{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.13,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.8383,
      "step": 2
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.8509,
      "step": 4
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00016,
      "loss": 0.8373,
      "step": 6
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.8067,
      "step": 8
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0002666666666666667,
      "loss": 0.5774,
      "step": 10
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00032,
      "loss": 0.5267,
      "step": 12
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0003733333333333334,
      "loss": 0.551,
      "step": 14
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00039703703703703705,
      "loss": 0.4659,
      "step": 16
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0003911111111111111,
      "loss": 0.4995,
      "step": 18
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0003851851851851852,
      "loss": 0.5013,
      "step": 20
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0003792592592592593,
      "loss": 0.3762,
      "step": 22
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0003733333333333334,
      "loss": 0.3964,
      "step": 24
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00036740740740740744,
      "loss": 0.4134,
      "step": 26
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0003614814814814815,
      "loss": 0.3222,
      "step": 28
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00035555555555555557,
      "loss": 0.2872,
      "step": 30
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00034962962962962964,
      "loss": 0.3529,
      "step": 32
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0003437037037037037,
      "loss": 0.282,
      "step": 34
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00033777777777777777,
      "loss": 0.2544,
      "step": 36
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00033185185185185184,
      "loss": 0.212,
      "step": 38
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0003259259259259259,
      "loss": 0.1252,
      "step": 40
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00032,
      "loss": 0.1933,
      "step": 42
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0003140740740740741,
      "loss": 0.1614,
      "step": 44
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.00030814814814814816,
      "loss": 0.1694,
      "step": 46
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0003022222222222222,
      "loss": 0.1645,
      "step": 48
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0002962962962962963,
      "loss": 0.1532,
      "step": 50
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.0002903703703703704,
      "loss": 0.1115,
      "step": 52
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.0002844444444444445,
      "loss": 0.0873,
      "step": 54
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.00027851851851851855,
      "loss": 0.1409,
      "step": 56
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.0002725925925925926,
      "loss": 0.1363,
      "step": 58
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0002666666666666667,
      "loss": 0.12,
      "step": 60
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.0002607407407407408,
      "loss": 0.0858,
      "step": 62
    },
    {
      "epoch": 4.27,
      "learning_rate": 0.0002548148148148148,
      "loss": 0.1151,
      "step": 64
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.0631,
      "step": 66
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00024296296296296297,
      "loss": 0.1468,
      "step": 68
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.00023703703703703704,
      "loss": 0.1032,
      "step": 70
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.0002311111111111111,
      "loss": 0.0626,
      "step": 72
    },
    {
      "epoch": 4.93,
      "learning_rate": 0.0002251851851851852,
      "loss": 0.0551,
      "step": 74
    },
    {
      "epoch": 5.07,
      "learning_rate": 0.00021925925925925927,
      "loss": 0.0717,
      "step": 76
    },
    {
      "epoch": 5.2,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.0711,
      "step": 78
    },
    {
      "epoch": 5.33,
      "learning_rate": 0.0002074074074074074,
      "loss": 0.0717,
      "step": 80
    },
    {
      "epoch": 5.47,
      "learning_rate": 0.00020148148148148147,
      "loss": 0.0886,
      "step": 82
    },
    {
      "epoch": 5.6,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.0527,
      "step": 84
    },
    {
      "epoch": 5.73,
      "learning_rate": 0.00018962962962962965,
      "loss": 0.0633,
      "step": 86
    },
    {
      "epoch": 5.87,
      "learning_rate": 0.00018370370370370372,
      "loss": 0.0424,
      "step": 88
    },
    {
      "epoch": 6.0,
      "learning_rate": 0.00017777777777777779,
      "loss": 0.0431,
      "step": 90
    },
    {
      "epoch": 6.13,
      "learning_rate": 0.00017185185185185185,
      "loss": 0.0538,
      "step": 92
    },
    {
      "epoch": 6.27,
      "learning_rate": 0.00016592592592592592,
      "loss": 0.0526,
      "step": 94
    },
    {
      "epoch": 6.4,
      "learning_rate": 0.00016,
      "loss": 0.055,
      "step": 96
    },
    {
      "epoch": 6.53,
      "learning_rate": 0.00015407407407407408,
      "loss": 0.0518,
      "step": 98
    },
    {
      "epoch": 6.67,
      "learning_rate": 0.00014814814814814815,
      "loss": 0.0484,
      "step": 100
    },
    {
      "epoch": 6.8,
      "learning_rate": 0.00014222222222222224,
      "loss": 0.0526,
      "step": 102
    },
    {
      "epoch": 6.93,
      "learning_rate": 0.0001362962962962963,
      "loss": 0.0544,
      "step": 104
    },
    {
      "epoch": 7.07,
      "learning_rate": 0.0001303703703703704,
      "loss": 0.036,
      "step": 106
    },
    {
      "epoch": 7.2,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.0475,
      "step": 108
    },
    {
      "epoch": 7.33,
      "learning_rate": 0.00011851851851851852,
      "loss": 0.0466,
      "step": 110
    },
    {
      "epoch": 7.47,
      "learning_rate": 0.0001125925925925926,
      "loss": 0.0416,
      "step": 112
    },
    {
      "epoch": 7.6,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.0387,
      "step": 114
    },
    {
      "epoch": 7.73,
      "learning_rate": 0.00010074074074074073,
      "loss": 0.0494,
      "step": 116
    },
    {
      "epoch": 7.87,
      "learning_rate": 9.481481481481483e-05,
      "loss": 0.0451,
      "step": 118
    },
    {
      "epoch": 8.0,
      "learning_rate": 8.888888888888889e-05,
      "loss": 0.0348,
      "step": 120
    },
    {
      "epoch": 8.13,
      "learning_rate": 8.296296296296296e-05,
      "loss": 0.0487,
      "step": 122
    },
    {
      "epoch": 8.27,
      "learning_rate": 7.703703703703704e-05,
      "loss": 0.0331,
      "step": 124
    },
    {
      "epoch": 8.4,
      "learning_rate": 7.111111111111112e-05,
      "loss": 0.0261,
      "step": 126
    },
    {
      "epoch": 8.53,
      "learning_rate": 6.51851851851852e-05,
      "loss": 0.0353,
      "step": 128
    },
    {
      "epoch": 8.67,
      "learning_rate": 5.925925925925926e-05,
      "loss": 0.0548,
      "step": 130
    },
    {
      "epoch": 8.8,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.0324,
      "step": 132
    },
    {
      "epoch": 8.93,
      "learning_rate": 4.740740740740741e-05,
      "loss": 0.0362,
      "step": 134
    },
    {
      "epoch": 9.07,
      "learning_rate": 4.148148148148148e-05,
      "loss": 0.0258,
      "step": 136
    },
    {
      "epoch": 9.2,
      "learning_rate": 3.555555555555556e-05,
      "loss": 0.0328,
      "step": 138
    },
    {
      "epoch": 9.33,
      "learning_rate": 2.962962962962963e-05,
      "loss": 0.0325,
      "step": 140
    },
    {
      "epoch": 9.47,
      "learning_rate": 2.3703703703703707e-05,
      "loss": 0.0245,
      "step": 142
    },
    {
      "epoch": 9.6,
      "learning_rate": 1.777777777777778e-05,
      "loss": 0.0294,
      "step": 144
    },
    {
      "epoch": 9.73,
      "learning_rate": 1.1851851851851853e-05,
      "loss": 0.0439,
      "step": 146
    },
    {
      "epoch": 9.87,
      "learning_rate": 5.925925925925927e-06,
      "loss": 0.0412,
      "step": 148
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 0.0352,
      "step": 150
    }
  ],
  "max_steps": 150,
  "num_train_epochs": 10,
  "total_flos": 2.407321351225344e+16,
  "trial_name": null,
  "trial_params": null
}