{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 500,
  "global_step": 160,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 2.1504,
      "step": 1
    },
    {
      "epoch": 0.25,
      "learning_rate": 4e-05,
      "loss": 2.3459,
      "step": 2
    },
    {
      "epoch": 0.38,
      "learning_rate": 6e-05,
      "loss": 2.222,
      "step": 3
    },
    {
      "epoch": 0.5,
      "learning_rate": 8e-05,
      "loss": 2.3004,
      "step": 4
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001,
      "loss": 2.2496,
      "step": 5
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.935483870967742e-05,
      "loss": 1.8725,
      "step": 6
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.870967741935484e-05,
      "loss": 2.1768,
      "step": 7
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.806451612903226e-05,
      "loss": 2.1073,
      "step": 8
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.741935483870968e-05,
      "loss": 1.8229,
      "step": 9
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.677419354838711e-05,
      "loss": 1.6071,
      "step": 10
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.612903225806452e-05,
      "loss": 1.8617,
      "step": 11
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.548387096774195e-05,
      "loss": 1.7659,
      "step": 12
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.483870967741936e-05,
      "loss": 1.533,
      "step": 13
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.419354838709677e-05,
      "loss": 1.4105,
      "step": 14
    },
    {
      "epoch": 1.88,
      "learning_rate": 9.35483870967742e-05,
      "loss": 1.4407,
      "step": 15
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.35483870967742e-05,
      "loss": 1.3275,
      "step": 16
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.290322580645162e-05,
      "loss": 1.2213,
      "step": 17
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.225806451612904e-05,
      "loss": 1.137,
      "step": 18
    },
    {
      "epoch": 2.38,
      "learning_rate": 9.161290322580646e-05,
      "loss": 1.4059,
      "step": 19
    },
    {
      "epoch": 2.5,
      "learning_rate": 9.096774193548387e-05,
      "loss": 1.4018,
      "step": 20
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.032258064516129e-05,
      "loss": 1.4079,
      "step": 21
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.967741935483871e-05,
      "loss": 1.3643,
      "step": 22
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.903225806451614e-05,
      "loss": 1.394,
      "step": 23
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.838709677419355e-05,
      "loss": 1.085,
      "step": 24
    },
    {
      "epoch": 3.12,
      "learning_rate": 8.774193548387098e-05,
      "loss": 1.2241,
      "step": 25
    },
    {
      "epoch": 3.25,
      "learning_rate": 8.709677419354839e-05,
      "loss": 1.2023,
      "step": 26
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.645161290322581e-05,
      "loss": 1.2069,
      "step": 27
    },
    {
      "epoch": 3.5,
      "learning_rate": 8.580645161290323e-05,
      "loss": 0.8765,
      "step": 28
    },
    {
      "epoch": 3.62,
      "learning_rate": 8.516129032258064e-05,
      "loss": 1.0117,
      "step": 29
    },
    {
      "epoch": 3.75,
      "learning_rate": 8.451612903225808e-05,
      "loss": 1.0511,
      "step": 30
    },
    {
      "epoch": 3.88,
      "learning_rate": 8.387096774193549e-05,
      "loss": 0.9173,
      "step": 31
    },
    {
      "epoch": 4.0,
      "learning_rate": 8.32258064516129e-05,
      "loss": 0.8711,
      "step": 32
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.258064516129033e-05,
      "loss": 0.8497,
      "step": 33
    },
    {
      "epoch": 4.25,
      "learning_rate": 8.193548387096774e-05,
      "loss": 0.9193,
      "step": 34
    },
    {
      "epoch": 4.38,
      "learning_rate": 8.129032258064517e-05,
      "loss": 0.768,
      "step": 35
    },
    {
      "epoch": 4.5,
      "learning_rate": 8.064516129032258e-05,
      "loss": 0.6524,
      "step": 36
    },
    {
      "epoch": 4.62,
      "learning_rate": 8e-05,
      "loss": 0.8045,
      "step": 37
    },
    {
      "epoch": 4.75,
      "learning_rate": 7.935483870967743e-05,
      "loss": 0.7216,
      "step": 38
    },
    {
      "epoch": 4.88,
      "learning_rate": 7.870967741935484e-05,
      "loss": 0.5946,
      "step": 39
    },
    {
      "epoch": 5.0,
      "learning_rate": 7.806451612903226e-05,
      "loss": 0.7398,
      "step": 40
    },
    {
      "epoch": 5.12,
      "learning_rate": 7.741935483870968e-05,
      "loss": 0.5371,
      "step": 41
    },
    {
      "epoch": 5.25,
      "learning_rate": 7.67741935483871e-05,
      "loss": 0.5425,
      "step": 42
    },
    {
      "epoch": 5.38,
      "learning_rate": 7.612903225806451e-05,
      "loss": 0.4687,
      "step": 43
    },
    {
      "epoch": 5.5,
      "learning_rate": 7.548387096774195e-05,
      "loss": 0.4539,
      "step": 44
    },
    {
      "epoch": 5.62,
      "learning_rate": 7.483870967741936e-05,
      "loss": 0.3761,
      "step": 45
    },
    {
      "epoch": 5.75,
      "learning_rate": 7.419354838709677e-05,
      "loss": 0.3375,
      "step": 46
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.35483870967742e-05,
      "loss": 0.2759,
      "step": 47
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.290322580645161e-05,
      "loss": 0.3232,
      "step": 48
    },
    {
      "epoch": 6.12,
      "learning_rate": 7.225806451612904e-05,
      "loss": 0.2162,
      "step": 49
    },
    {
      "epoch": 6.25,
      "learning_rate": 7.161290322580646e-05,
      "loss": 0.1798,
      "step": 50
    },
    {
      "epoch": 6.38,
      "learning_rate": 7.096774193548388e-05,
      "loss": 0.1543,
      "step": 51
    },
    {
      "epoch": 6.5,
      "learning_rate": 7.03225806451613e-05,
      "loss": 0.239,
      "step": 52
    },
    {
      "epoch": 6.62,
      "learning_rate": 6.967741935483871e-05,
      "loss": 0.2297,
      "step": 53
    },
    {
      "epoch": 6.75,
      "learning_rate": 6.903225806451613e-05,
      "loss": 0.2006,
      "step": 54
    },
    {
      "epoch": 6.88,
      "learning_rate": 6.838709677419355e-05,
      "loss": 0.1463,
      "step": 55
    },
    {
      "epoch": 7.0,
      "learning_rate": 6.774193548387096e-05,
      "loss": 0.2489,
      "step": 56
    },
    {
      "epoch": 7.12,
      "learning_rate": 6.709677419354839e-05,
      "loss": 0.1289,
      "step": 57
    },
    {
      "epoch": 7.25,
      "learning_rate": 6.645161290322582e-05,
      "loss": 0.1069,
      "step": 58
    },
    {
      "epoch": 7.38,
      "learning_rate": 6.580645161290323e-05,
      "loss": 0.092,
      "step": 59
    },
    {
      "epoch": 7.5,
      "learning_rate": 6.516129032258065e-05,
      "loss": 0.0909,
      "step": 60
    },
    {
      "epoch": 7.62,
      "learning_rate": 6.451612903225807e-05,
      "loss": 0.1182,
      "step": 61
    },
    {
      "epoch": 7.75,
      "learning_rate": 6.387096774193548e-05,
      "loss": 0.0968,
      "step": 62
    },
    {
      "epoch": 7.88,
      "learning_rate": 6.32258064516129e-05,
      "loss": 0.0951,
      "step": 63
    },
    {
      "epoch": 8.0,
      "learning_rate": 6.258064516129033e-05,
      "loss": 0.0895,
      "step": 64
    },
    {
      "epoch": 8.12,
      "learning_rate": 6.193548387096774e-05,
      "loss": 0.0821,
      "step": 65
    },
    {
      "epoch": 8.25,
      "learning_rate": 6.129032258064517e-05,
      "loss": 0.0888,
      "step": 66
    },
    {
      "epoch": 8.38,
      "learning_rate": 6.064516129032258e-05,
      "loss": 0.0455,
      "step": 67
    },
    {
      "epoch": 8.5,
      "learning_rate": 6e-05,
      "loss": 0.0482,
      "step": 68
    },
    {
      "epoch": 8.62,
      "learning_rate": 5.935483870967742e-05,
      "loss": 0.0634,
      "step": 69
    },
    {
      "epoch": 8.75,
      "learning_rate": 5.870967741935483e-05,
      "loss": 0.06,
      "step": 70
    },
    {
      "epoch": 8.88,
      "learning_rate": 5.8064516129032266e-05,
      "loss": 0.0598,
      "step": 71
    },
    {
      "epoch": 9.0,
      "learning_rate": 5.7419354838709685e-05,
      "loss": 0.0679,
      "step": 72
    },
    {
      "epoch": 9.12,
      "learning_rate": 5.67741935483871e-05,
      "loss": 0.0327,
      "step": 73
    },
    {
      "epoch": 9.25,
      "learning_rate": 5.612903225806452e-05,
      "loss": 0.0443,
      "step": 74
    },
    {
      "epoch": 9.38,
      "learning_rate": 5.5483870967741936e-05,
      "loss": 0.0393,
      "step": 75
    },
    {
      "epoch": 9.5,
      "learning_rate": 5.4838709677419355e-05,
      "loss": 0.0396,
      "step": 76
    },
    {
      "epoch": 9.62,
      "learning_rate": 5.419354838709678e-05,
      "loss": 0.0483,
      "step": 77
    },
    {
      "epoch": 9.75,
      "learning_rate": 5.35483870967742e-05,
      "loss": 0.0541,
      "step": 78
    },
    {
      "epoch": 9.88,
      "learning_rate": 5.290322580645162e-05,
      "loss": 0.0455,
      "step": 79
    },
    {
      "epoch": 10.0,
      "learning_rate": 5.225806451612903e-05,
      "loss": 0.0392,
      "step": 80
    },
    {
      "epoch": 10.12,
      "learning_rate": 5.161290322580645e-05,
      "loss": 0.028,
      "step": 81
    },
    {
      "epoch": 10.25,
      "learning_rate": 5.096774193548387e-05,
      "loss": 0.0393,
      "step": 82
    },
    {
      "epoch": 10.38,
      "learning_rate": 5.032258064516129e-05,
      "loss": 0.0343,
      "step": 83
    },
    {
      "epoch": 10.5,
      "learning_rate": 4.967741935483871e-05,
      "loss": 0.026,
      "step": 84
    },
    {
      "epoch": 10.62,
      "learning_rate": 4.903225806451613e-05,
      "loss": 0.0258,
      "step": 85
    },
    {
      "epoch": 10.75,
      "learning_rate": 4.8387096774193554e-05,
      "loss": 0.024,
      "step": 86
    },
    {
      "epoch": 10.88,
      "learning_rate": 4.774193548387097e-05,
      "loss": 0.0457,
      "step": 87
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.7096774193548385e-05,
      "loss": 0.0348,
      "step": 88
    },
    {
      "epoch": 11.12,
      "learning_rate": 4.645161290322581e-05,
      "loss": 0.0277,
      "step": 89
    },
    {
      "epoch": 11.25,
      "learning_rate": 4.580645161290323e-05,
      "loss": 0.0247,
      "step": 90
    },
    {
      "epoch": 11.38,
      "learning_rate": 4.516129032258064e-05,
      "loss": 0.0331,
      "step": 91
    },
    {
      "epoch": 11.5,
      "learning_rate": 4.451612903225807e-05,
      "loss": 0.02,
      "step": 92
    },
    {
      "epoch": 11.62,
      "learning_rate": 4.387096774193549e-05,
      "loss": 0.0259,
      "step": 93
    },
    {
      "epoch": 11.75,
      "learning_rate": 4.322580645161291e-05,
      "loss": 0.0229,
      "step": 94
    },
    {
      "epoch": 11.88,
      "learning_rate": 4.258064516129032e-05,
      "loss": 0.0326,
      "step": 95
    },
    {
      "epoch": 12.0,
      "learning_rate": 4.1935483870967746e-05,
      "loss": 0.0237,
      "step": 96
    },
    {
      "epoch": 12.12,
      "learning_rate": 4.1290322580645165e-05,
      "loss": 0.0259,
      "step": 97
    },
    {
      "epoch": 12.25,
      "learning_rate": 4.0645161290322584e-05,
      "loss": 0.0216,
      "step": 98
    },
    {
      "epoch": 12.38,
      "learning_rate": 4e-05,
      "loss": 0.0265,
      "step": 99
    },
    {
      "epoch": 12.5,
      "learning_rate": 3.935483870967742e-05,
      "loss": 0.0229,
      "step": 100
    },
    {
      "epoch": 12.62,
      "learning_rate": 3.870967741935484e-05,
      "loss": 0.0185,
      "step": 101
    },
    {
      "epoch": 12.75,
      "learning_rate": 3.8064516129032254e-05,
      "loss": 0.0228,
      "step": 102
    },
    {
      "epoch": 12.88,
      "learning_rate": 3.741935483870968e-05,
      "loss": 0.0246,
      "step": 103
    },
    {
      "epoch": 13.0,
      "learning_rate": 3.67741935483871e-05,
      "loss": 0.0205,
      "step": 104
    },
    {
      "epoch": 13.12,
      "learning_rate": 3.612903225806452e-05,
      "loss": 0.0197,
      "step": 105
    },
    {
      "epoch": 13.25,
      "learning_rate": 3.548387096774194e-05,
      "loss": 0.0213,
      "step": 106
    },
    {
      "epoch": 13.38,
      "learning_rate": 3.483870967741936e-05,
      "loss": 0.0173,
      "step": 107
    },
    {
      "epoch": 13.5,
      "learning_rate": 3.4193548387096776e-05,
      "loss": 0.0244,
      "step": 108
    },
    {
      "epoch": 13.62,
      "learning_rate": 3.3548387096774195e-05,
      "loss": 0.0193,
      "step": 109
    },
    {
      "epoch": 13.75,
      "learning_rate": 3.2903225806451614e-05,
      "loss": 0.0246,
      "step": 110
    },
    {
      "epoch": 13.88,
      "learning_rate": 3.2258064516129034e-05,
      "loss": 0.0276,
      "step": 111
    },
    {
      "epoch": 14.0,
      "learning_rate": 3.161290322580645e-05,
      "loss": 0.0234,
      "step": 112
    },
    {
      "epoch": 14.12,
      "learning_rate": 3.096774193548387e-05,
      "loss": 0.0218,
      "step": 113
    },
    {
      "epoch": 14.25,
      "learning_rate": 3.032258064516129e-05,
      "loss": 0.0212,
      "step": 114
    },
    {
      "epoch": 14.38,
      "learning_rate": 2.967741935483871e-05,
      "loss": 0.0177,
      "step": 115
    },
    {
      "epoch": 14.5,
      "learning_rate": 2.9032258064516133e-05,
      "loss": 0.0221,
      "step": 116
    },
    {
      "epoch": 14.62,
      "learning_rate": 2.838709677419355e-05,
      "loss": 0.0211,
      "step": 117
    },
    {
      "epoch": 14.75,
      "learning_rate": 2.7741935483870968e-05,
      "loss": 0.0235,
      "step": 118
    },
    {
      "epoch": 14.88,
      "learning_rate": 2.709677419354839e-05,
      "loss": 0.0205,
      "step": 119
    },
    {
      "epoch": 15.0,
      "learning_rate": 2.645161290322581e-05,
      "loss": 0.0205,
      "step": 120
    },
    {
      "epoch": 15.12,
      "learning_rate": 2.5806451612903226e-05,
      "loss": 0.0194,
      "step": 121
    },
    {
      "epoch": 15.25,
      "learning_rate": 2.5161290322580645e-05,
      "loss": 0.0209,
      "step": 122
    },
    {
      "epoch": 15.38,
      "learning_rate": 2.4516129032258064e-05,
      "loss": 0.0186,
      "step": 123
    },
    {
      "epoch": 15.5,
      "learning_rate": 2.3870967741935486e-05,
      "loss": 0.019,
      "step": 124
    },
    {
      "epoch": 15.62,
      "learning_rate": 2.3225806451612906e-05,
      "loss": 0.0199,
      "step": 125
    },
    {
      "epoch": 15.75,
      "learning_rate": 2.258064516129032e-05,
      "loss": 0.0253,
      "step": 126
    },
    {
      "epoch": 15.88,
      "learning_rate": 2.1935483870967744e-05,
      "loss": 0.0238,
      "step": 127
    },
    {
      "epoch": 16.0,
      "learning_rate": 2.129032258064516e-05,
      "loss": 0.0243,
      "step": 128
    },
    {
      "epoch": 16.12,
      "learning_rate": 2.0645161290322582e-05,
      "loss": 0.0211,
      "step": 129
    },
    {
      "epoch": 16.25,
      "learning_rate": 2e-05,
      "loss": 0.0193,
      "step": 130
    },
    {
      "epoch": 16.38,
      "learning_rate": 1.935483870967742e-05,
      "loss": 0.0172,
      "step": 131
    },
    {
      "epoch": 16.5,
      "learning_rate": 1.870967741935484e-05,
      "loss": 0.018,
      "step": 132
    },
    {
      "epoch": 16.62,
      "learning_rate": 1.806451612903226e-05,
      "loss": 0.0208,
      "step": 133
    },
    {
      "epoch": 16.75,
      "learning_rate": 1.741935483870968e-05,
      "loss": 0.0221,
      "step": 134
    },
    {
      "epoch": 16.88,
      "learning_rate": 1.6774193548387098e-05,
      "loss": 0.023,
      "step": 135
    },
    {
      "epoch": 17.0,
      "learning_rate": 1.6129032258064517e-05,
      "loss": 0.0242,
      "step": 136
    },
    {
      "epoch": 17.12,
      "learning_rate": 1.5483870967741936e-05,
      "loss": 0.0197,
      "step": 137
    },
    {
      "epoch": 17.25,
      "learning_rate": 1.4838709677419355e-05,
      "loss": 0.0173,
      "step": 138
    },
    {
      "epoch": 17.38,
      "learning_rate": 1.4193548387096774e-05,
      "loss": 0.0196,
      "step": 139
    },
    {
      "epoch": 17.5,
      "learning_rate": 1.3548387096774195e-05,
      "loss": 0.0219,
      "step": 140
    },
    {
      "epoch": 17.62,
      "learning_rate": 1.2903225806451613e-05,
      "loss": 0.0212,
      "step": 141
    },
    {
      "epoch": 17.75,
      "learning_rate": 1.2258064516129032e-05,
      "loss": 0.0228,
      "step": 142
    },
    {
      "epoch": 17.88,
      "learning_rate": 1.1612903225806453e-05,
      "loss": 0.0208,
      "step": 143
    },
    {
      "epoch": 18.0,
      "learning_rate": 1.0967741935483872e-05,
      "loss": 0.0188,
      "step": 144
    },
    {
      "epoch": 18.12,
      "learning_rate": 1.0322580645161291e-05,
      "loss": 0.0198,
      "step": 145
    },
    {
      "epoch": 18.25,
      "learning_rate": 9.67741935483871e-06,
      "loss": 0.0165,
      "step": 146
    },
    {
      "epoch": 18.38,
      "learning_rate": 9.03225806451613e-06,
      "loss": 0.0219,
      "step": 147
    },
    {
      "epoch": 18.5,
      "learning_rate": 8.387096774193549e-06,
      "loss": 0.0216,
      "step": 148
    },
    {
      "epoch": 18.62,
      "learning_rate": 7.741935483870968e-06,
      "loss": 0.0197,
      "step": 149
    },
    {
      "epoch": 18.75,
      "learning_rate": 7.096774193548387e-06,
      "loss": 0.0179,
      "step": 150
    },
    {
      "epoch": 18.88,
      "learning_rate": 6.451612903225806e-06,
      "loss": 0.0209,
      "step": 151
    },
    {
      "epoch": 19.0,
      "learning_rate": 5.806451612903226e-06,
      "loss": 0.0203,
      "step": 152
    },
    {
      "epoch": 19.12,
      "learning_rate": 5.161290322580646e-06,
      "loss": 0.0213,
      "step": 153
    },
    {
      "epoch": 19.25,
      "learning_rate": 4.516129032258065e-06,
      "loss": 0.0216,
      "step": 154
    },
    {
      "epoch": 19.38,
      "learning_rate": 3.870967741935484e-06,
      "loss": 0.0179,
      "step": 155
    },
    {
      "epoch": 19.5,
      "learning_rate": 3.225806451612903e-06,
      "loss": 0.0224,
      "step": 156
    },
    {
      "epoch": 19.62,
      "learning_rate": 2.580645161290323e-06,
      "loss": 0.0183,
      "step": 157
    },
    {
      "epoch": 19.75,
      "learning_rate": 1.935483870967742e-06,
      "loss": 0.0199,
      "step": 158
    },
    {
      "epoch": 19.88,
      "learning_rate": 1.2903225806451614e-06,
      "loss": 0.0186,
      "step": 159
    },
    {
      "epoch": 20.0,
      "learning_rate": 6.451612903225807e-07,
      "loss": 0.0233,
      "step": 160
    },
    {
      "epoch": 20.0,
      "step": 160,
      "total_flos": 1.209918701961216e+16,
      "train_loss": 0.3965469028917141,
      "train_runtime": 374.4336,
      "train_samples_per_second": 1.709,
      "train_steps_per_second": 0.427
    }
  ],
  "logging_steps": 1,
  "max_steps": 160,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 1.209918701961216e+16,
  "trial_name": null,
  "trial_params": null
}