.14865685999393463,
      "learning_rate": 1.86046511627907e-06,
      "loss": 0.6865,
      "step": 8
    },
    {
      "epoch": 0.06323185011709602,
      "grad_norm": 0.15419331192970276,
      "learning_rate": 2.0930232558139536e-06,
      "loss": 0.7145,
      "step": 9
    },
    {
      "epoch": 0.0702576112412178,
      "grad_norm": 0.1144702211022377,
      "learning_rate": 2.3255813953488376e-06,
      "loss": 0.6761,
      "step": 10
    },
    {
      "epoch": 0.07728337236533958,
      "grad_norm": 0.12051534652709961,
      "learning_rate": 2.558139534883721e-06,
      "loss": 0.7301,
      "step": 11
    },
    {
      "epoch": 0.08430913348946135,
      "grad_norm": 0.11249984055757523,
      "learning_rate": 2.790697674418605e-06,
      "loss": 0.7126,
      "step": 12
    },
    {
      "epoch": 0.09133489461358314,
      "grad_norm": 0.11579480022192001,
      "learning_rate": 3.0232558139534885e-06,
      "loss": 0.6869,
      "step": 13
    },
    {
      "epoch": 0.09836065573770492,
      "grad_norm": 0.08192656189203262,
      "learning_rate": 3.2558139534883724e-06,
      "loss": 0.6698,
      "step": 14
    },
    {
      "epoch": 0.1053864168618267,
      "grad_norm": 0.08397164940834045,
      "learning_rate": 3.4883720930232564e-06,
      "loss": 0.6257,
      "step": 15
    },
    {
      "epoch": 0.11241217798594848,
      "grad_norm": 0.07774341851472855,
      "learning_rate": 3.72093023255814e-06,
      "loss": 0.5955,
      "step": 16
    },
    {
      "epoch": 0.11943793911007025,
      "grad_norm": 0.08383660763502121,
      "learning_rate": 3.953488372093024e-06,
      "loss": 0.5698,
      "step": 17
    },
    {
      "epoch": 0.12646370023419204,
      "grad_norm": 0.08330123126506805,
      "learning_rate": 4.186046511627907e-06,
      "loss": 0.6513,
      "step": 18
    },
    {
      "epoch": 0.13348946135831383,
      "grad_norm": 0.07081824541091919,
      "learning_rate": 4.418604651162791e-06,
      "loss": 0.5878,
      "step": 19
    },
    {
      "epoch": 0.1405152224824356,
      "grad_norm": 0.09581199288368225,
      "learning_rate": 4.651162790697675e-06,
      "loss": 0.6923,
      "step": 20
    },
    {
      "epoch": 0.14754098360655737,
      "grad_norm": 0.06888371706008911,
      "learning_rate": 4.883720930232559e-06,
      "loss": 0.5683,
      "step": 21
    },
    {
      "epoch": 0.15456674473067916,
      "grad_norm": 0.07578740268945694,
      "learning_rate": 5.116279069767442e-06,
      "loss": 0.5844,
      "step": 22
    },
    {
      "epoch": 0.16159250585480095,
      "grad_norm": 0.0799824595451355,
      "learning_rate": 5.348837209302326e-06,
      "loss": 0.6688,
      "step": 23
    },
    {
      "epoch": 0.1686182669789227,
      "grad_norm": 0.06833522021770477,
      "learning_rate": 5.58139534883721e-06,
      "loss": 0.5623,
      "step": 24
    },
    {
      "epoch": 0.1756440281030445,
      "grad_norm": 0.07280978560447693,
      "learning_rate": 5.8139534883720935e-06,
      "loss": 0.6319,
      "step": 25
    },
    {
      "epoch": 0.18266978922716628,
      "grad_norm": 0.06204582750797272,
      "learning_rate": 6.046511627906977e-06,
      "loss": 0.563,
      "step": 26
    },
    {
      "epoch": 0.18969555035128804,
      "grad_norm": 0.06134812906384468,
      "learning_rate": 6.279069767441861e-06,
      "loss": 0.5847,
      "step": 27
    },
    {
      "epoch": 0.19672131147540983,
      "grad_norm": 0.05748824402689934,
      "learning_rate": 6.511627906976745e-06,
      "loss": 0.5311,
      "step": 28
    },
    {
      "epoch": 0.20374707259953162,
      "grad_norm": 0.06215423345565796,
      "learning_rate": 6.744186046511628e-06,
      "loss": 0.5514,
      "step": 29
    },
    {
      "epoch": 0.2107728337236534,
      "grad_norm": 0.06185278296470642,
      "learning_rate": 6.976744186046513e-06,
      "loss": 0.5563,
      "step": 30
    },
    {
      "epoch": 0.21779859484777517,
      "grad_norm": 0.05966101959347725,
      "learning_rate": 7.209302325581395e-06,
      "loss": 0.5758,
      "step": 31
    },
    {
      "epoch": 0.22482435597189696,
      "grad_norm": 0.06163804978132248,
      "learning_rate": 7.44186046511628e-06,
      "loss": 0.5063,
      "step": 32
    },
    {
      "epoch": 0.23185011709601874,
      "grad_norm": 0.06333377212285995,
      "learning_rate": 7.674418604651164e-06,
      "loss": 0.5583,
      "step": 33
    },
    {
      "epoch": 0.2388758782201405,
      "grad_norm": 0.07251594960689545,
      "learning_rate": 7.906976744186048e-06,
      "loss": 0.5886,
      "step": 34
    },
    {
      "epoch": 0.2459016393442623,
      "grad_norm": 0.0646405890583992,
      "learning_rate": 8.139534883720931e-06,
      "loss": 0.5421,
      "step": 35
    },
    {
      "epoch": 0.2529274004683841,
      "grad_norm": 0.08668368309736252,
      "learning_rate": 8.372093023255815e-06,
      "loss": 0.5354,
      "step": 36
    },
    {
      "epoch": 0.25995316159250587,
      "grad_norm": 0.06333053857088089,
      "learning_rate": 8.604651162790698e-06,
      "loss": 0.5425,
      "step": 37
    },
    {
      "epoch": 0.26697892271662765,
      "grad_norm": 0.053554244339466095,
      "learning_rate": 8.837209302325582e-06,
      "loss": 0.4706,
      "step": 38
    },
    {
      "epoch": 0.27400468384074944,
      "grad_norm": 0.05727384239435196,
      "learning_rate": 9.069767441860465e-06,
      "loss": 0.4781,
      "step": 39
    },
    {
      "epoch": 0.2810304449648712,
      "grad_norm": 0.05613735318183899,
      "learning_rate": 9.30232558139535e-06,
      "loss": 0.4938,
      "step": 40
    },
    {
      "epoch": 0.28805620608899296,
      "grad_norm": 0.05665242299437523,
      "learning_rate": 9.534883720930234e-06,
      "loss": 0.516,
      "step": 41
    },
    {
      "epoch": 0.29508196721311475,
      "grad_norm": 0.06407799571752548,
      "learning_rate": 9.767441860465117e-06,
      "loss": 0.5219,
      "step": 42
    },
    {
      "epoch": 0.30210772833723654,
      "grad_norm": 0.059850167483091354,
      "learning_rate": 1e-05,
      "loss": 0.5281,
      "step": 43
    },
    {
      "epoch": 0.3091334894613583,
      "grad_norm": 0.058536890894174576,
      "learning_rate": 9.99983179466314e-06,
      "loss": 0.4646,
      "step": 44
    },
    {
      "epoch": 0.3161592505854801,
      "grad_norm": 0.053681179881095886,
      "learning_rate": 9.999327189969768e-06,
      "loss": 0.4662,
      "step": 45
    },
    {
      "epoch": 0.3231850117096019,
      "grad_norm": 0.05520308017730713,
      "learning_rate": 9.998486219870769e-06,
      "loss": 0.4972,
      "step": 46
    },
    {
      "epoch": 0.33021077283372363,
      "grad_norm": 0.06100783869624138,
      "learning_rate": 9.997308940948405e-06,
      "loss": 0.5449,
      "step": 47
    },
    {
      "epoch": 0.3372365339578454,
      "grad_norm": 0.05527447536587715,
      "learning_rate": 9.995795432412513e-06,
      "loss": 0.517,
      "step": 48
    },
    {
      "epoch": 0.3442622950819672,
      "grad_norm": 0.06016694754362106,
      "learning_rate": 9.993945796095183e-06,
      "loss": 0.4996,
      "step": 49
    },
    {
      "epoch": 0.351288056206089,
      "grad_norm": 0.061447110027074814,
      "learning_rate": 9.991760156443892e-06,
      "loss": 0.4678,
      "step": 50
    },
    {
      "epoch": 0.3583138173302108,
      "grad_norm": 0.05987882614135742,
      "learning_rate": 9.989238660513141e-06,
      "loss": 0.5699,
      "step": 51
    },
    {
      "epoch": 0.36533957845433257,
      "grad_norm": 0.05668003484606743,
      "learning_rate": 9.98638147795456e-06,
      "loss": 0.505,
      "step": 52
    },
    {
      "epoch": 0.37236533957845436,
      "grad_norm": 0.05434773117303848,
      "learning_rate": 9.983188801005492e-06,
      "loss": 0.4742,
      "step": 53
    },
    {
      "epoch": 0.3793911007025761,
      "grad_norm": 0.05131086707115173,
      "learning_rate": 9.979660844476056e-06,
      "loss": 0.5096,
      "step": 54
    },
    {
      "epoch": 0.3864168618266979,
      "grad_norm": 0.058555662631988525,
      "learning_rate": 9.975797845734699e-06,
      "loss": 0.5277,
      "step": 55
    },
    {
      "epoch": 0.39344262295081966,
      "grad_norm": 0.0570974200963974,
      "learning_rate": 9.971600064692222e-06,
      "loss": 0.4801,
      "step": 56
    },
    {
      "epoch": 0.40046838407494145,
      "grad_norm": 0.056893255561590195,
      "learning_rate": 9.967067783784297e-06,
      "loss": 0.4857,
      "step": 57
    },
    {
      "epoch": 0.40749414519906324,
      "grad_norm": 0.05847252905368805,
      "learning_rate": 9.962201307952455e-06,
      "loss": 0.5099,
      "step": 58
    },
    {
      "epoch": 0.41451990632318503,
      "grad_norm": 0.06154610961675644,
      "learning_rate": 9.957000964623585e-06,
      "loss": 0.5099,
      "step": 59
    },
    {
      "epoch": 0.4215456674473068,
      "grad_norm": 0.05544036626815796,
      "learning_rate": 9.951467103687879e-06,
      "loss": 0.4713,
      "step": 60
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.04792679101228714,
      "learning_rate": 9.945600097475322e-06,
      "loss": 0.4576,
      "step": 61
    },
    {
      "epoch": 0.43559718969555034,
      "grad_norm": 0.05695069581270218,
      "learning_rate": 9.939400340730611e-06,
      "loss": 0.5169,
      "step": 62
    },
    {
      "epoch": 0.4426229508196721,
      "grad_norm": 0.049334120005369186,
      "learning_rate": 9.932868250586619e-06,
      "loss": 0.4828,
      "step": 63
    },
    {
      "epoch": 0.4496487119437939,
      "grad_norm": 0.045351915061473846,
      "learning_rate": 9.926004266536314e-06,
      "loss": 0.4305,
      "step": 64
    },
    {
      "epoch": 0.4566744730679157,
      "grad_norm": 0.049619413912296295,
      "learning_rate": 9.918808850403192e-06,
      "loss": 0.4892,
      "step": 65
    },
    {
      "epoch": 0.4637002341920375,
      "grad_norm": 0.05420510098338127,
      "learning_rate": 9.911282486310214e-06,
      "loss": 0.4722,
      "step": 66
    },
    {
      "epoch": 0.4707259953161593,
      "grad_norm": 0.06276430934667587,
      "learning_rate": 9.903425680647225e-06,
      "loss": 0.5461,
      "step": 67
    },
    {
      "epoch": 0.477751756440281,
      "grad_norm": 0.06031210348010063,
      "learning_rate": 9.895238962036878e-06,
      "loss": 0.426,
      "step": 68
    },
    {
      "epoch": 0.4847775175644028,
      "grad_norm": 0.05641568452119827,
      "learning_rate": 9.88672288129908e-06,
      "loss": 0.4954,
      "step": 69
    },
    {
      "epoch": 0.4918032786885246,
      "grad_norm": 0.05691906809806824,
      "learning_rate": 9.877878011413924e-06,
      "loss": 0.4511,
      "step": 70
    },
    {
      "epoch": 0.49882903981264637,
      "grad_norm": 0.06182294711470604,
      "learning_rate": 9.868704947483134e-06,
      "loss": 0.5105,
      "step": 71
    },
    {
      "epoch": 0.5058548009367682,
      "grad_norm": 0.053402017802000046,
      "learning_rate": 9.859204306690038e-06,
      "loss": 0.4792,
      "step": 72
    },
    {
      "epoch": 0.5128805620608899,
      "grad_norm": 0.08146236836910248,
      "learning_rate": 9.849376728258024e-06,
      "loss": 0.496,
      "step": 73
    },
    {
      "epoch": 0.5199063231850117,
      "grad_norm": 0.051524728536605835,
      "learning_rate": 9.839222873407553e-06,
      "loss": 0.4674,
      "step": 74
    },
    {
      "epoch": 0.5269320843091335,
      "grad_norm": 0.059568360447883606,
      "learning_rate": 9.828743425311654e-06,
      "loss": 0.4675,
      "step": 75
    },
    {
      "epoch": 0.5339578454332553,
      "grad_norm": 0.05749664828181267,
      "learning_rate": 9.817939089049964e-06,
      "loss": 0.449,
      "step": 76
    },
    {
      "epoch": 0.5409836065573771,
      "grad_norm": 0.051995862275362015,
      "learning_rate": 9.806810591561295e-06,
      "loss": 0.4323,
      "step": 77
    },
    {
      "epoch": 0.5480093676814989,
      "grad_norm": 0.051814232021570206,
      "learning_rate": 9.795358681594712e-06,
      "loss": 0.4724,
      "step": 78
    },
    {
      "epoch": 0.5550351288056206,
      "grad_norm": 0.056114014238119125,
      "learning_rate": 9.783584129659162e-06,
      "loss": 0.4983,
      "step": 79
    },
    {
      "epoch": 0.5620608899297423,
      "grad_norm": 0.06239385902881622,
      "learning_rate": 9.771487727971642e-06,
      "loss": 0.4808,
      "step": 80
    },
    {
      "epoch": 0.5690866510538641,
      "grad_norm": 0.054439231753349304,
      "learning_rate": 9.759070290403873e-06,
      "loss": 0.4642,
      "step": 81
    },
    {
      "epoch": 0.5761124121779859,
      "grad_norm": 0.05044344440102577,
      "learning_rate": 9.746332652427566e-06,
      "loss": 0.4654,
      "step": 82
    },
    {
      "epoch": 0.5831381733021077,
      "grad_norm": 0.06456462293863297,
      "learning_rate": 9.733275671058195e-06,
      "loss": 0.5002,
      "step": 83
    },
    {
      "epoch": 0.5901639344262295,
      "grad_norm": 0.05796905234456062,
      "learning_rate": 9.71990022479734e-06,
      "loss": 0.4735,
      "step": 84
    },
    {
      "epoch": 0.5971896955503513,
      "grad_norm": 0.055456362664699554,
      "learning_rate": 9.70620721357358e-06,
      "loss": 0.4612,
      "step": 85
    },
    {
      "epoch": 0.6042154566744731,
      "grad_norm": 0.07052718102931976,
      "learning_rate": 9.69219755868194e-06,
      "loss": 0.4645,
      "step": 86
    },
    {
      "epoch": 0.6112412177985949,
      "grad_norm": 0.057100776582956314,
      "learning_rate": 9.677872202721906e-06,
      "loss": 0.4949,
      "step": 87
    },
    {
      "epoch": 0.6182669789227166,
      "grad_norm": 0.056776516139507294,
      "learning_rate": 9.663232109534011e-06,
      "loss": 0.5361,
      "step": 88
    },
    {
      "epoch": 0.6252927400468384,
      "grad_norm": 0.05490226671099663,
      "learning_rate": 9.648278264134977e-06,
      "loss": 0.441,
      "step": 89
    },
    {
      "epoch": 0.6323185011709602,
      "grad_norm": 0.0555921345949173,
      "learning_rate": 9.633011672651443e-06,
      "loss": 0.4116,
      "step": 90
    },
    {
      "epoch": 0.639344262295082,
      "grad_norm": 0.053403131663799286,
      "learning_rate": 9.617433362252277e-06,
      "loss": 0.482,
      "step": 91
    },
    {
      "epoch": 0.6463700234192038,
      "grad_norm": 0.04735114425420761,
      "learning_rate": 9.601544381079457e-06,
      "loss": 0.4476,
      "step": 92
    },
    {
      "epoch": 0.6533957845433255,
      "grad_norm": 0.051030732691287994,
      "learning_rate": 9.585345798177557e-06,
      "loss": 0.4997,
      "step": 93
    },
    {
      "epoch": 0.6604215456674473,
      "grad_norm": 0.054940689355134964,
      "learning_rate": 9.56883870342181e-06,
      "loss": 0.4406,
      "step": 94
    },
    {
      "epoch": 0.667447306791569,
      "grad_norm": 0.06611790508031845,
      "learning_rate": 9.552024207444794e-06,
      "loss": 0.4291,
      "step": 95
    },
    {
      "epoch": 0.6744730679156908,
      "grad_norm": 0.06092395260930061,
      "learning_rate": 9.534903441561693e-06,
      "loss": 0.5012,
      "step": 96
    },
    {
      "epoch": 0.6814988290398126,
      "grad_norm": 0.0560709647834301,
      "learning_rate": 9.517477557694182e-06,
      "loss": 0.5025,
      "step": 97
    },
    {
      "epoch": 0.6885245901639344,
      "grad_norm": 0.06219569593667984,
      "learning_rate": 9.499747728292928e-06,
      "loss": 0.5078,
      "step": 98
    },
    {
      "epoch": 0.6955503512880562,
      "grad_norm": 0.05389304459095001,
      "learning_rate": 9.481715146258699e-06,
      "loss": 0.4275,
      "step": 99
    },
    {
      "epoch": 0.702576112412178,
      "grad_norm": 0.055709365755319595,
      "learning_rate": 9.463381024862116e-06,
      "loss": 0.5152,
      "step": 100
    },
    {
      "epoch": 0.7096018735362998,
      "grad_norm": 0.07741156220436096,
      "learning_rate": 9.444746597662e-06,
      "loss": 0.504,
      "step": 101
    },
    {
      "epoch": 0.7166276346604216,
      "grad_norm": 0.049752719700336456,
      "learning_rate": 9.425813118422393e-06,
      "loss": 0.4613,
      "step": 102
    },
    {
      "epoch": 0.7236533957845434,
      "grad_norm": 0.056723710149526596,
      "learning_rate": 9.406581861028199e-06,
      "loss": 0.5293,
      "step": 103
    },
    {
      "epoch": 0.7306791569086651,
      "grad_norm": 0.05891204625368118,
      "learning_rate": 9.387054119399466e-06,
      "loss": 0.4229,
      "step": 104
    },
    {
      "epoch": 0.7377049180327869,
      "grad_norm": 0.053549159318208694,
      "learning_rate": 9.36723120740434e-06,
      "loss": 0.4525,
      "step": 105
    },
    {
      "epoch": 0.7447306791569087,
      "grad_norm": 0.05781884491443634,
      "learning_rate": 9.347114458770656e-06,
      "loss": 0.4756,
      "step": 106
    },
    {
      "epoch": 0.7517564402810304,
      "grad_norm": 0.052228234708309174,
      "learning_rate": 9.326705226996207e-06,
      "loss": 0.4577,
      "step": 107
    },
    {
      "epoch": 0.7587822014051522,
      "grad_norm": 0.05756888911128044,
      "learning_rate": 9.306004885257675e-06,
      "loss": 0.4604,
      "step": 108
    },
    {
      "epoch": 0.765807962529274,
      "grad_norm": 0.07348482310771942,
      "learning_rate": 9.28501482631824e-06,
      "loss": 0.4768,
      "step": 109
    },
    {
      "epoch": 0.7728337236533958,
      "grad_norm": 0.060214102268218994,
      "learning_rate": 9.26373646243388e-06,
      "loss": 0.4891,
      "step": 110
    },
    {
      "epoch": 0.7798594847775175,
      "grad_norm": 0.05229140818119049,
      "learning_rate": 9.242171225258336e-06,
      "loss": 0.4719,
      "step": 111
    },
    {
      "epoch": 0.7868852459016393,
      "grad_norm": 0.053298093378543854,
      "learning_rate": 9.220320565746806e-06,
      "loss": 0.4449,
      "step": 112
    },
    {
      "epoch": 0.7939110070257611,
      "grad_norm": 0.04904168099164963,
      "learning_rate": 9.198185954058305e-06,
      "loss": 0.4682,
      "step": 113
    },
    {
      "epoch": 0.8009367681498829,
      "grad_norm": 0.055706869810819626,
      "learning_rate": 9.175768879456759e-06,
      "loss": 0.4522,
      "step": 114
    },
    {
      "epoch": 0.8079625292740047,
      "grad_norm": 0.057728879153728485,
      "learning_rate": 9.153070850210803e-06,
      "loss": 0.4465,
      "step": 115
    },
    {
      "epoch": 0.8149882903981265,
      "grad_norm": 0.04915739968419075,
      "learning_rate": 9.130093393492302e-06,
      "loss": 0.4254,
      "step": 116
    },
    {
      "epoch": 0.8220140515222483,
      "grad_norm": 0.04908665642142296,
      "learning_rate": 9.106838055273589e-06,
      "loss": 0.4492,
      "step": 117
    },
    {
      "epoch": 0.8290398126463701,
      "grad_norm": 0.05924678593873978,
      "learning_rate": 9.083306400223465e-06,
      "loss": 0.4821,
      "step": 118
    },
    {
      "epoch": 0.8360655737704918,
      "grad_norm": 0.05097891017794609,
      "learning_rate": 9.059500011601919e-06,
      "loss": 0.4454,
      "step": 119
    },
    {
      "epoch": 0.8430913348946136,
      "grad_norm": 0.05687716603279114,
      "learning_rate": 9.035420491153596e-06,
      "loss": 0.4329,
      "step": 120
    },
    {
      "epoch": 0.8501170960187353,
      "grad_norm": 0.0736871287226677,
      "learning_rate": 9.011069459000035e-06,
      "loss": 0.4685,
      "step": 121
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.05752350762486458,
      "learning_rate": 8.986448553530665e-06,
      "loss": 0.4282,
      "step": 122
    },
    {
      "epoch": 0.8641686182669789,
      "grad_norm": 0.05238118767738342,
      "learning_rate": 8.961559431292562e-06,
      "loss": 0.4381,
      "step": 123
    },
    {
      "epoch": 0.8711943793911007,
      "grad_norm": 0.055724114179611206,
      "learning_rate": 8.936403766879003e-06,
      "loss": 0.4239,
      "step": 124
    },
    {
      "epoch": 0.8782201405152225,
      "grad_norm": 0.05202113091945648,
      "learning_rate": 8.910983252816794e-06,
      "loss": 0.4127,
      "step": 125
    },
    {
      "epoch": 0.8852459016393442,
      "grad_norm": 0.05094996467232704,
      "learning_rate": 8.885299599452381e-06,
      "loss": 0.447,
      "step": 126
    },
    {
      "epoch": 0.892271662763466,
      "grad_norm": 0.060567859560251236,
      "learning_rate": 8.859354534836797e-06,
      "loss": 0.4433,
      "step": 127
    },
    {
      "epoch": 0.8992974238875878,
      "grad_norm": 0.05051707103848457,
      "learning_rate": 8.833149804609372e-06,
      "loss": 0.409,
      "step": 128
    },
    {
      "epoch": 0.9063231850117096,
      "grad_norm": 0.05201994627714157,
      "learning_rate": 8.806687171880298e-06,
      "loss": 0.4511,
      "step": 129
    },
    {
      "epoch": 0.9133489461358314,
      "grad_norm": 0.049249548465013504,
      "learning_rate": 8.779968417111991e-06,
      "loss": 0.4365,
      "step": 130
    },
    {
      "epoch": 0.9203747072599532,
      "grad_norm": 0.05699831247329712,
      "learning_rate": 8.752995337999316e-06,
      "loss": 0.5187,
      "step": 131
    },
    {
      "epoch": 0.927400468384075,
      "grad_norm": 0.05579648166894913,
      "learning_rate": 8.725769749348612e-06,
      "loss": 0.4587,
      "step": 132
    },
    {
      "epoch": 0.9344262295081968,
      "grad_norm": 0.05366150662302971,
      "learning_rate": 8.698293482955605e-06,
      "loss": 0.4599,
      "step": 133
    },
    {
      "epoch": 0.9414519906323185,
      "grad_norm": 0.05521006882190704,
      "learning_rate": 8.670568387482153e-06,
      "loss": 0.4634,
      "step": 134
    },
    {
      "epoch": 0.9484777517564403,
      "grad_norm": 0.058564692735672,
      "learning_rate": 8.642596328331864e-06,
      "loss": 0.4546,
      "step": 135
    },
    {
      "epoch": 0.955503512880562,
      "grad_norm": 0.22925680875778198,
      "learning_rate": 8.614379187524593e-06,
      "loss": 0.4392,
      "step": 136
    },
    {
      "epoch": 0.9625292740046838,
      "grad_norm": 0.05385170876979828,
      "learning_rate": 8.585918863569806e-06,
      "loss": 0.4316,
      "step": 137
    },
    {
      "epoch": 0.9695550351288056,
      "grad_norm": 0.05049740523099899,
      "learning_rate": 8.55721727133886e-06,
      "loss": 0.4467,
      "step": 138
    },
    {
      "epoch": 0.9765807962529274,
      "grad_norm": 0.0547189861536026,
      "learning_rate": 8.528276341936146e-06,
      "loss": 0.4707,
      "step": 139
    },
    {
      "epoch": 0.9836065573770492,
      "grad_norm": 0.057287830859422684,
      "learning_rate": 8.499098022569177e-06,
      "loss": 0.4413,
      "step": 140
    },
    {
      "epoch": 0.990632318501171,
      "grad_norm": 0.05672721937298775,
      "learning_rate": 8.469684276417568e-06,
      "loss": 0.485,
      "step": 141
    },
    {
      "epoch": 0.9976580796252927,
      "grad_norm": 0.05014313757419586,
      "learning_rate": 8.440037082500953e-06,
      "loss": 0.441,
      "step": 142
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.05014313757419586,
      "learning_rate": 8.410158435545825e-06,
      "loss": 0.4919,
      "step": 143
    },
    {
      "epoch": 1.0070257611241218,
      "grad_norm": 0.09903538227081299,
      "learning_rate": 8.380050345851338e-06,
      "loss": 0.3345,
      "step": 144
    },
    {
      "epoch": 1.0140515222482436,
      "grad_norm": 0.04598068818449974,
      "learning_rate": 8.349714839154035e-06,
      "loss": 0.3456,
      "step": 145
    },
    {
      "epoch": 1.0210772833723654,
      "grad_norm": 0.05103905871510506,
      "learning_rate": 8.319153956491567e-06,
      "loss": 0.3466,
      "step": 146
    },
    {
      "epoch": 1.0281030444964872,
      "grad_norm": 0.050713758915662766,
      "learning_rate": 8.288369754065362e-06,
      "loss": 0.3337,
      "step": 147
    },
    {
      "epoch": 1.035128805620609,
      "grad_norm": 0.04552711173892021,
      "learning_rate": 8.257364303102275e-06,
      "loss": 0.3678,
      "step": 148
    },
    {
      "epoch": 1.0421545667447307,
      "grad_norm": 0.05452848598361015,
      "learning_rate": 8.226139689715233e-06,
      "loss": 0.3589,
      "step": 149
    },
    {
      "epoch": 1.0491803278688525,
      "grad_norm": 0.048317231237888336,
      "learning_rate": 8.19469801476288e-06,
      "loss": 0.3848,
      "step": 150
    },
    {
      "epoch": 1.0562060889929743,
      "grad_norm": 0.04934811219573021,
      "learning_rate": 8.16304139370823e-06,
      "loss": 0.3017,
      "step": 151
    },
    {
      "epoch": 1.063231850117096,
      "grad_norm": 0.04967602342367172,
      "learning_rate": 8.131171956476328e-06,
      "loss": 0.3647,
      "step": 152
    },
    {
      "epoch": 1.0702576112412179,
      "grad_norm": 0.048318326473236084,
      "learning_rate": 8.09909184731094e-06,
      "loss": 0.3386,
      "step": 153
    },
    {
      "epoch": 1.0772833723653397,
      "grad_norm": 0.06288264691829681,
      "learning_rate": 8.066803224630295e-06,
      "loss": 0.334,
      "step": 154
    },
    {
      "epoch": 1.0843091334894615,
      "grad_norm": 0.05197209119796753,
      "learning_rate": 8.034308260881854e-06,
      "loss": 0.3538,
      "step": 155
    },
    {
      "epoch": 1.0913348946135832,
      "grad_norm": 0.0542137436568737,
      "learning_rate": 8.00160914239615e-06,
      "loss": 0.3296,
      "step": 156
    },
    {
      "epoch": 1.098360655737705,
      "grad_norm": 0.05365780368447304,
      "learning_rate": 7.968708069239672e-06,
      "loss": 0.3932,
      "step": 157
    },
    {
      "epoch": 1.1053864168618266,
      "grad_norm": 0.060223035514354706,
      "learning_rate": 7.935607255066867e-06,
      "loss": 0.3679,
      "step": 158
    },
    {
      "epoch": 1.1124121779859484,
      "grad_norm": 0.05058826506137848,
      "learning_rate": 7.902308926971166e-06,
      "loss": 0.3321,
      "step": 159
    },
    {
      "epoch": 1.1194379391100702,
      "grad_norm": 0.054012175649404526,
      "learning_rate": 7.868815325335168e-06,
      "loss": 0.3501,
      "step": 160
    },
    {
      "epoch": 1.126463700234192,
      "grad_norm": 0.061976704746484756,
      "learning_rate": 7.835128703679896e-06,
      "loss": 0.35,
      "step": 161
    },
    {
      "epoch": 1.1334894613583137,
      "grad_norm": 0.05166299641132355,
      "learning_rate": 7.801251328513164e-06,
      "loss": 0.3521,
      "step": 162
    },
    {
      "epoch": 1.1405152224824355,
      "grad_norm": 0.05794499069452286,
      "learning_rate": 7.767185479177092e-06,
      "loss": 0.3318,
      "step": 163
    },
    {
      "epoch": 1.1475409836065573,
      "grad_norm": 0.05970500782132149,
      "learning_rate": 7.732933447694748e-06,
      "loss": 0.3601,
      "step": 164
    },
    {
      "epoch": 1.154566744730679,
      "grad_norm": 0.04941580072045326,
      "learning_rate": 7.698497538615928e-06,
      "loss": 0.321,
      "step": 165
    },
    {
      "epoch": 1.161592505854801,
      "grad_norm": 0.061307549476623535,
      "learning_rate": 7.663880068862106e-06,
      "loss": 0.3103,
      "step": 166
    },
    {
      "epoch": 1.1686182669789227,
      "grad_norm": 0.055390097200870514,
      "learning_rate": 7.629083367570547e-06,
      "loss": 0.3561,
      "step": 167
    },
    {
      "epoch": 1.1756440281030445,
      "grad_norm": 0.06362055987119675,
      "learning_rate": 7.594109775937595e-06,
      "loss": 0.3477,
      "step": 168
    },
    {
      "epoch": 1.1826697892271663,
      "grad_norm": 0.05179616063833237,
      "learning_rate": 7.558961647061156e-06,
      "loss": 0.3534,
      "step": 169
    },
    {
      "epoch": 1.189695550351288,
      "grad_norm": 0.0547102652490139,
      "learning_rate": 7.5236413457823745e-06,
      "loss": 0.3023,
      "step": 170
    },
    {
      "epoch": 1.1967213114754098,
      "grad_norm": 0.0540938526391983,
      "learning_rate": 7.488151248526518e-06,
      "loss": 0.3144,
      "step": 171
    },
    {
      "epoch": 1.2037470725995316,
      "grad_norm": 0.04804334416985512,
      "learning_rate": 7.452493743143092e-06,
      "loss": 0.317,
      "step": 172
    },
    {
      "epoch": 1.2107728337236534,
      "grad_norm": 0.05935240909457207,
      "learning_rate": 7.416671228745181e-06,
      "loss": 0.3801,
      "step": 173
    },
    {
      "epoch": 1.2177985948477752,
      "grad_norm": 0.05715787410736084,
      "learning_rate": 7.380686115548024e-06,
      "loss": 0.3286,
      "step": 174
    },
    {
      "epoch": 1.224824355971897,
      "grad_norm": 0.05142643675208092,
      "learning_rate": 7.344540824706855e-06,
      "loss": 0.3351,
      "step": 175
    },
    {
      "epoch": 1.2318501170960188,
      "grad_norm": 0.0654173269867897,
      "learning_rate": 7.3082377881540025e-06,
      "loss": 0.344,
      "step": 176
    },
    {
      "epoch": 1.2388758782201406,
      "grad_norm": 0.05274007469415665,
      "learning_rate": 7.271779448435265e-06,
      "loss": 0.3427,
      "step": 177
    },
    {
      "epoch": 1.2459016393442623,
      "grad_norm": 0.05858859047293663,
      "learning_rate": 7.235168258545569e-06,
      "loss": 0.3477,
      "step": 178
    },
    {
      "epoch": 1.2529274004683841,
      "grad_norm": 0.05919253081083298,
      "learning_rate": 7.198406681763925e-06,
      "loss": 0.348,
      "step": 179
    },
    {
      "epoch": 1.259953161592506,
      "grad_norm": 0.05598839744925499,
      "learning_rate": 7.161497191487693e-06,
      "loss": 0.3529,
      "step": 180
    },
    {
      "epoch": 1.2669789227166277,
      "grad_norm": 0.04682192578911781,
      "learning_rate": 7.124442271066174e-06,
      "loss": 0.3346,
      "step": 181
    },
    {
      "epoch": 1.2740046838407495,
      "grad_norm": 0.05666663125157356,
      "learning_rate": 7.087244413633516e-06,
      "loss": 0.3372,
      "step": 182
    },
    {
      "epoch": 1.281030444964871,
      "grad_norm": 0.05192001909017563,
      "learning_rate": 7.049906121940974e-06,
      "loss": 0.3495,
      "step": 183
    },
    {
      "epoch": 1.288056206088993,
      "grad_norm": 0.056901462376117706,
      "learning_rate": 7.012429908188523e-06,
      "loss": 0.3189,
      "step": 184
    },
    {
      "epoch": 1.2950819672131146,
      "grad_norm": 0.05573394522070885,
      "learning_rate": 6.9748182938558225e-06,
      "loss": 0.3074,
      "step": 185
    },
    {
      "epoch": 1.3021077283372366,
      "grad_norm": 0.051766857504844666,
      "learning_rate": 6.937073809532581e-06,
      "loss": 0.3801,
      "step": 186
    },
    {
      "epoch": 1.3091334894613582,
      "grad_norm": 0.0547168143093586,
      "learning_rate": 6.899198994748274e-06,
      "loss": 0.3197,
      "step": 187
    },
    {
      "epoch": 1.3161592505854802,
      "grad_norm": 0.055122897028923035,
      "learning_rate": 6.861196397801297e-06,
      "loss": 0.2924,
      "step": 188
    },
    {
      "epoch": 1.3231850117096018,
      "grad_norm": 0.04852576553821564,
      "learning_rate": 6.823068575587496e-06,
      "loss": 0.3215,
      "step": 189
    },
    {
      "epoch": 1.3302107728337236,
      "grad_norm": 0.04942547157406807,
      "learning_rate": 6.784818093428144e-06,
      "loss": 0.31,
      "step": 190
    },
    {
      "epoch": 1.3372365339578454,
      "grad_norm": 0.05302649736404419,
      "learning_rate": 6.746447524897335e-06,
      "loss": 0.3386,
      "step": 191
    },
    {
      "epoch": 1.3442622950819672,
      "grad_norm": 0.04981255903840065,
      "learning_rate": 6.70795945164883e-06,
      "loss": 0.3429,
      "step": 192
    },
    {
      "epoch": 1.351288056206089,
      "grad_norm": 0.061598002910614014,
      "learning_rate": 6.6693564632423626e-06,
      "loss": 0.3444,
      "step": 193
    },
    {
      "epoch": 1.3583138173302107,
      "grad_norm": 0.06926045566797256,
      "learning_rate": 6.630641156969397e-06,
      "loss": 0.3119,
      "step": 194
    },
    {
      "epoch": 1.3653395784543325,
      "grad_norm": 0.05165793001651764,
      "learning_rate": 6.591816137678388e-06,
      "loss": 0.3309,
      "step": 195
    },
    {
      "epoch": 1.3723653395784543,
      "grad_norm": 0.05770070105791092,
      "learning_rate": 6.552884017599517e-06,
      "loss": 0.3527,
      "step": 196
    },
    {
      "epoch": 1.379391100702576,
      "grad_norm": 0.06011437624692917,
      "learning_rate": 6.513847416168929e-06,
      "loss": 0.3574,
      "step": 197
    },
    {
      "epoch": 1.3864168618266979,
      "grad_norm": 0.05262259393930435,
      "learning_rate": 6.474708959852504e-06,
      "loss": 0.2922,
      "step": 198
    },
    {
      "epoch": 1.3934426229508197,
      "grad_norm": 0.05124557390809059,
      "learning_rate": 6.435471281969133e-06,
      "loss": 0.3053,
      "step": 199
    },
    {
      "epoch": 1.4004683840749415,
      "grad_norm": 0.054529063403606415,
      "learning_rate": 6.396137022513545e-06,
      "loss": 0.3311,
      "step": 200
    },
    {
      "epoch": 1.4074941451990632,
      "grad_norm": 0.05387987568974495,
      "learning_rate": 6.3567088279786885e-06,
      "loss": 0.3508,
      "step": 201
    },
    {
      "epoch": 1.414519906323185,
      "grad_norm": 0.052831970155239105,
      "learning_rate": 6.317189351177657e-06,
      "loss": 0.3425,
      "step": 202
    },
    {
      "epoch": 1.4215456674473068,
      "grad_norm": 0.0602416917681694,
      "learning_rate": 6.277581251065217e-06,
      "loss": 0.3448,
      "step": 203
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.04873733967542648,
      "learning_rate": 6.237887192558894e-06,
      "loss": 0.361,
      "step": 204
    },
    {
      "epoch": 1.4355971896955504,
      "grad_norm": 0.05855415388941765,
      "learning_rate": 6.198109846359682e-06,
      "loss": 0.3324,
      "step": 205
    },
    {
      "epoch": 1.4426229508196722,
      "grad_norm": 0.05669476091861725,
      "learning_rate": 6.15825188877235e-06,
      "loss": 0.3892,
      "step": 206
    },
    {
      "epoch": 1.449648711943794,
      "grad_norm": 0.0551174134016037,
      "learning_rate": 6.118316001525368e-06,
      "loss": 0.3579,
      "step": 207
    },
    {
      "epoch": 1.4566744730679158,
      "grad_norm": 0.05648890882730484,
      "learning_rate": 6.078304871590485e-06,
      "loss": 0.3183,
      "step": 208
    },
    {
      "epoch": 1.4637002341920375,
      "grad_norm": 0.05304768308997154,
      "learning_rate": 6.038221191001935e-06,
      "loss": 0.3346,
      "step": 209
    },
    {
      "epoch": 1.4707259953161593,
      "grad_norm": 0.04858344793319702,
      "learning_rate": 5.998067656675318e-06,
      "loss": 0.3545,
      "step": 210
    },
    {
      "epoch": 1.4777517564402811,
      "grad_norm": 0.060655366629362106,
      "learning_rate": 5.95784697022614e-06,
      "loss": 0.3321,
      "step": 211
    },
    {
      "epoch": 1.4847775175644027,
      "grad_norm": 0.053603459149599075,
      "learning_rate": 5.917561837788046e-06,
      "loss": 0.3308,
      "step": 212
    },
    {
      "epoch": 1.4918032786885247,
      "grad_norm": 0.04807604104280472,
      "learning_rate": 5.877214969830746e-06,
      "loss": 0.2965,
      "step": 213
    },
    {
      "epoch": 1.4988290398126463,
      "grad_norm": 0.05581279471516609,
      "learning_rate": 5.836809080977644e-06,
      "loss": 0.3358,
      "step": 214
    },
    {
      "epoch": 1.5058548009367683,
      "grad_norm": 0.054010357707738876,
      "learning_rate": 5.7963468898232026e-06,
      "loss": 0.3361,
      "step": 215
    },
    {
      "epoch": 1.5128805620608898,
      "grad_norm": 0.06058075278997421,
      "learning_rate": 5.755831118750016e-06,
      "loss": 0.3657,
      "step": 216
    },
    {
      "epoch": 1.5199063231850118,
      "grad_norm": 0.05514070391654968,
      "learning_rate": 5.715264493745652e-06,
      "loss": 0.3355,
      "step": 217
    },
    {
      "epoch": 1.5269320843091334,
      "grad_norm": 0.056756455451250076,
      "learning_rate": 5.6746497442192425e-06,
      "loss": 0.3322,
      "step": 218
    },
    {
      "epoch": 1.5339578454332554,
      "grad_norm": 0.051219794899225235,
      "learning_rate": 5.633989602817837e-06,
      "loss": 0.314,
      "step": 219
    },
    {
      "epoch": 1.540983606557377,
      "grad_norm": 0.05676623433828354,
      "learning_rate": 5.593286805242549e-06,
      "loss": 0.3442,
      "step": 220
    },
    {
      "epoch": 1.548009367681499,
      "grad_norm": 0.059845082461833954,
      "learning_rate": 5.552544090064487e-06,
      "loss": 0.3441,
      "step": 221
    },
    {
      "epoch": 1.5550351288056206,
      "grad_norm": 0.05877413973212242,
      "learning_rate": 5.5117641985405055e-06,
      "loss": 0.3165,
      "step": 222
    },
    {
      "epoch": 1.5620608899297423,
      "grad_norm": 0.04743430018424988,
      "learning_rate": 5.47094987442876e-06,
      "loss": 0.3253,
      "step": 223
    },
    {
      "epoch": 1.5690866510538641,
      "grad_norm": 0.05630961060523987,
      "learning_rate": 5.430103863804107e-06,
      "loss": 0.3384,
      "step": 224
    },
    {
      "epoch": 1.576112412177986,
      "grad_norm": 0.05597255006432533,
      "learning_rate": 5.389228914873334e-06,
      "loss": 0.3488,
      "step": 225
    },
    {
      "epoch": 1.5831381733021077,
      "grad_norm": 0.045843351632356644,
      "learning_rate": 5.348327777790262e-06,
      "loss": 0.3169,
      "step": 226
    },
    {
      "epoch": 1.5901639344262295,
      "grad_norm": 0.049752555787563324,
      "learning_rate": 5.307403204470711e-06,
      "loss": 0.3393,
      "step": 227
    },
    {
      "epoch": 1.5971896955503513,
      "grad_norm": 0.059289898723363876,
      "learning_rate": 5.266457948407336e-06,
      "loss": 0.3531,
      "step": 228
    },
    {
      "epoch": 1.604215456674473,
      "grad_norm": 0.05609236657619476,
      "learning_rate": 5.2254947644843735e-06,
      "loss": 0.3384,
      "step": 229
    },
    {
      "epoch": 1.6112412177985949,
      "grad_norm": 0.053165484219789505,
      "learning_rate": 5.18451640879228e-06,
      "loss": 0.32,
      "step": 230
    },
    {
      "epoch": 1.6182669789227166,
      "grad_norm": 0.05502665415406227,
      "learning_rate": 5.14352563844231e-06,
      "loss": 0.2832,
      "step": 231
    },
    {
      "epoch": 1.6252927400468384,
      "grad_norm": 0.04645336791872978,
      "learning_rate": 5.1025252113809945e-06,
      "loss": 0.3047,
      "step": 232
    },
    {
      "epoch": 1.6323185011709602,
      "grad_norm": 0.05163678526878357,
      "learning_rate": 5.061517886204592e-06,
      "loss": 0.3143,
      "step": 233
    },
    {
      "epoch": 1.639344262295082,
      "grad_norm": 0.05668428912758827,
      "learning_rate": 5.02050642197348e-06,
      "loss": 0.334,
      "step": 234
    },
    {
      "epoch": 1.6463700234192038,
      "grad_norm": 0.062112439423799515,
      "learning_rate": 4.979493578026523e-06,
      "loss": 0.3259,
      "step": 235
    },
    {
      "epoch": 1.6533957845433256,
      "grad_norm": 0.050369080156087875,
      "learning_rate": 4.9384821137954106e-06,
      "loss": 0.3506,
      "step": 236
    },
    {
      "epoch": 1.6604215456674472,
      "grad_norm": 0.0596189871430397,
      "learning_rate": 4.897474788619007e-06,
      "loss": 0.3211,
      "step": 237
    },
    {
      "epoch": 1.6674473067915692,
      "grad_norm": 0.05669752135872841,
      "learning_rate": 4.856474361557692e-06,
      "loss": 0.3037,
      "step": 238
    },
    {
      "epoch": 1.6744730679156907,
      "grad_norm": 0.048729464411735535,
      "learning_rate": 4.815483591207721e-06,
      "loss": 0.3546,
      "step": 239
    },
    {
      "epoch": 1.6814988290398127,
      "grad_norm": 0.05216846615076065,
      "learning_rate": 4.774505235515628e-06,
      "loss": 0.3619,
      "step": 240
    },
    {
      "epoch": 1.6885245901639343,
      "grad_norm": 0.05519293621182442,
      "learning_rate": 4.733542051592665e-06,
      "loss": 0.3097,
      "step": 241
    },
    {
      "epoch": 1.6955503512880563,
      "grad_norm": 0.05697103589773178,
      "learning_rate": 4.69259679552929e-06,
      "loss": 0.343,
      "step": 242
    },
    {
      "epoch": 1.7025761124121779,
      "grad_norm": 0.057407498359680176,
      "learning_rate": 4.651672222209738e-06,
      "loss": 0.3594,
      "step": 243
    },
    {
      "epoch": 1.7096018735362999,
      "grad_norm": 0.0562685951590538,
      "learning_rate": 4.6107710851266695e-06,
      "loss": 0.3212,
      "step": 244
    },
    {
      "epoch": 1.7166276346604215,
      "grad_norm": 0.0513756088912487,
      "learning_rate": 4.5698961361958955e-06,
      "loss": 0.3164,
      "step": 245
    },
    {
      "epoch": 1.7236533957845435,
      "grad_norm": 0.05124720558524132,
      "learning_rate": 4.529050125571241e-06,
      "loss": 0.3212,
      "step": 246
    },
    {
      "epoch": 1.730679156908665,
      "grad_norm": 0.052108533680438995,
      "learning_rate": 4.488235801459495e-06,
      "loss": 0.3056,
      "step": 247
    },
    {
      "epoch": 1.737704918032787,
      "grad_norm": 0.058752771466970444,
      "learning_rate": 4.447455909935513e-06,
      "loss": 0.3539,
      "step": 248
    },
    {
      "epoch": 1.7447306791569086,
      "grad_norm": 0.053762953728437424,
      "learning_rate": 4.4067131947574515e-06,
      "loss": 0.3157,
      "step": 249
    },
    {
      "epoch": 1.7517564402810304,
      "grad_norm": 0.04892832413315773,
      "learning_rate": 4.3660103971821635e-06,
      "loss": 0.3279,
      "step": 250
    },
    {
      "epoch": 1.7587822014051522,
      "grad_norm": 0.0599125474691391,
      "learning_rate": 4.3253502557807575e-06,
      "loss": 0.3255,
      "step": 251
    },
    {
      "epoch": 1.765807962529274,
      "grad_norm": 0.05481419712305069,
      "learning_rate": 4.28473550625435e-06,
      "loss": 0.3502,
      "step": 252
    },
    {
      "epoch": 1.7728337236533958,
      "grad_norm": 0.05331256613135338,
      "learning_rate": 4.244168881249986e-06,
      "loss": 0.3565,
      "step": 253
    },
    {
      "epoch": 1.7798594847775175,
      "grad_norm": 0.053473543375730515,
      "learning_rate": 4.203653110176798e-06,
      "loss": 0.2904,
      "step": 254
    },
    {
      "epoch": 1.7868852459016393,
      "grad_norm": 0.053232353180646896,
      "learning_rate": 4.163190919022357e-06,
      "loss": 0.321,
      "step": 255
    },
    {
      "epoch": 1.7939110070257611,
      "grad_norm": 0.05323438718914986,
      "learning_rate": 4.122785030169256e-06,
      "loss": 0.3562,
      "step": 256
    },
    {
      "epoch": 1.800936768149883,
      "grad_norm": 0.049881111830472946,
      "learning_rate": 4.082438162211955e-06,
      "loss": 0.3218,
      "step": 257
    },
    {
      "epoch": 1.8079625292740047,
      "grad_norm": 0.2167116105556488,
      "learning_rate": 4.042153029773861e-06,
      "loss": 0.3218,
      "step": 258
    },
    {
      "epoch": 1.8149882903981265,
      "grad_norm": 0.051224831491708755,
      "learning_rate": 4.001932343324683e-06,
      "loss": 0.3317,
      "step": 259
    },
    {
      "epoch": 1.8220140515222483,
      "grad_norm": 0.10497119277715683,
      "learning_rate": 3.961778808998066e-06,
      "loss": 0.3676,
      "step": 260
    },
    {
      "epoch": 1.82903981264637,
      "grad_norm": 0.05344713106751442,
      "learning_rate": 3.921695128409517e-06,
      "loss": 0.3619,
      "step": 261
    },
    {
      "epoch": 1.8360655737704918,
      "grad_norm": 0.04895958676934242,
      "learning_rate": 3.8816839984746334e-06,
      "loss": 0.2942,
      "step": 262
    },
    {
      "epoch": 1.8430913348946136,
      "grad_norm": 0.05228881910443306,
      "learning_rate": 3.841748111227652e-06,
      "loss": 0.3259,
      "step": 263
    },
    {
      "epoch": 1.8501170960187352,
      "grad_norm": 0.04963938146829605,
      "learning_rate": 3.8018901536403198e-06,
      "loss": 0.3582,
      "step": 264
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.056527718901634216,
      "learning_rate": 3.762112807441108e-06,
      "loss": 0.3681,
      "step": 265
    },
    {
      "epoch": 1.8641686182669788,
      "grad_norm": 0.05749901384115219,
      "learning_rate": 3.7224187489347847e-06,
      "loss": 0.3329,
      "step": 266
    },
    {
      "epoch": 1.8711943793911008,
      "grad_norm": 0.05997895449399948,
      "learning_rate": 3.682810648822343e-06,
      "loss": 0.3636,
      "step": 267
    },
    {
      "epoch": 1.8782201405152223,
      "grad_norm": 0.05262213572859764,
      "learning_rate": 3.6432911720213127e-06,
      "loss": 0.3473,
      "step": 268
    },
    {
      "epoch": 1.8852459016393444,
      "grad_norm": 0.05787108838558197,
      "learning_rate": 3.6038629774864563e-06,
      "loss": 0.3231,
      "step": 269
    },
    {
      "epoch": 1.892271662763466,
      "grad_norm": 0.06611717492341995,
      "learning_rate": 3.56452871803087e-06,
      "loss": 0.3687,
      "step": 270
    },
    {
      "epoch": 1.899297423887588,
      "grad_norm": 0.056811776012182236,
      "learning_rate": 3.525291040147498e-06,
      "loss": 0.298,
      "step": 271
    },
    {
      "epoch": 1.9063231850117095,
      "grad_norm": 0.0592680424451828,
      "learning_rate": 3.486152583831072e-06,
      "loss": 0.3253,
      "step": 272
    },
    {
      "epoch": 1.9133489461358315,
      "grad_norm": 0.05636563152074814,
      "learning_rate": 3.447115982400485e-06,
      "loss": 0.3421,
      "step": 273
    },
    {
      "epoch": 1.920374707259953,
      "grad_norm": 0.05371502414345741,
      "learning_rate": 3.4081838623216124e-06,
      "loss": 0.3643,
      "step": 274
    },
    {
      "epoch": 1.927400468384075,
      "grad_norm": 0.05231010541319847,
      "learning_rate": 3.3693588430306035e-06,
      "loss": 0.2898,
      "step": 275
    },
    {
      "epoch": 1.9344262295081966,
      "grad_norm": 0.055993061512708664,
      "learning_rate": 3.330643536757638e-06,
      "loss": 0.3231,
      "step": 276
    },
    {
      "epoch": 1.9414519906323187,
      "grad_norm": 0.05135795846581459,
      "learning_rate": 3.2920405483511702e-06,
      "loss": 0.3241,
      "step": 277
    },
    {
      "epoch": 1.9484777517564402,
      "grad_norm": 0.05628650262951851,
      "learning_rate": 3.253552475102668e-06,
      "loss": 0.3325,
      "step": 278
    },
    {
      "epoch": 1.955503512880562,
      "grad_norm": 0.05219852551817894,
      "learning_rate": 3.215181906571858e-06,
      "loss": 0.3499,
      "step": 279
    },
    {
      "epoch": 1.9625292740046838,
      "grad_norm": 0.05198737978935242,
      "learning_rate": 3.1769314244125056e-06,
      "loss": 0.3252,
      "step": 280
    },
    {
      "epoch": 1.9695550351288056,
      "grad_norm": 0.0620955154299736,
      "learning_rate": 3.1388036021987047e-06,
      "loss": 0.38,
      "step": 281
    },
    {
      "epoch": 1.9765807962529274,
      "grad_norm": 0.05170467123389244,
      "learning_rate": 3.100801005251727e-06,
      "loss": 0.3206,
      "step": 282
    },
    {
      "epoch": 1.9836065573770492,
      "grad_norm": 0.05948978289961815,
      "learning_rate": 3.0629261904674206e-06,
      "loss": 0.3355,
      "step": 283
    },
    {
      "epoch": 1.990632318501171,
      "grad_norm": 0.055122461169958115,
      "learning_rate": 3.025181706144178e-06,
      "loss": 0.2983,
      "step": 284
    },
    {
      "epoch": 1.9976580796252927,
      "grad_norm": 0.05636753514409065,
      "learning_rate": 2.987570091811479e-06,
      "loss": 0.3636,
      "step": 285
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.05636753514409065,
      "learning_rate": 2.9500938780590276e-06,
      "loss": 0.3186,
      "step": 286
    },
    {
      "epoch": 2.0070257611241216,
      "grad_norm": 0.10458986461162567,
      "learning_rate": 2.9127555863664857e-06,
      "loss": 0.254,
      "step": 287
    },
    {
      "epoch": 2.0140515222482436,
      "grad_norm": 0.059524934738874435,
      "learning_rate": 2.8755577289338267e-06,
      "loss": 0.2405,
      "step": 288
    },
    {
      "epoch": 2.021077283372365,
      "grad_norm": 0.0579247772693634,
      "learning_rate": 2.838502808512309e-06,
      "loss": 0.2159,
      "step": 289
    },
    {
      "epoch": 2.028103044496487,
      "grad_norm": 0.06685949116945267,
      "learning_rate": 2.801593318236078e-06,
      "loss": 0.2268,
      "step": 290
    },
    {
      "epoch": 2.0351288056206087,
      "grad_norm": 0.05411679297685623,
      "learning_rate": 2.764831741454432e-06,
      "loss": 0.2579,
      "step": 291
    },
    {
      "epoch": 2.0421545667447307,
      "grad_norm": 0.058176521211862564,
      "learning_rate": 2.7282205515647348e-06,
      "loss": 0.195,
      "step": 292
    },
    {
      "epoch": 2.0491803278688523,
      "grad_norm": 0.051991887390613556,
      "learning_rate": 2.6917622118459975e-06,
      "loss": 0.2002,
      "step": 293
    },
    {
      "epoch": 2.0562060889929743,
      "grad_norm": 0.06430847942829132,
      "learning_rate": 2.655459175293146e-06,
      "loss": 0.1863,
      "step": 294
    },
    {
      "epoch": 2.063231850117096,
      "grad_norm": 0.06031821295619011,
      "learning_rate": 2.6193138844519785e-06,
      "loss": 0.2521,
      "step": 295
    },
    {
      "epoch": 2.070257611241218,
      "grad_norm": 0.07283060252666473,
      "learning_rate": 2.58332877125482e-06,
      "loss": 0.2138,
      "step": 296
    },
    {
      "epoch": 2.0772833723653394,
      "grad_norm": 0.05646536499261856,
      "learning_rate": 2.5475062568569077e-06,
      "loss": 0.2463,
      "step": 297
    },
    {
      "epoch": 2.0843091334894615,
      "grad_norm": 0.0762052983045578,
      "learning_rate": 2.511848751473485e-06,
      "loss": 0.2135,
      "step": 298
    },
    {
      "epoch": 2.091334894613583,
      "grad_norm": 0.04977189749479294,
      "learning_rate": 2.476358654217627e-06,
      "loss": 0.1957,
      "step": 299
    },
    {
      "epoch": 2.098360655737705,
      "grad_norm": 0.05424179881811142,
      "learning_rate": 2.4410383529388448e-06,
      "loss": 0.2154,
      "step": 300
    },
    {
      "epoch": 2.1053864168618266,
      "grad_norm": 0.05405616760253906,
      "learning_rate": 2.405890224062406e-06,
      "loss": 0.2171,
      "step": 301
    },
    {
      "epoch": 2.1124121779859486,
      "grad_norm": 0.050341855734586716,
      "learning_rate": 2.370916632429455e-06,
      "loss": 0.1926,
      "step": 302
    },
    {
      "epoch": 2.11943793911007,
      "grad_norm": 0.05347389727830887,
      "learning_rate": 2.336119931137897e-06,
      "loss": 0.2238,
      "step": 303
    },
    {
      "epoch": 2.126463700234192,
      "grad_norm": 0.061148934066295624,
      "learning_rate": 2.3015024613840742e-06,
      "loss": 0.2218,
      "step": 304
    },
    {
      "epoch": 2.1334894613583137,
      "grad_norm": 0.056503091007471085,
      "learning_rate": 2.2670665523052534e-06,
      "loss": 0.2147,
      "step": 305
    },
    {
      "epoch": 2.1405152224824358,
      "grad_norm": 0.05360401049256325,
      "learning_rate": 2.2328145208229096e-06,
      "loss": 0.1889,
      "step": 306
    },
    {
      "epoch": 2.1475409836065573,
      "grad_norm": 0.05166396126151085,
      "learning_rate": 2.1987486714868384e-06,
      "loss": 0.1911,
      "step": 307
    },
    {
      "epoch": 2.1545667447306793,
      "grad_norm": 0.05574238300323486,
      "learning_rate": 2.164871296320106e-06,
      "loss": 0.2184,
      "step": 308
    },
    {
      "epoch": 2.161592505854801,
      "grad_norm": 0.05423097312450409,
      "learning_rate": 2.1311846746648325e-06,
      "loss": 0.1904,
      "step": 309
    },
    {
      "epoch": 2.168618266978923,
      "grad_norm": 0.04529114067554474,
      "learning_rate": 2.097691073028836e-06,
      "loss": 0.2049,
      "step": 310
    },
    {
      "epoch": 2.1756440281030445,
      "grad_norm": 0.050600163638591766,
      "learning_rate": 2.064392744933135e-06,
      "loss": 0.2136,
      "step": 311
    },
    {
      "epoch": 2.1826697892271665,
      "grad_norm": 0.04993721842765808,
      "learning_rate": 2.0312919307603286e-06,
      "loss": 0.1835,
      "step": 312
    },
    {
      "epoch": 2.189695550351288,
      "grad_norm": 0.050447553396224976,
      "learning_rate": 1.998390857603853e-06,
      "loss": 0.202,
      "step": 313
    },
    {
      "epoch": 2.19672131147541,
      "grad_norm": 0.05513535067439079,
      "learning_rate": 1.965691739118146e-06,
      "loss": 0.1964,
      "step": 314
    },
    {
      "epoch": 2.2037470725995316,
      "grad_norm": 0.05925602465867996,
      "learning_rate": 1.9331967753697077e-06,
      "loss": 0.1935,
      "step": 315
    },
    {
      "epoch": 2.210772833723653,
      "grad_norm": 0.05441617593169212,
      "learning_rate": 1.9009081526890622e-06,
      "loss": 0.1894,
      "step": 316
    },
    {
      "epoch": 2.217798594847775,
      "grad_norm": 0.06225774437189102,
      "learning_rate": 1.8688280435236732e-06,
      "loss": 0.1849,
      "step": 317
    },
    {
      "epoch": 2.2248243559718968,
      "grad_norm": 0.05718431621789932,
      "learning_rate": 1.8369586062917693e-06,
      "loss": 0.2327,
      "step": 318
    },
    {
      "epoch": 2.2318501170960188,
      "grad_norm": 0.05463183671236038,
      "learning_rate": 1.8053019852371195e-06,
      "loss": 0.2159,
      "step": 319
    },
    {
      "epoch": 2.2388758782201403,
      "grad_norm": 0.05712981894612312,
      "learning_rate": 1.7738603102847696e-06,
      "loss": 0.2057,
      "step": 320
    },
    {
      "epoch": 2.2459016393442623,
      "grad_norm": 0.05346499755978584,
      "learning_rate": 1.7426356968977265e-06,
      "loss": 0.2406,
      "step": 321
    },
    {
      "epoch": 2.252927400468384,
      "grad_norm": 0.06216704845428467,
      "learning_rate": 1.711630245934638e-06,
      "loss": 0.2307,
      "step": 322
    },
    {
      "epoch": 2.259953161592506,
      "grad_norm": 0.04996800422668457,
      "learning_rate": 1.6808460435084316e-06,
      "loss": 0.2145,
      "step": 323
    },
    {
      "epoch": 2.2669789227166275,
      "grad_norm": 0.05448152869939804,
      "learning_rate": 1.6502851608459668e-06,
      "loss": 0.2114,
      "step": 324
    },
    {
      "epoch": 2.2740046838407495,
      "grad_norm": 0.05008429288864136,
      "learning_rate": 1.6199496541486647e-06,
      "loss": 0.1881,
      "step": 325
    },
    {
      "epoch": 2.281030444964871,
      "grad_norm": 0.04923682287335396,
      "learning_rate": 1.589841564454176e-06,
      "loss": 0.2114,
      "step": 326
    },
    {
      "epoch": 2.288056206088993,
      "grad_norm": 0.04919268935918808,
      "learning_rate": 1.5599629174990482e-06,
      "loss": 0.2149,
      "step": 327
    },
    {
      "epoch": 2.2950819672131146,
      "grad_norm": 0.05411992594599724,
      "learning_rate": 1.5303157235824323e-06,
      "loss": 0.2034,
      "step": 328
    },
    {
      "epoch": 2.3021077283372366,
      "grad_norm": 0.047855135053396225,
      "learning_rate": 1.5009019774308249e-06,
      "loss": 0.1977,
      "step": 329
    },
    {
      "epoch": 2.309133489461358,
      "grad_norm": 0.05060090869665146,
      "learning_rate": 1.471723658063856e-06,
      "loss": 0.226,
      "step": 330
    },
    {
      "epoch": 2.3161592505854802,
      "grad_norm": 0.05514250695705414,
      "learning_rate": 1.4427827286611412e-06,
      "loss": 0.2213,
      "step": 331
    },
    {
      "epoch": 2.323185011709602,
      "grad_norm": 0.05435176566243172,
      "learning_rate": 1.4140811364301931e-06,
      "loss": 0.1782,
      "step": 332
    },
    {
      "epoch": 2.330210772833724,
      "grad_norm": 0.05435670167207718,
      "learning_rate": 1.385620812475409e-06,
      "loss": 0.2196,
      "step": 333
    },
    {
      "epoch": 2.3372365339578454,
      "grad_norm": 0.060686711221933365,
      "learning_rate": 1.3574036716681366e-06,
      "loss": 0.2386,
      "step": 334
    },
    {
      "epoch": 2.3442622950819674,
      "grad_norm": 0.045962344855070114,
      "learning_rate": 1.3294316125178474e-06,
      "loss": 0.2073,
      "step": 335
    },
    {
      "epoch": 2.351288056206089,
      "grad_norm": 0.05140933021903038,
      "learning_rate": 1.301706517044395e-06,
      "loss": 0.2094,
      "step": 336
    },
    {
      "epoch": 2.358313817330211,
      "grad_norm": 0.052996646612882614,
      "learning_rate": 1.2742302506513894e-06,
      "loss": 0.2183,
      "step": 337
    },
    {
      "epoch": 2.3653395784543325,
      "grad_norm": 0.060029152780771255,
      "learning_rate": 1.247004662000686e-06,
      "loss": 0.2298,
      "step": 338
    },
    {
      "epoch": 2.3723653395784545,
      "grad_norm": 0.0602019801735878,
      "learning_rate": 1.2200315828880094e-06,
      "loss": 0.2162,
      "step": 339
    },
    {
      "epoch": 2.379391100702576,
      "grad_norm": 0.0632694885134697,
      "learning_rate": 1.1933128281197042e-06,
      "loss": 0.2032,
      "step": 340
    },
    {
      "epoch": 2.3864168618266977,
      "grad_norm": 0.05728806555271149,
      "learning_rate": 1.166850195390628e-06,
      "loss": 0.238,
      "step": 341
    },
    {
      "epoch": 2.3934426229508197,
      "grad_norm": 0.05771248787641525,
      "learning_rate": 1.1406454651632042e-06,
      "loss": 0.2385,
      "step": 342
    },
    {
      "epoch": 2.4004683840749417,
      "grad_norm": 0.09708061069250107,
      "learning_rate": 1.1147004005476192e-06,
      "loss": 0.2367,
      "step": 343
    },
    {
      "epoch": 2.4074941451990632,
      "grad_norm": 0.05926092714071274,
      "learning_rate": 1.089016747183208e-06,
      "loss": 0.1787,
      "step": 344
    },
    {
      "epoch": 2.414519906323185,
      "grad_norm": 0.05011197552084923,
      "learning_rate": 1.063596233120997e-06,
      "loss": 0.2119,
      "step": 345
    },
    {
      "epoch": 2.421545667447307,
      "grad_norm": 0.05037648603320122,
      "learning_rate": 1.03844056870744e-06,
      "loss": 0.2177,
      "step": 346
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 0.05578481778502464,
      "learning_rate": 1.013551446469337e-06,
      "loss": 0.1988,
      "step": 347
    },
    {
      "epoch": 2.4355971896955504,
      "grad_norm": 0.05308612063527107,
      "learning_rate": 9.889305409999656e-07,
      "loss": 0.2006,
      "step": 348
    },
    {
      "epoch": 2.442622950819672,
      "grad_norm": 0.05964551866054535,
      "learning_rate": 9.64579508846405e-07,
      "loss": 0.2598,
      "step": 349
    },
    {
      "epoch": 2.449648711943794,
      "grad_norm": 0.057544246315956116,
      "learning_rate": 9.40499988398082e-07,
      "loss": 0.2095,
      "step": 350
    },
    {
      "epoch": 2.4566744730679155,
      "grad_norm": 0.05650878697633743,
      "learning_rate": 9.166935997765364e-07,
      "loss": 0.2347,
      "step": 351
    },
    {
      "epoch": 2.4637002341920375,
      "grad_norm": 0.07671873271465302,
      "learning_rate": 8.93161944726414e-07,
      "loss": 0.2145,
      "step": 352
    },
    {
      "epoch": 2.470725995316159,
      "grad_norm": 0.0586552731692791,
      "learning_rate": 8.699066065077005e-07,
      "loss": 0.2078,
      "step": 353
    },
    {
      "epoch": 2.477751756440281,
      "grad_norm": 0.045289915055036545,
      "learning_rate": 8.469291497891979e-07,
      "loss": 0.2129,
      "step": 354
    },
    {
      "epoch": 2.4847775175644027,
      "grad_norm": 0.05548376590013504,
      "learning_rate": 8.242311205432418e-07,
      "loss": 0.1944,
      "step": 355
    },
    {
      "epoch": 2.4918032786885247,
      "grad_norm": 0.04866700619459152,
      "learning_rate": 8.018140459416962e-07,
      "loss": 0.2228,
      "step": 356
    },
    {
      "epoch": 2.4988290398126463,
      "grad_norm": 0.05934524163603783,
      "learning_rate": 7.796794342531949e-07,
      "loss": 0.2155,
      "step": 357
    },
    {
      "epoch": 2.5058548009367683,
      "grad_norm": 0.056303609162569046,
      "learning_rate": 7.57828774741664e-07,
      "loss": 0.2048,
      "step": 358
    },
    {
      "epoch": 2.51288056206089,
      "grad_norm": 0.05041569098830223,
      "learning_rate": 7.362635375661225e-07,
      "loss": 0.224,
      "step": 359
    },
    {
      "epoch": 2.519906323185012,
      "grad_norm": 0.05846166983246803,
      "learning_rate": 7.149851736817609e-07,
      "loss": 0.2502,
      "step": 360
    },
    {
      "epoch": 2.5269320843091334,
      "grad_norm": 0.0575077049434185,
      "learning_rate": 6.939951147423269e-07,
      "loss": 0.1915,
      "step": 361
    },
    {
      "epoch": 2.5339578454332554,
      "grad_norm": 0.05687614157795906,
      "learning_rate": 6.732947730037936e-07,
      "loss": 0.2019,
      "step": 362
    },
    {
      "epoch": 2.540983606557377,
      "grad_norm": 0.05574915185570717,
      "learning_rate": 6.52885541229345e-07,
      "loss": 0.2085,
      "step": 363
    },
    {
      "epoch": 2.548009367681499,
      "grad_norm": 0.05865642428398132,
      "learning_rate": 6.327687925956616e-07,
      "loss": 0.1975,
      "step": 364
    },
    {
      "epoch": 2.5550351288056206,
      "grad_norm": 0.04809865728020668,
      "learning_rate": 6.12945880600535e-07,
      "loss": 0.1823,
      "step": 365
    },
    {
      "epoch": 2.562060889929742,
      "grad_norm": 0.05387086421251297,
      "learning_rate": 5.93418138971803e-07,
      "loss": 0.2394,
      "step": 366
    },
    {
      "epoch": 2.569086651053864,
      "grad_norm": 0.05231115594506264,
      "learning_rate": 5.741868815776081e-07,
      "loss": 0.2235,
      "step": 367
    },
    {
      "epoch": 2.576112412177986,
      "grad_norm": 0.07717155665159225,
      "learning_rate": 5.552534023380024e-07,
      "loss": 0.2279,
      "step": 368
    },
    {
      "epoch": 2.5831381733021077,
      "grad_norm": 0.05653085187077522,
      "learning_rate": 5.366189751378858e-07,
      "loss": 0.1869,
      "step": 369
    },
    {
      "epoch": 2.5901639344262293,
      "grad_norm": 0.05067911744117737,
      "learning_rate": 5.18284853741301e-07,
      "loss": 0.2376,
      "step": 370
    },
    {
      "epoch": 2.5971896955503513,
      "grad_norm": 0.05797469988465309,
      "learning_rate": 5.002522717070751e-07,
      "loss": 0.1973,
      "step": 371
    },
    {
      "epoch": 2.6042154566744733,
      "grad_norm": 0.05374491214752197,
      "learning_rate": 4.8252244230582e-07,
      "loss": 0.2014,
      "step": 372
    },
    {
      "epoch": 2.611241217798595,
      "grad_norm": 0.05649881809949875,
      "learning_rate": 4.6509655843830827e-07,
      "loss": 0.2027,
      "step": 373
    },
    {
      "epoch": 2.6182669789227164,
      "grad_norm": 0.054554861038923264,
      "learning_rate": 4.4797579255520585e-07,
      "loss": 0.2092,
      "step": 374
    },
    {
      "epoch": 2.6252927400468384,
      "grad_norm": 0.054530441761016846,
      "learning_rate": 4.311612965781903e-07,
      "loss": 0.2196,
      "step": 375
    },
    {
      "epoch": 2.6323185011709604,
      "grad_norm": 0.0523928701877594,
      "learning_rate": 4.1465420182244476e-07,
      "loss": 0.2104,
      "step": 376
    },
    {
      "epoch": 2.639344262295082,
      "grad_norm": 0.057155121117830276,
      "learning_rate": 3.984556189205441e-07,
      "loss": 0.1966,
      "step": 377
    },
    {
      "epoch": 2.6463700234192036,
      "grad_norm": 0.05527928099036217,
      "learning_rate": 3.8256663774772383e-07,
      "loss": 0.235,
      "step": 378
    },
    {
      "epoch": 2.6533957845433256,
      "grad_norm": 0.05507628992199898,
      "learning_rate": 3.669883273485575e-07,
      "loss": 0.2284,
      "step": 379
    },
    {
      "epoch": 2.660421545667447,
      "grad_norm": 0.05377492681145668,
      "learning_rate": 3.5172173586502543e-07,
      "loss": 0.1792,
      "step": 380
    },
    {
      "epoch": 2.667447306791569,
      "grad_norm": 0.05631445720791817,
      "learning_rate": 3.3676789046599045e-07,
      "loss": 0.1963,
      "step": 381
    },
    {
      "epoch": 2.6744730679156907,
      "grad_norm": 0.05693857744336128,
      "learning_rate": 3.2212779727809504e-07,
      "loss": 0.2265,
      "step": 382
    },
    {
      "epoch": 2.6814988290398127,
      "grad_norm": 0.052210524678230286,
      "learning_rate": 3.0780244131806193e-07,
      "loss": 0.2292,
      "step": 383
    },
    {
      "epoch": 2.6885245901639343,
      "grad_norm": 0.06421205401420593,
      "learning_rate": 2.937927864264206e-07,
      "loss": 0.2247,
      "step": 384
    },
    {
      "epoch": 2.6955503512880563,
      "grad_norm": 0.054567910730838776,
      "learning_rate": 2.800997752026596e-07,
      "loss": 0.2115,
      "step": 385
    },
    {
      "epoch": 2.702576112412178,
      "grad_norm": 0.05900174379348755,
      "learning_rate": 2.667243289418059e-07,
      "loss": 0.2014,
      "step": 386
    },
    {
      "epoch": 2.7096018735363,
      "grad_norm": 0.05543851852416992,
      "learning_rate": 2.5366734757243496e-07,
      "loss": 0.1795,
      "step": 387
    },
    {
      "epoch": 2.7166276346604215,
      "grad_norm": 0.05802013352513313,
      "learning_rate": 2.4092970959612885e-07,
      "loss": 0.195,
      "step": 388
    },
    {
      "epoch": 2.7236533957845435,
      "grad_norm": 0.04725894704461098,
      "learning_rate": 2.2851227202836002e-07,
      "loss": 0.2301,
      "step": 389
    },
    {
      "epoch": 2.730679156908665,
      "grad_norm": 0.05542716756463051,
      "learning_rate": 2.1641587034083756e-07,
      "loss": 0.2115,
      "step": 390
    },
    {
      "epoch": 2.737704918032787,
      "grad_norm": 0.048591192811727524,
      "learning_rate": 2.0464131840528978e-07,
      "loss": 0.1922,
      "step": 391
    },
    {
      "epoch": 2.7447306791569086,
      "grad_norm": 0.04870427027344704,
      "learning_rate": 1.9318940843870594e-07,
      "loss": 0.1852,
      "step": 392
    },
    {
      "epoch": 2.7517564402810306,
      "grad_norm": 0.05185515433549881,
      "learning_rate": 1.8206091095003543e-07,
      "loss": 0.2278,
      "step": 393
    },
    {
      "epoch": 2.758782201405152,
      "grad_norm": 0.056988395750522614,
      "learning_rate": 1.7125657468834656e-07,
      "loss": 0.1981,
      "step": 394
    },
    {
      "epoch": 2.7658079625292737,
      "grad_norm": 0.050774361938238144,
      "learning_rate": 1.6077712659244792e-07,
      "loss": 0.2147,
      "step": 395
    },
    {
      "epoch": 2.7728337236533958,
      "grad_norm": 0.053753916174173355,
      "learning_rate": 1.5062327174197645e-07,
      "loss": 0.1922,
      "step": 396
    },
    {
      "epoch": 2.7798594847775178,
      "grad_norm": 0.06190885975956917,
      "learning_rate": 1.4079569330996412e-07,
      "loss": 0.2409,
      "step": 397
    },
    {
      "epoch": 2.7868852459016393,
      "grad_norm": 0.05707972124218941,
      "learning_rate": 1.3129505251686603e-07,
      "loss": 0.1889,
      "step": 398
    },
    {
      "epoch": 2.793911007025761,
      "grad_norm": 0.05355476588010788,
      "learning_rate": 1.2212198858607694e-07,
      "loss": 0.2026,
      "step": 399
    },
    {
      "epoch": 2.800936768149883,
      "grad_norm": 0.055988870561122894,
      "learning_rate": 1.1327711870091963e-07,
      "loss": 0.2114,
      "step": 400
    },
    {
      "epoch": 2.807962529274005,
      "grad_norm": 0.056202538311481476,
      "learning_rate": 1.0476103796312254e-07,
      "loss": 0.2217,
      "step": 401
    },
    {
      "epoch": 2.8149882903981265,
      "grad_norm": 0.06274823844432831,
      "learning_rate": 9.657431935277629e-08,
      "loss": 0.2113,
      "step": 402
    },
    {
      "epoch": 2.822014051522248,
      "grad_norm": 0.049836281687021255,
      "learning_rate": 8.871751368978554e-08,
      "loss": 0.2032,
      "step": 403
    },
    {
      "epoch": 2.82903981264637,
      "grad_norm": 0.05938475951552391,
      "learning_rate": 8.119114959680929e-08,
      "loss": 0.19,
      "step": 404
    },
    {
      "epoch": 2.836065573770492,
      "grad_norm": 0.06170409545302391,
      "learning_rate": 7.399573346368871e-08,
      "loss": 0.2377,
      "step": 405
    },
    {
      "epoch": 2.8430913348946136,
      "grad_norm": 0.05599289387464523,
      "learning_rate": 6.713174941338163e-08,
      "loss": 0.2307,
      "step": 406
    },
    {
      "epoch": 2.850117096018735,
      "grad_norm": 0.05716150254011154,
      "learning_rate": 6.05996592693886e-08,
      "loss": 0.1509,
      "step": 407
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 0.04617927595973015,
      "learning_rate": 5.439990252467886e-08,
      "loss": 0.2429,
      "step": 408
    },
    {
      "epoch": 2.8641686182669788,
      "grad_norm": 0.05094977468252182,
      "learning_rate": 4.853289631212066e-08,
      "loss": 0.2302,
      "step": 409
    },
    {
      "epoch": 2.871194379391101,
      "grad_norm": 0.05809462442994118,
      "learning_rate": 4.299903537641703e-08,
      "loss": 0.1985,
      "step": 410
    },
    {
      "epoch": 2.8782201405152223,
      "grad_norm": 0.0512741394340992,
      "learning_rate": 3.779869204754427e-08,
      "loss": 0.2277,
      "step": 411
    },
    {
      "epoch": 2.8852459016393444,
      "grad_norm": 0.06057005748152733,
      "learning_rate": 3.2932216215704195e-08,
      "loss": 0.1957,
      "step": 412
    },
    {
      "epoch": 2.892271662763466,
      "grad_norm": 0.05182475969195366,
      "learning_rate": 2.8399935307778516e-08,
      "loss": 0.1988,
      "step": 413
    },
    {
      "epoch": 2.899297423887588,
      "grad_norm": 0.05228479206562042,
      "learning_rate": 2.420215426530259e-08,
      "loss": 0.1873,
      "step": 414
    },
    {
      "epoch": 2.9063231850117095,
      "grad_norm": 0.05834292992949486,
      "learning_rate": 2.0339155523945164e-08,
      "loss": 0.1845,
      "step": 415
    },
    {
      "epoch": 2.9133489461358315,
      "grad_norm": 0.053505126386880875,
      "learning_rate": 1.681119899450856e-08,
      "loss": 0.208,
      "step": 416
    },
    {
      "epoch": 2.920374707259953,
      "grad_norm": 0.05727040395140648,
      "learning_rate": 1.3618522045439897e-08,
      "loss": 0.2026,
      "step": 417
    },
    {
      "epoch": 2.927400468384075,
      "grad_norm": 0.15337811410427094,
      "learning_rate": 1.0761339486859424e-08,
      "loss": 0.2024,
      "step": 418
    },
    {
      "epoch": 2.9344262295081966,
      "grad_norm": 0.05246254801750183,
      "learning_rate": 8.239843556108739e-09,
      "loss": 0.1698,
      "step": 419
    },
    {
      "epoch": 2.9414519906323187,
      "grad_norm": 0.06088126823306084,
      "learning_rate": 6.054203904817812e-09,
      "loss": 0.2337,
      "step": 420
    },
    {
      "epoch": 2.9484777517564402,
      "grad_norm": 0.05842256173491478,
      "learning_rate": 4.204567587486885e-09,
      "loss": 0.2318,
      "step": 421
    },
    {
      "epoch": 2.9555035128805622,
      "grad_norm": 0.05337844043970108,
      "learning_rate": 2.6910590515966117e-09,
      "loss": 0.1856,
      "step": 422
    },
    {
      "epoch": 2.962529274004684,
      "grad_norm": 0.05508222058415413,
      "learning_rate": 1.5137801292325338e-09,
      "loss": 0.227,
      "step": 423
    },
    {
      "epoch": 2.9695550351288054,
      "grad_norm": 0.05756290256977081,
      "learning_rate": 6.728100302327844e-10,
      "loss": 0.1856,
      "step": 424
    },
    {
      "epoch": 2.9765807962529274,
      "grad_norm": 0.05223892256617546,
      "learning_rate": 1.6820533686179308e-10,
      "loss": 0.2299,
      "step": 425
    },
    {
      "epoch": 2.9836065573770494,
      "grad_norm": 0.0553586483001709,
      "learning_rate": 0.0,
      "loss": 0.1971,
      "step": 426
    },
    {
      "epoch": 2.9836065573770494,
      "step": 426,
      "total_flos": 1.0638855097443942e+17,
      "train_loss": 0.35510572714145194,
      "train_runtime": 58523.3937,
      "train_samples_per_second": 0.175,
      "train_steps_per_second": 0.007
    }
  ],
  "logging_steps": 1,
  "max_steps": 426,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0638855097443942e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}