{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 11.594202898550725,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 3.75e-06,
      "loss": 2.586,
      "step": 1
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.5e-06,
      "loss": 2.5846,
      "step": 2
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.125e-05,
      "loss": 2.5617,
      "step": 3
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5e-05,
      "loss": 2.5385,
      "step": 4
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 2.5426,
      "step": 5
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.25e-05,
      "loss": 2.5664,
      "step": 6
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.625e-05,
      "loss": 2.5914,
      "step": 7
    },
    {
      "epoch": 0.19,
      "learning_rate": 3e-05,
      "loss": 2.4594,
      "step": 8
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.9939024390243903e-05,
      "loss": 2.5858,
      "step": 9
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.9878048780487805e-05,
      "loss": 2.5069,
      "step": 10
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.9817073170731707e-05,
      "loss": 2.5146,
      "step": 11
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.975609756097561e-05,
      "loss": 2.4025,
      "step": 12
    },
    {
      "epoch": 0.3,
      "learning_rate": 2.9695121951219515e-05,
      "loss": 2.4719,
      "step": 13
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.9634146341463413e-05,
      "loss": 2.4496,
      "step": 14
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.9573170731707316e-05,
      "loss": 2.3463,
      "step": 15
    },
    {
      "epoch": 0.37,
      "learning_rate": 2.951219512195122e-05,
      "loss": 2.4185,
      "step": 16
    },
    {
      "epoch": 0.39,
      "learning_rate": 2.9451219512195123e-05,
      "loss": 2.3562,
      "step": 17
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.9390243902439022e-05,
      "loss": 2.3574,
      "step": 18
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.9329268292682927e-05,
      "loss": 2.3392,
      "step": 19
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.926829268292683e-05,
      "loss": 2.2794,
      "step": 20
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.9207317073170735e-05,
      "loss": 2.2295,
      "step": 21
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.9146341463414634e-05,
      "loss": 2.2596,
      "step": 22
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.9085365853658536e-05,
      "loss": 2.1936,
      "step": 23
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.902439024390244e-05,
      "loss": 2.1734,
      "step": 24
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.8963414634146343e-05,
      "loss": 2.1466,
      "step": 25
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8902439024390242e-05,
      "loss": 2.119,
      "step": 26
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.8841463414634148e-05,
      "loss": 2.0868,
      "step": 27
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.878048780487805e-05,
      "loss": 2.1283,
      "step": 28
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.8719512195121952e-05,
      "loss": 2.0504,
      "step": 29
    },
    {
      "epoch": 0.7,
      "learning_rate": 2.8658536585365854e-05,
      "loss": 1.9765,
      "step": 30
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.8597560975609756e-05,
      "loss": 1.9256,
      "step": 31
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.8536585365853658e-05,
      "loss": 2.0544,
      "step": 32
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.8475609756097564e-05,
      "loss": 2.0002,
      "step": 33
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.8414634146341462e-05,
      "loss": 1.9999,
      "step": 34
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.8353658536585365e-05,
      "loss": 2.0487,
      "step": 35
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.829268292682927e-05,
      "loss": 2.0628,
      "step": 36
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.8231707317073172e-05,
      "loss": 2.0396,
      "step": 37
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.817073170731707e-05,
      "loss": 1.9948,
      "step": 38
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.8109756097560976e-05,
      "loss": 1.9887,
      "step": 39
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.804878048780488e-05,
      "loss": 2.0597,
      "step": 40
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.7987804878048784e-05,
      "loss": 2.0228,
      "step": 41
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.7926829268292683e-05,
      "loss": 2.0505,
      "step": 42
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.7865853658536585e-05,
      "loss": 1.9953,
      "step": 43
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.780487804878049e-05,
      "loss": 2.061,
      "step": 44
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.7743902439024393e-05,
      "loss": 2.01,
      "step": 45
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.768292682926829e-05,
      "loss": 1.9348,
      "step": 46
    },
    {
      "epoch": 1.09,
      "learning_rate": 2.7621951219512197e-05,
      "loss": 2.0096,
      "step": 47
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.75609756097561e-05,
      "loss": 1.9776,
      "step": 48
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.75e-05,
      "loss": 1.9766,
      "step": 49
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.7439024390243903e-05,
      "loss": 1.963,
      "step": 50
    },
    {
      "epoch": 1.18,
      "learning_rate": 2.7378048780487805e-05,
      "loss": 1.939,
      "step": 51
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.7317073170731707e-05,
      "loss": 1.9372,
      "step": 52
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.7256097560975613e-05,
      "loss": 1.9914,
      "step": 53
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.719512195121951e-05,
      "loss": 1.9472,
      "step": 54
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.7134146341463414e-05,
      "loss": 1.9727,
      "step": 55
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.707317073170732e-05,
      "loss": 1.9598,
      "step": 56
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.701219512195122e-05,
      "loss": 1.913,
      "step": 57
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.695121951219512e-05,
      "loss": 1.9835,
      "step": 58
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.6890243902439026e-05,
      "loss": 1.9365,
      "step": 59
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6829268292682928e-05,
      "loss": 1.9472,
      "step": 60
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6768292682926833e-05,
      "loss": 1.9362,
      "step": 61
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6707317073170732e-05,
      "loss": 1.9834,
      "step": 62
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.6646341463414634e-05,
      "loss": 1.9174,
      "step": 63
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.658536585365854e-05,
      "loss": 1.9143,
      "step": 64
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.652439024390244e-05,
      "loss": 1.9332,
      "step": 65
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.646341463414634e-05,
      "loss": 1.9531,
      "step": 66
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.6402439024390246e-05,
      "loss": 1.9712,
      "step": 67
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.6341463414634148e-05,
      "loss": 1.8767,
      "step": 68
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.628048780487805e-05,
      "loss": 1.9734,
      "step": 69
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.6219512195121952e-05,
      "loss": 1.9125,
      "step": 70
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.6158536585365854e-05,
      "loss": 1.9498,
      "step": 71
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.6097560975609756e-05,
      "loss": 1.9364,
      "step": 72
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.603658536585366e-05,
      "loss": 1.9647,
      "step": 73
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.597560975609756e-05,
      "loss": 1.9455,
      "step": 74
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.5914634146341463e-05,
      "loss": 2.0236,
      "step": 75
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.5853658536585368e-05,
      "loss": 1.941,
      "step": 76
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.5792682926829267e-05,
      "loss": 1.9002,
      "step": 77
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.573170731707317e-05,
      "loss": 1.9697,
      "step": 78
    },
    {
      "epoch": 1.83,
      "learning_rate": 2.5670731707317075e-05,
      "loss": 1.8974,
      "step": 79
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.5609756097560977e-05,
      "loss": 1.9521,
      "step": 80
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.554878048780488e-05,
      "loss": 1.9044,
      "step": 81
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.548780487804878e-05,
      "loss": 1.9168,
      "step": 82
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.5426829268292683e-05,
      "loss": 1.9546,
      "step": 83
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.536585365853659e-05,
      "loss": 1.9659,
      "step": 84
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.5304878048780487e-05,
      "loss": 1.9973,
      "step": 85
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.524390243902439e-05,
      "loss": 1.9115,
      "step": 86
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.5182926829268295e-05,
      "loss": 1.9176,
      "step": 87
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.5121951219512197e-05,
      "loss": 1.8841,
      "step": 88
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.5060975609756096e-05,
      "loss": 1.9046,
      "step": 89
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.5e-05,
      "loss": 1.8913,
      "step": 90
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.4939024390243903e-05,
      "loss": 1.8652,
      "step": 91
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.4878048780487805e-05,
      "loss": 1.9712,
      "step": 92
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.4817073170731708e-05,
      "loss": 1.8976,
      "step": 93
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.475609756097561e-05,
      "loss": 1.8578,
      "step": 94
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.4695121951219512e-05,
      "loss": 1.892,
      "step": 95
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.4634146341463417e-05,
      "loss": 1.8431,
      "step": 96
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.4573170731707316e-05,
      "loss": 1.9123,
      "step": 97
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.4512195121951218e-05,
      "loss": 1.9426,
      "step": 98
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.4451219512195124e-05,
      "loss": 1.9891,
      "step": 99
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.4390243902439026e-05,
      "loss": 1.878,
      "step": 100
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4329268292682928e-05,
      "loss": 1.9455,
      "step": 101
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.426829268292683e-05,
      "loss": 1.8838,
      "step": 102
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.4207317073170732e-05,
      "loss": 1.8222,
      "step": 103
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.4146341463414638e-05,
      "loss": 1.9379,
      "step": 104
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.4085365853658536e-05,
      "loss": 1.8979,
      "step": 105
    },
    {
      "epoch": 2.46,
      "learning_rate": 2.402439024390244e-05,
      "loss": 1.9112,
      "step": 106
    },
    {
      "epoch": 2.48,
      "learning_rate": 2.3963414634146344e-05,
      "loss": 1.8726,
      "step": 107
    },
    {
      "epoch": 2.5,
      "learning_rate": 2.3902439024390246e-05,
      "loss": 1.952,
      "step": 108
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.3841463414634145e-05,
      "loss": 1.8966,
      "step": 109
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.378048780487805e-05,
      "loss": 1.951,
      "step": 110
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.3719512195121952e-05,
      "loss": 1.9121,
      "step": 111
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.3658536585365854e-05,
      "loss": 1.8444,
      "step": 112
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.3597560975609757e-05,
      "loss": 1.8815,
      "step": 113
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.353658536585366e-05,
      "loss": 1.9449,
      "step": 114
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.347560975609756e-05,
      "loss": 1.9857,
      "step": 115
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.3414634146341466e-05,
      "loss": 1.8547,
      "step": 116
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3353658536585365e-05,
      "loss": 1.9372,
      "step": 117
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.3292682926829267e-05,
      "loss": 1.8957,
      "step": 118
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.3231707317073173e-05,
      "loss": 1.8939,
      "step": 119
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.3170731707317075e-05,
      "loss": 1.9126,
      "step": 120
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.3109756097560977e-05,
      "loss": 1.9055,
      "step": 121
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.304878048780488e-05,
      "loss": 1.8628,
      "step": 122
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.298780487804878e-05,
      "loss": 1.9196,
      "step": 123
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.2926829268292687e-05,
      "loss": 1.9766,
      "step": 124
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.2865853658536585e-05,
      "loss": 1.9367,
      "step": 125
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.2804878048780487e-05,
      "loss": 1.8564,
      "step": 126
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.2743902439024393e-05,
      "loss": 1.9548,
      "step": 127
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.2682926829268295e-05,
      "loss": 1.8375,
      "step": 128
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.2621951219512194e-05,
      "loss": 1.8987,
      "step": 129
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.25609756097561e-05,
      "loss": 1.8881,
      "step": 130
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.25e-05,
      "loss": 1.8478,
      "step": 131
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.2439024390243904e-05,
      "loss": 1.9343,
      "step": 132
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.2378048780487806e-05,
      "loss": 1.9425,
      "step": 133
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.2317073170731708e-05,
      "loss": 1.8641,
      "step": 134
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.225609756097561e-05,
      "loss": 1.8896,
      "step": 135
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.2195121951219512e-05,
      "loss": 1.8852,
      "step": 136
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.2134146341463414e-05,
      "loss": 1.8599,
      "step": 137
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.2073170731707316e-05,
      "loss": 1.916,
      "step": 138
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.2012195121951222e-05,
      "loss": 1.9417,
      "step": 139
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.195121951219512e-05,
      "loss": 1.916,
      "step": 140
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.1890243902439023e-05,
      "loss": 1.7946,
      "step": 141
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.1829268292682928e-05,
      "loss": 1.8801,
      "step": 142
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.176829268292683e-05,
      "loss": 1.9045,
      "step": 143
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.1707317073170732e-05,
      "loss": 1.9059,
      "step": 144
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.1646341463414634e-05,
      "loss": 1.8927,
      "step": 145
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.1585365853658537e-05,
      "loss": 2.0249,
      "step": 146
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.1524390243902442e-05,
      "loss": 1.8718,
      "step": 147
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.146341463414634e-05,
      "loss": 1.9046,
      "step": 148
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.1402439024390243e-05,
      "loss": 1.9481,
      "step": 149
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.134146341463415e-05,
      "loss": 1.8769,
      "step": 150
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.128048780487805e-05,
      "loss": 1.753,
      "step": 151
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.121951219512195e-05,
      "loss": 1.9065,
      "step": 152
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.1158536585365855e-05,
      "loss": 1.947,
      "step": 153
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.1097560975609757e-05,
      "loss": 1.9284,
      "step": 154
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.103658536585366e-05,
      "loss": 1.8341,
      "step": 155
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.097560975609756e-05,
      "loss": 1.8711,
      "step": 156
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.0914634146341463e-05,
      "loss": 1.8183,
      "step": 157
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.0853658536585365e-05,
      "loss": 1.8935,
      "step": 158
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.079268292682927e-05,
      "loss": 1.8397,
      "step": 159
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.073170731707317e-05,
      "loss": 1.8481,
      "step": 160
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.067073170731707e-05,
      "loss": 1.8134,
      "step": 161
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.0609756097560977e-05,
      "loss": 1.9221,
      "step": 162
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.054878048780488e-05,
      "loss": 1.8334,
      "step": 163
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.048780487804878e-05,
      "loss": 1.899,
      "step": 164
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.0426829268292683e-05,
      "loss": 1.8769,
      "step": 165
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.0365853658536586e-05,
      "loss": 1.8427,
      "step": 166
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.030487804878049e-05,
      "loss": 1.8423,
      "step": 167
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.024390243902439e-05,
      "loss": 1.9441,
      "step": 168
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.0182926829268292e-05,
      "loss": 1.7726,
      "step": 169
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.0121951219512197e-05,
      "loss": 1.8858,
      "step": 170
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.00609756097561e-05,
      "loss": 1.8227,
      "step": 171
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 1.8622,
      "step": 172
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.9939024390243904e-05,
      "loss": 1.8685,
      "step": 173
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.9878048780487806e-05,
      "loss": 1.92,
      "step": 174
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.9817073170731708e-05,
      "loss": 1.925,
      "step": 175
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.975609756097561e-05,
      "loss": 1.8486,
      "step": 176
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.9695121951219512e-05,
      "loss": 1.8687,
      "step": 177
    },
    {
      "epoch": 4.13,
      "learning_rate": 1.9634146341463414e-05,
      "loss": 1.8679,
      "step": 178
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.957317073170732e-05,
      "loss": 1.8379,
      "step": 179
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.951219512195122e-05,
      "loss": 1.8829,
      "step": 180
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.945121951219512e-05,
      "loss": 1.924,
      "step": 181
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9390243902439026e-05,
      "loss": 1.9461,
      "step": 182
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.9329268292682928e-05,
      "loss": 1.8702,
      "step": 183
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.926829268292683e-05,
      "loss": 1.8518,
      "step": 184
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9207317073170733e-05,
      "loss": 1.9101,
      "step": 185
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9146341463414635e-05,
      "loss": 1.7935,
      "step": 186
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.908536585365854e-05,
      "loss": 1.8736,
      "step": 187
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.902439024390244e-05,
      "loss": 1.8667,
      "step": 188
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.896341463414634e-05,
      "loss": 1.8707,
      "step": 189
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8902439024390246e-05,
      "loss": 1.9036,
      "step": 190
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.884146341463415e-05,
      "loss": 1.7947,
      "step": 191
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8780487804878047e-05,
      "loss": 1.8098,
      "step": 192
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.8719512195121953e-05,
      "loss": 1.8323,
      "step": 193
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.8658536585365855e-05,
      "loss": 1.8219,
      "step": 194
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.8597560975609754e-05,
      "loss": 1.8358,
      "step": 195
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.853658536585366e-05,
      "loss": 1.873,
      "step": 196
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.847560975609756e-05,
      "loss": 1.8871,
      "step": 197
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.8414634146341463e-05,
      "loss": 1.8926,
      "step": 198
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.8353658536585365e-05,
      "loss": 1.8976,
      "step": 199
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.8292682926829268e-05,
      "loss": 1.8112,
      "step": 200
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.823170731707317e-05,
      "loss": 1.8891,
      "step": 201
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.8170731707317075e-05,
      "loss": 1.8788,
      "step": 202
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.8109756097560974e-05,
      "loss": 1.787,
      "step": 203
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.804878048780488e-05,
      "loss": 1.8208,
      "step": 204
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.798780487804878e-05,
      "loss": 1.9106,
      "step": 205
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.7926829268292684e-05,
      "loss": 1.8922,
      "step": 206
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.7865853658536586e-05,
      "loss": 1.9014,
      "step": 207
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.7804878048780488e-05,
      "loss": 1.8015,
      "step": 208
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.774390243902439e-05,
      "loss": 1.8836,
      "step": 209
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.7682926829268296e-05,
      "loss": 1.8423,
      "step": 210
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.7621951219512194e-05,
      "loss": 1.8525,
      "step": 211
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.7560975609756096e-05,
      "loss": 1.8726,
      "step": 212
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 1.8582,
      "step": 213
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.7439024390243904e-05,
      "loss": 1.8846,
      "step": 214
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.7378048780487803e-05,
      "loss": 1.8683,
      "step": 215
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.7317073170731708e-05,
      "loss": 1.8713,
      "step": 216
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.725609756097561e-05,
      "loss": 1.8359,
      "step": 217
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.7195121951219512e-05,
      "loss": 1.8415,
      "step": 218
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.7134146341463415e-05,
      "loss": 1.9431,
      "step": 219
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.7073170731707317e-05,
      "loss": 1.8503,
      "step": 220
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.701219512195122e-05,
      "loss": 1.8799,
      "step": 221
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.6951219512195124e-05,
      "loss": 1.8415,
      "step": 222
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.6890243902439023e-05,
      "loss": 1.8431,
      "step": 223
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.682926829268293e-05,
      "loss": 1.8891,
      "step": 224
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.676829268292683e-05,
      "loss": 1.8438,
      "step": 225
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.6707317073170733e-05,
      "loss": 1.9094,
      "step": 226
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.6646341463414635e-05,
      "loss": 1.8045,
      "step": 227
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.6585365853658537e-05,
      "loss": 1.9617,
      "step": 228
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.652439024390244e-05,
      "loss": 1.8788,
      "step": 229
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.6463414634146345e-05,
      "loss": 1.8641,
      "step": 230
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.6402439024390243e-05,
      "loss": 1.8191,
      "step": 231
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.6341463414634145e-05,
      "loss": 1.8844,
      "step": 232
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.628048780487805e-05,
      "loss": 1.8835,
      "step": 233
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.6219512195121953e-05,
      "loss": 1.8899,
      "step": 234
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.6158536585365852e-05,
      "loss": 1.7983,
      "step": 235
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.6097560975609757e-05,
      "loss": 1.8025,
      "step": 236
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.603658536585366e-05,
      "loss": 1.8975,
      "step": 237
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.597560975609756e-05,
      "loss": 1.817,
      "step": 238
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.5914634146341464e-05,
      "loss": 1.768,
      "step": 239
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.5853658536585366e-05,
      "loss": 1.8033,
      "step": 240
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.5792682926829268e-05,
      "loss": 1.8734,
      "step": 241
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.5731707317073173e-05,
      "loss": 1.8962,
      "step": 242
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.5670731707317072e-05,
      "loss": 1.8442,
      "step": 243
    },
    {
      "epoch": 5.66,
      "learning_rate": 1.5609756097560978e-05,
      "loss": 1.8266,
      "step": 244
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.554878048780488e-05,
      "loss": 1.8717,
      "step": 245
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.5487804878048782e-05,
      "loss": 1.9014,
      "step": 246
    },
    {
      "epoch": 5.73,
      "learning_rate": 1.5426829268292684e-05,
      "loss": 1.8647,
      "step": 247
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.5365853658536586e-05,
      "loss": 1.8806,
      "step": 248
    },
    {
      "epoch": 5.77,
      "learning_rate": 1.5304878048780488e-05,
      "loss": 1.8166,
      "step": 249
    },
    {
      "epoch": 5.8,
      "learning_rate": 1.5243902439024392e-05,
      "loss": 1.8547,
      "step": 250
    },
    {
      "epoch": 5.82,
      "learning_rate": 1.5182926829268294e-05,
      "loss": 1.855,
      "step": 251
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.5121951219512194e-05,
      "loss": 1.8207,
      "step": 252
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.5060975609756098e-05,
      "loss": 1.8739,
      "step": 253
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.5e-05,
      "loss": 1.9057,
      "step": 254
    },
    {
      "epoch": 5.91,
      "learning_rate": 1.4939024390243902e-05,
      "loss": 1.8652,
      "step": 255
    },
    {
      "epoch": 5.94,
      "learning_rate": 1.4878048780487805e-05,
      "loss": 1.8042,
      "step": 256
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.4817073170731707e-05,
      "loss": 1.8646,
      "step": 257
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.475609756097561e-05,
      "loss": 1.7727,
      "step": 258
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.4695121951219511e-05,
      "loss": 1.8401,
      "step": 259
    },
    {
      "epoch": 6.03,
      "learning_rate": 1.4634146341463415e-05,
      "loss": 1.8323,
      "step": 260
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.4573170731707317e-05,
      "loss": 1.798,
      "step": 261
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.451219512195122e-05,
      "loss": 1.8498,
      "step": 262
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.4451219512195121e-05,
      "loss": 1.8613,
      "step": 263
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.4390243902439025e-05,
      "loss": 1.8581,
      "step": 264
    },
    {
      "epoch": 6.14,
      "learning_rate": 1.4329268292682927e-05,
      "loss": 1.8183,
      "step": 265
    },
    {
      "epoch": 6.17,
      "learning_rate": 1.4268292682926829e-05,
      "loss": 1.8341,
      "step": 266
    },
    {
      "epoch": 6.19,
      "learning_rate": 1.4207317073170731e-05,
      "loss": 1.8379,
      "step": 267
    },
    {
      "epoch": 6.21,
      "learning_rate": 1.4146341463414635e-05,
      "loss": 1.8602,
      "step": 268
    },
    {
      "epoch": 6.24,
      "learning_rate": 1.4085365853658535e-05,
      "loss": 1.8839,
      "step": 269
    },
    {
      "epoch": 6.26,
      "learning_rate": 1.402439024390244e-05,
      "loss": 1.8391,
      "step": 270
    },
    {
      "epoch": 6.28,
      "learning_rate": 1.3963414634146341e-05,
      "loss": 1.8329,
      "step": 271
    },
    {
      "epoch": 6.31,
      "learning_rate": 1.3902439024390245e-05,
      "loss": 1.8749,
      "step": 272
    },
    {
      "epoch": 6.33,
      "learning_rate": 1.3841463414634146e-05,
      "loss": 1.8137,
      "step": 273
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.378048780487805e-05,
      "loss": 1.8471,
      "step": 274
    },
    {
      "epoch": 6.38,
      "learning_rate": 1.3719512195121952e-05,
      "loss": 1.8196,
      "step": 275
    },
    {
      "epoch": 6.4,
      "learning_rate": 1.3658536585365854e-05,
      "loss": 1.8122,
      "step": 276
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.3597560975609756e-05,
      "loss": 1.827,
      "step": 277
    },
    {
      "epoch": 6.45,
      "learning_rate": 1.353658536585366e-05,
      "loss": 1.8483,
      "step": 278
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.347560975609756e-05,
      "loss": 1.8586,
      "step": 279
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.3414634146341464e-05,
      "loss": 1.8376,
      "step": 280
    },
    {
      "epoch": 6.52,
      "learning_rate": 1.3353658536585366e-05,
      "loss": 1.9202,
      "step": 281
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.329268292682927e-05,
      "loss": 1.8513,
      "step": 282
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.323170731707317e-05,
      "loss": 1.8526,
      "step": 283
    },
    {
      "epoch": 6.59,
      "learning_rate": 1.3170731707317074e-05,
      "loss": 1.8625,
      "step": 284
    },
    {
      "epoch": 6.61,
      "learning_rate": 1.3109756097560976e-05,
      "loss": 1.8891,
      "step": 285
    },
    {
      "epoch": 6.63,
      "learning_rate": 1.3048780487804878e-05,
      "loss": 1.8396,
      "step": 286
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.298780487804878e-05,
      "loss": 1.8735,
      "step": 287
    },
    {
      "epoch": 6.68,
      "learning_rate": 1.2926829268292684e-05,
      "loss": 1.8588,
      "step": 288
    },
    {
      "epoch": 6.7,
      "learning_rate": 1.2865853658536585e-05,
      "loss": 1.8354,
      "step": 289
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.2804878048780488e-05,
      "loss": 1.7961,
      "step": 290
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.274390243902439e-05,
      "loss": 1.8739,
      "step": 291
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.2682926829268294e-05,
      "loss": 1.755,
      "step": 292
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.2621951219512195e-05,
      "loss": 1.819,
      "step": 293
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.2560975609756098e-05,
      "loss": 1.873,
      "step": 294
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.25e-05,
      "loss": 1.8856,
      "step": 295
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.2439024390243903e-05,
      "loss": 1.7957,
      "step": 296
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.2378048780487805e-05,
      "loss": 1.9105,
      "step": 297
    },
    {
      "epoch": 6.91,
      "learning_rate": 1.2317073170731709e-05,
      "loss": 1.9567,
      "step": 298
    },
    {
      "epoch": 6.93,
      "learning_rate": 1.2256097560975609e-05,
      "loss": 1.8302,
      "step": 299
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.2195121951219513e-05,
      "loss": 1.8012,
      "step": 300
    },
    {
      "epoch": 6.98,
      "learning_rate": 1.2134146341463415e-05,
      "loss": 1.8738,
      "step": 301
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.2073170731707319e-05,
      "loss": 1.8451,
      "step": 302
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.201219512195122e-05,
      "loss": 1.8725,
      "step": 303
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.1951219512195123e-05,
      "loss": 1.8076,
      "step": 304
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.1890243902439025e-05,
      "loss": 1.8252,
      "step": 305
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.1829268292682927e-05,
      "loss": 1.8556,
      "step": 306
    },
    {
      "epoch": 7.12,
      "learning_rate": 1.176829268292683e-05,
      "loss": 1.9182,
      "step": 307
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.1707317073170733e-05,
      "loss": 1.8325,
      "step": 308
    },
    {
      "epoch": 7.17,
      "learning_rate": 1.1646341463414634e-05,
      "loss": 1.8132,
      "step": 309
    },
    {
      "epoch": 7.19,
      "learning_rate": 1.1585365853658537e-05,
      "loss": 1.8424,
      "step": 310
    },
    {
      "epoch": 7.21,
      "learning_rate": 1.152439024390244e-05,
      "loss": 1.8617,
      "step": 311
    },
    {
      "epoch": 7.23,
      "learning_rate": 1.1463414634146343e-05,
      "loss": 1.7966,
      "step": 312
    },
    {
      "epoch": 7.26,
      "learning_rate": 1.1402439024390244e-05,
      "loss": 1.8628,
      "step": 313
    },
    {
      "epoch": 7.28,
      "learning_rate": 1.1341463414634148e-05,
      "loss": 1.8711,
      "step": 314
    },
    {
      "epoch": 7.3,
      "learning_rate": 1.128048780487805e-05,
      "loss": 1.8142,
      "step": 315
    },
    {
      "epoch": 7.33,
      "learning_rate": 1.1219512195121952e-05,
      "loss": 1.8658,
      "step": 316
    },
    {
      "epoch": 7.35,
      "learning_rate": 1.1158536585365854e-05,
      "loss": 1.8233,
      "step": 317
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.1097560975609756e-05,
      "loss": 1.8624,
      "step": 318
    },
    {
      "epoch": 7.4,
      "learning_rate": 1.1036585365853658e-05,
      "loss": 1.8298,
      "step": 319
    },
    {
      "epoch": 7.42,
      "learning_rate": 1.097560975609756e-05,
      "loss": 1.869,
      "step": 320
    },
    {
      "epoch": 7.44,
      "learning_rate": 1.0914634146341464e-05,
      "loss": 1.884,
      "step": 321
    },
    {
      "epoch": 7.47,
      "learning_rate": 1.0853658536585366e-05,
      "loss": 1.8007,
      "step": 322
    },
    {
      "epoch": 7.49,
      "learning_rate": 1.0792682926829268e-05,
      "loss": 1.8457,
      "step": 323
    },
    {
      "epoch": 7.51,
      "learning_rate": 1.073170731707317e-05,
      "loss": 1.8589,
      "step": 324
    },
    {
      "epoch": 7.54,
      "learning_rate": 1.0670731707317074e-05,
      "loss": 1.8211,
      "step": 325
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.0609756097560975e-05,
      "loss": 1.8594,
      "step": 326
    },
    {
      "epoch": 7.58,
      "learning_rate": 1.0548780487804878e-05,
      "loss": 1.8588,
      "step": 327
    },
    {
      "epoch": 7.61,
      "learning_rate": 1.048780487804878e-05,
      "loss": 1.845,
      "step": 328
    },
    {
      "epoch": 7.63,
      "learning_rate": 1.0426829268292683e-05,
      "loss": 1.7974,
      "step": 329
    },
    {
      "epoch": 7.65,
      "learning_rate": 1.0365853658536585e-05,
      "loss": 1.8571,
      "step": 330
    },
    {
      "epoch": 7.68,
      "learning_rate": 1.0304878048780489e-05,
      "loss": 1.7917,
      "step": 331
    },
    {
      "epoch": 7.7,
      "learning_rate": 1.024390243902439e-05,
      "loss": 1.8581,
      "step": 332
    },
    {
      "epoch": 7.72,
      "learning_rate": 1.0182926829268293e-05,
      "loss": 1.8404,
      "step": 333
    },
    {
      "epoch": 7.74,
      "learning_rate": 1.0121951219512195e-05,
      "loss": 1.8323,
      "step": 334
    },
    {
      "epoch": 7.77,
      "learning_rate": 1.0060975609756099e-05,
      "loss": 1.8368,
      "step": 335
    },
    {
      "epoch": 7.79,
      "learning_rate": 9.999999999999999e-06,
      "loss": 1.8568,
      "step": 336
    },
    {
      "epoch": 7.81,
      "learning_rate": 9.939024390243903e-06,
      "loss": 1.8771,
      "step": 337
    },
    {
      "epoch": 7.84,
      "learning_rate": 9.878048780487805e-06,
      "loss": 1.8877,
      "step": 338
    },
    {
      "epoch": 7.86,
      "learning_rate": 9.817073170731707e-06,
      "loss": 1.8309,
      "step": 339
    },
    {
      "epoch": 7.88,
      "learning_rate": 9.75609756097561e-06,
      "loss": 1.8153,
      "step": 340
    },
    {
      "epoch": 7.91,
      "learning_rate": 9.695121951219513e-06,
      "loss": 1.9367,
      "step": 341
    },
    {
      "epoch": 7.93,
      "learning_rate": 9.634146341463415e-06,
      "loss": 1.7258,
      "step": 342
    },
    {
      "epoch": 7.95,
      "learning_rate": 9.573170731707317e-06,
      "loss": 1.825,
      "step": 343
    },
    {
      "epoch": 7.98,
      "learning_rate": 9.51219512195122e-06,
      "loss": 1.8158,
      "step": 344
    },
    {
      "epoch": 8.0,
      "learning_rate": 9.451219512195123e-06,
      "loss": 1.8606,
      "step": 345
    },
    {
      "epoch": 8.02,
      "learning_rate": 9.390243902439024e-06,
      "loss": 1.881,
      "step": 346
    },
    {
      "epoch": 8.05,
      "learning_rate": 9.329268292682927e-06,
      "loss": 1.8339,
      "step": 347
    },
    {
      "epoch": 8.07,
      "learning_rate": 9.26829268292683e-06,
      "loss": 1.8034,
      "step": 348
    },
    {
      "epoch": 8.09,
      "learning_rate": 9.207317073170732e-06,
      "loss": 1.8334,
      "step": 349
    },
    {
      "epoch": 8.12,
      "learning_rate": 9.146341463414634e-06,
      "loss": 1.8411,
      "step": 350
    },
    {
      "epoch": 8.14,
      "learning_rate": 9.085365853658538e-06,
      "loss": 1.8157,
      "step": 351
    },
    {
      "epoch": 8.16,
      "learning_rate": 9.02439024390244e-06,
      "loss": 1.7685,
      "step": 352
    },
    {
      "epoch": 8.19,
      "learning_rate": 8.963414634146342e-06,
      "loss": 1.8318,
      "step": 353
    },
    {
      "epoch": 8.21,
      "learning_rate": 8.902439024390244e-06,
      "loss": 1.8124,
      "step": 354
    },
    {
      "epoch": 8.23,
      "learning_rate": 8.841463414634148e-06,
      "loss": 1.8656,
      "step": 355
    },
    {
      "epoch": 8.26,
      "learning_rate": 8.780487804878048e-06,
      "loss": 1.8696,
      "step": 356
    },
    {
      "epoch": 8.28,
      "learning_rate": 8.719512195121952e-06,
      "loss": 1.8821,
      "step": 357
    },
    {
      "epoch": 8.3,
      "learning_rate": 8.658536585365854e-06,
      "loss": 1.8023,
      "step": 358
    },
    {
      "epoch": 8.32,
      "learning_rate": 8.597560975609756e-06,
      "loss": 1.8643,
      "step": 359
    },
    {
      "epoch": 8.35,
      "learning_rate": 8.536585365853658e-06,
      "loss": 1.9178,
      "step": 360
    },
    {
      "epoch": 8.37,
      "learning_rate": 8.475609756097562e-06,
      "loss": 1.8703,
      "step": 361
    },
    {
      "epoch": 8.39,
      "learning_rate": 8.414634146341464e-06,
      "loss": 1.8673,
      "step": 362
    },
    {
      "epoch": 8.42,
      "learning_rate": 8.353658536585366e-06,
      "loss": 1.8394,
      "step": 363
    },
    {
      "epoch": 8.44,
      "learning_rate": 8.292682926829268e-06,
      "loss": 1.7786,
      "step": 364
    },
    {
      "epoch": 8.46,
      "learning_rate": 8.231707317073172e-06,
      "loss": 1.8853,
      "step": 365
    },
    {
      "epoch": 8.49,
      "learning_rate": 8.170731707317073e-06,
      "loss": 1.7903,
      "step": 366
    },
    {
      "epoch": 8.51,
      "learning_rate": 8.109756097560977e-06,
      "loss": 1.8189,
      "step": 367
    },
    {
      "epoch": 8.53,
      "learning_rate": 8.048780487804879e-06,
      "loss": 1.8252,
      "step": 368
    },
    {
      "epoch": 8.56,
      "learning_rate": 7.98780487804878e-06,
      "loss": 1.8291,
      "step": 369
    },
    {
      "epoch": 8.58,
      "learning_rate": 7.926829268292683e-06,
      "loss": 1.8915,
      "step": 370
    },
    {
      "epoch": 8.6,
      "learning_rate": 7.865853658536587e-06,
      "loss": 1.8891,
      "step": 371
    },
    {
      "epoch": 8.63,
      "learning_rate": 7.804878048780489e-06,
      "loss": 1.8618,
      "step": 372
    },
    {
      "epoch": 8.65,
      "learning_rate": 7.743902439024391e-06,
      "loss": 1.8479,
      "step": 373
    },
    {
      "epoch": 8.67,
      "learning_rate": 7.682926829268293e-06,
      "loss": 1.8519,
      "step": 374
    },
    {
      "epoch": 8.7,
      "learning_rate": 7.621951219512196e-06,
      "loss": 1.8327,
      "step": 375
    },
    {
      "epoch": 8.72,
      "learning_rate": 7.560975609756097e-06,
      "loss": 1.8108,
      "step": 376
    },
    {
      "epoch": 8.74,
      "learning_rate": 7.5e-06,
      "loss": 1.7837,
      "step": 377
    },
    {
      "epoch": 8.77,
      "learning_rate": 7.439024390243902e-06,
      "loss": 1.85,
      "step": 378
    },
    {
      "epoch": 8.79,
      "learning_rate": 7.378048780487805e-06,
      "loss": 1.8255,
      "step": 379
    },
    {
      "epoch": 8.81,
      "learning_rate": 7.317073170731707e-06,
      "loss": 1.8321,
      "step": 380
    },
    {
      "epoch": 8.83,
      "learning_rate": 7.25609756097561e-06,
      "loss": 1.8171,
      "step": 381
    },
    {
      "epoch": 8.86,
      "learning_rate": 7.1951219512195125e-06,
      "loss": 1.8386,
      "step": 382
    },
    {
      "epoch": 8.88,
      "learning_rate": 7.1341463414634146e-06,
      "loss": 1.8255,
      "step": 383
    },
    {
      "epoch": 8.9,
      "learning_rate": 7.0731707317073175e-06,
      "loss": 1.8649,
      "step": 384
    },
    {
      "epoch": 8.93,
      "learning_rate": 7.01219512195122e-06,
      "loss": 1.83,
      "step": 385
    },
    {
      "epoch": 8.95,
      "learning_rate": 6.951219512195123e-06,
      "loss": 1.7527,
      "step": 386
    },
    {
      "epoch": 8.97,
      "learning_rate": 6.890243902439025e-06,
      "loss": 1.9175,
      "step": 387
    },
    {
      "epoch": 9.0,
      "learning_rate": 6.829268292682927e-06,
      "loss": 1.8951,
      "step": 388
    },
    {
      "epoch": 9.02,
      "learning_rate": 6.76829268292683e-06,
      "loss": 1.8545,
      "step": 389
    },
    {
      "epoch": 9.04,
      "learning_rate": 6.707317073170732e-06,
      "loss": 1.9206,
      "step": 390
    },
    {
      "epoch": 9.07,
      "learning_rate": 6.646341463414635e-06,
      "loss": 1.8398,
      "step": 391
    },
    {
      "epoch": 9.09,
      "learning_rate": 6.585365853658537e-06,
      "loss": 1.7943,
      "step": 392
    },
    {
      "epoch": 9.11,
      "learning_rate": 6.524390243902439e-06,
      "loss": 1.8321,
      "step": 393
    },
    {
      "epoch": 9.14,
      "learning_rate": 6.463414634146342e-06,
      "loss": 1.7706,
      "step": 394
    },
    {
      "epoch": 9.16,
      "learning_rate": 6.402439024390244e-06,
      "loss": 1.9105,
      "step": 395
    },
    {
      "epoch": 9.18,
      "learning_rate": 6.341463414634147e-06,
      "loss": 1.8658,
      "step": 396
    },
    {
      "epoch": 9.21,
      "learning_rate": 6.280487804878049e-06,
      "loss": 1.8315,
      "step": 397
    },
    {
      "epoch": 9.23,
      "learning_rate": 6.219512195121951e-06,
      "loss": 1.909,
      "step": 398
    },
    {
      "epoch": 9.25,
      "learning_rate": 6.158536585365854e-06,
      "loss": 1.8242,
      "step": 399
    },
    {
      "epoch": 9.28,
      "learning_rate": 6.0975609756097564e-06,
      "loss": 1.8862,
      "step": 400
    },
    {
      "epoch": 9.3,
      "learning_rate": 6.036585365853659e-06,
      "loss": 1.8389,
      "step": 401
    },
    {
      "epoch": 9.32,
      "learning_rate": 5.9756097560975615e-06,
      "loss": 1.8099,
      "step": 402
    },
    {
      "epoch": 9.34,
      "learning_rate": 5.914634146341464e-06,
      "loss": 1.8162,
      "step": 403
    },
    {
      "epoch": 9.37,
      "learning_rate": 5.853658536585367e-06,
      "loss": 1.8323,
      "step": 404
    },
    {
      "epoch": 9.39,
      "learning_rate": 5.792682926829269e-06,
      "loss": 1.8445,
      "step": 405
    },
    {
      "epoch": 9.41,
      "learning_rate": 5.731707317073172e-06,
      "loss": 1.7648,
      "step": 406
    },
    {
      "epoch": 9.44,
      "learning_rate": 5.670731707317074e-06,
      "loss": 1.862,
      "step": 407
    },
    {
      "epoch": 9.46,
      "learning_rate": 5.609756097560976e-06,
      "loss": 1.818,
      "step": 408
    },
    {
      "epoch": 9.48,
      "learning_rate": 5.548780487804878e-06,
      "loss": 1.8372,
      "step": 409
    },
    {
      "epoch": 9.51,
      "learning_rate": 5.48780487804878e-06,
      "loss": 1.8508,
      "step": 410
    },
    {
      "epoch": 9.53,
      "learning_rate": 5.426829268292683e-06,
      "loss": 1.8789,
      "step": 411
    },
    {
      "epoch": 9.55,
      "learning_rate": 5.365853658536585e-06,
      "loss": 1.8296,
      "step": 412
    },
    {
      "epoch": 9.58,
      "learning_rate": 5.304878048780487e-06,
      "loss": 1.7921,
      "step": 413
    },
    {
      "epoch": 9.6,
      "learning_rate": 5.24390243902439e-06,
      "loss": 1.8492,
      "step": 414
    },
    {
      "epoch": 9.62,
      "learning_rate": 5.182926829268292e-06,
      "loss": 1.853,
      "step": 415
    },
    {
      "epoch": 9.65,
      "learning_rate": 5.121951219512195e-06,
      "loss": 1.883,
      "step": 416
    },
    {
      "epoch": 9.67,
      "learning_rate": 5.0609756097560974e-06,
      "loss": 1.7697,
      "step": 417
    },
    {
      "epoch": 9.69,
      "learning_rate": 4.9999999999999996e-06,
      "loss": 1.7551,
      "step": 418
    },
    {
      "epoch": 9.72,
      "learning_rate": 4.9390243902439025e-06,
      "loss": 1.8302,
      "step": 419
    },
    {
      "epoch": 9.74,
      "learning_rate": 4.878048780487805e-06,
      "loss": 1.8384,
      "step": 420
    },
    {
      "epoch": 9.76,
      "learning_rate": 4.817073170731708e-06,
      "loss": 1.7868,
      "step": 421
    },
    {
      "epoch": 9.79,
      "learning_rate": 4.75609756097561e-06,
      "loss": 1.912,
      "step": 422
    },
    {
      "epoch": 9.81,
      "learning_rate": 4.695121951219512e-06,
      "loss": 1.7863,
      "step": 423
    },
    {
      "epoch": 9.83,
      "learning_rate": 4.634146341463415e-06,
      "loss": 1.863,
      "step": 424
    },
    {
      "epoch": 9.86,
      "learning_rate": 4.573170731707317e-06,
      "loss": 1.8152,
      "step": 425
    },
    {
      "epoch": 9.88,
      "learning_rate": 4.51219512195122e-06,
      "loss": 1.8181,
      "step": 426
    },
    {
      "epoch": 9.9,
      "learning_rate": 4.451219512195122e-06,
      "loss": 1.8361,
      "step": 427
    },
    {
      "epoch": 9.92,
      "learning_rate": 4.390243902439024e-06,
      "loss": 1.9204,
      "step": 428
    },
    {
      "epoch": 9.95,
      "learning_rate": 4.329268292682927e-06,
      "loss": 1.8739,
      "step": 429
    },
    {
      "epoch": 9.97,
      "learning_rate": 4.268292682926829e-06,
      "loss": 1.8166,
      "step": 430
    },
    {
      "epoch": 9.99,
      "learning_rate": 4.207317073170732e-06,
      "loss": 1.8519,
      "step": 431
    },
    {
      "epoch": 10.02,
      "learning_rate": 4.146341463414634e-06,
      "loss": 1.7722,
      "step": 432
    },
    {
      "epoch": 10.04,
      "learning_rate": 4.085365853658536e-06,
      "loss": 1.8071,
      "step": 433
    },
    {
      "epoch": 10.06,
      "learning_rate": 4.024390243902439e-06,
      "loss": 1.8382,
      "step": 434
    },
    {
      "epoch": 10.09,
      "learning_rate": 3.9634146341463414e-06,
      "loss": 1.8579,
      "step": 435
    },
    {
      "epoch": 10.11,
      "learning_rate": 3.902439024390244e-06,
      "loss": 1.8477,
      "step": 436
    },
    {
      "epoch": 10.13,
      "learning_rate": 3.8414634146341465e-06,
      "loss": 1.8037,
      "step": 437
    },
    {
      "epoch": 10.16,
      "learning_rate": 3.7804878048780486e-06,
      "loss": 1.8286,
      "step": 438
    },
    {
      "epoch": 10.18,
      "learning_rate": 3.719512195121951e-06,
      "loss": 1.8407,
      "step": 439
    },
    {
      "epoch": 10.2,
      "learning_rate": 3.6585365853658537e-06,
      "loss": 1.8503,
      "step": 440
    },
    {
      "epoch": 10.23,
      "learning_rate": 3.5975609756097562e-06,
      "loss": 1.7788,
      "step": 441
    },
    {
      "epoch": 10.25,
      "learning_rate": 3.5365853658536588e-06,
      "loss": 1.8687,
      "step": 442
    },
    {
      "epoch": 10.27,
      "learning_rate": 3.4756097560975613e-06,
      "loss": 1.7877,
      "step": 443
    },
    {
      "epoch": 10.3,
      "learning_rate": 3.4146341463414634e-06,
      "loss": 1.8842,
      "step": 444
    },
    {
      "epoch": 10.32,
      "learning_rate": 3.353658536585366e-06,
      "loss": 1.7731,
      "step": 445
    },
    {
      "epoch": 10.34,
      "learning_rate": 3.2926829268292685e-06,
      "loss": 1.9421,
      "step": 446
    },
    {
      "epoch": 10.37,
      "learning_rate": 3.231707317073171e-06,
      "loss": 1.893,
      "step": 447
    },
    {
      "epoch": 10.39,
      "learning_rate": 3.1707317073170736e-06,
      "loss": 1.8234,
      "step": 448
    },
    {
      "epoch": 10.41,
      "learning_rate": 3.1097560975609757e-06,
      "loss": 1.7849,
      "step": 449
    },
    {
      "epoch": 10.43,
      "learning_rate": 3.0487804878048782e-06,
      "loss": 1.8656,
      "step": 450
    },
    {
      "epoch": 10.46,
      "learning_rate": 2.9878048780487808e-06,
      "loss": 1.9295,
      "step": 451
    },
    {
      "epoch": 10.48,
      "learning_rate": 2.9268292682926833e-06,
      "loss": 1.8294,
      "step": 452
    },
    {
      "epoch": 10.5,
      "learning_rate": 2.865853658536586e-06,
      "loss": 1.8313,
      "step": 453
    },
    {
      "epoch": 10.53,
      "learning_rate": 2.804878048780488e-06,
      "loss": 1.8892,
      "step": 454
    },
    {
      "epoch": 10.55,
      "learning_rate": 2.74390243902439e-06,
      "loss": 1.8483,
      "step": 455
    },
    {
      "epoch": 10.57,
      "learning_rate": 2.6829268292682926e-06,
      "loss": 1.7787,
      "step": 456
    },
    {
      "epoch": 10.6,
      "learning_rate": 2.621951219512195e-06,
      "loss": 1.8363,
      "step": 457
    },
    {
      "epoch": 10.62,
      "learning_rate": 2.5609756097560977e-06,
      "loss": 1.8556,
      "step": 458
    },
    {
      "epoch": 10.64,
      "learning_rate": 2.4999999999999998e-06,
      "loss": 1.8664,
      "step": 459
    },
    {
      "epoch": 10.67,
      "learning_rate": 2.4390243902439023e-06,
      "loss": 1.8107,
      "step": 460
    },
    {
      "epoch": 10.69,
      "learning_rate": 2.378048780487805e-06,
      "loss": 1.9115,
      "step": 461
    },
    {
      "epoch": 10.71,
      "learning_rate": 2.3170731707317074e-06,
      "loss": 1.8131,
      "step": 462
    },
    {
      "epoch": 10.74,
      "learning_rate": 2.25609756097561e-06,
      "loss": 1.8422,
      "step": 463
    },
    {
      "epoch": 10.76,
      "learning_rate": 2.195121951219512e-06,
      "loss": 1.7932,
      "step": 464
    },
    {
      "epoch": 10.78,
      "learning_rate": 2.1341463414634146e-06,
      "loss": 1.8752,
      "step": 465
    },
    {
      "epoch": 10.81,
      "learning_rate": 2.073170731707317e-06,
      "loss": 1.7811,
      "step": 466
    },
    {
      "epoch": 10.83,
      "learning_rate": 2.0121951219512197e-06,
      "loss": 1.7859,
      "step": 467
    },
    {
      "epoch": 10.85,
      "learning_rate": 1.951219512195122e-06,
      "loss": 1.8503,
      "step": 468
    },
    {
      "epoch": 10.88,
      "learning_rate": 1.8902439024390243e-06,
      "loss": 1.8373,
      "step": 469
    },
    {
      "epoch": 10.9,
      "learning_rate": 1.8292682926829268e-06,
      "loss": 1.9218,
      "step": 470
    },
    {
      "epoch": 10.92,
      "learning_rate": 1.7682926829268294e-06,
      "loss": 1.7876,
      "step": 471
    },
    {
      "epoch": 10.94,
      "learning_rate": 1.7073170731707317e-06,
      "loss": 1.859,
      "step": 472
    },
    {
      "epoch": 10.97,
      "learning_rate": 1.6463414634146342e-06,
      "loss": 1.8083,
      "step": 473
    },
    {
      "epoch": 10.99,
      "learning_rate": 1.5853658536585368e-06,
      "loss": 1.859,
      "step": 474
    },
    {
      "epoch": 11.01,
      "learning_rate": 1.5243902439024391e-06,
      "loss": 1.7685,
      "step": 475
    },
    {
      "epoch": 11.04,
      "learning_rate": 1.4634146341463416e-06,
      "loss": 1.7692,
      "step": 476
    },
    {
      "epoch": 11.06,
      "learning_rate": 1.402439024390244e-06,
      "loss": 1.8202,
      "step": 477
    },
    {
      "epoch": 11.08,
      "learning_rate": 1.3414634146341463e-06,
      "loss": 1.8052,
      "step": 478
    },
    {
      "epoch": 11.11,
      "learning_rate": 1.2804878048780488e-06,
      "loss": 1.8734,
      "step": 479
    },
    {
      "epoch": 11.13,
      "learning_rate": 1.2195121951219512e-06,
      "loss": 1.7973,
      "step": 480
    },
    {
      "epoch": 11.15,
      "learning_rate": 1.1585365853658537e-06,
      "loss": 1.8462,
      "step": 481
    },
    {
      "epoch": 11.18,
      "learning_rate": 1.097560975609756e-06,
      "loss": 1.8119,
      "step": 482
    },
    {
      "epoch": 11.2,
      "learning_rate": 1.0365853658536586e-06,
      "loss": 1.9105,
      "step": 483
    },
    {
      "epoch": 11.22,
      "learning_rate": 9.75609756097561e-07,
      "loss": 1.8709,
      "step": 484
    },
    {
      "epoch": 11.25,
      "learning_rate": 9.146341463414634e-07,
      "loss": 1.871,
      "step": 485
    },
    {
      "epoch": 11.27,
      "learning_rate": 8.536585365853659e-07,
      "loss": 1.7683,
      "step": 486
    },
    {
      "epoch": 11.29,
      "learning_rate": 7.926829268292684e-07,
      "loss": 1.8853,
      "step": 487
    },
    {
      "epoch": 11.32,
      "learning_rate": 7.317073170731708e-07,
      "loss": 1.7629,
      "step": 488
    },
    {
      "epoch": 11.34,
      "learning_rate": 6.707317073170731e-07,
      "loss": 1.7865,
      "step": 489
    },
    {
      "epoch": 11.36,
      "learning_rate": 6.097560975609756e-07,
      "loss": 1.9186,
      "step": 490
    },
    {
      "epoch": 11.39,
      "learning_rate": 5.48780487804878e-07,
      "loss": 1.8269,
      "step": 491
    },
    {
      "epoch": 11.41,
      "learning_rate": 4.878048780487805e-07,
      "loss": 1.8538,
      "step": 492
    },
    {
      "epoch": 11.43,
      "learning_rate": 4.2682926829268293e-07,
      "loss": 1.8228,
      "step": 493
    },
    {
      "epoch": 11.46,
      "learning_rate": 3.658536585365854e-07,
      "loss": 1.8196,
      "step": 494
    },
    {
      "epoch": 11.48,
      "learning_rate": 3.048780487804878e-07,
      "loss": 1.8104,
      "step": 495
    },
    {
      "epoch": 11.5,
      "learning_rate": 2.439024390243903e-07,
      "loss": 1.8221,
      "step": 496
    },
    {
      "epoch": 11.52,
      "learning_rate": 1.829268292682927e-07,
      "loss": 1.8234,
      "step": 497
    },
    {
      "epoch": 11.55,
      "learning_rate": 1.2195121951219514e-07,
      "loss": 1.8229,
      "step": 498
    },
    {
      "epoch": 11.57,
      "learning_rate": 6.097560975609757e-08,
      "loss": 1.9636,
      "step": 499
    },
    {
      "epoch": 11.59,
      "learning_rate": 0.0,
      "loss": 1.8427,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 12,
  "save_steps": 50,
  "total_flos": 1.3295810584164434e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}