{
  "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": 2.6536,
      "step": 1
    },
    {
      "epoch": 0.25,
      "learning_rate": 4e-05,
      "loss": 3.3995,
      "step": 2
    },
    {
      "epoch": 0.38,
      "learning_rate": 6e-05,
      "loss": 4.281,
      "step": 3
    },
    {
      "epoch": 0.5,
      "learning_rate": 8e-05,
      "loss": 2.4605,
      "step": 4
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001,
      "loss": 2.538,
      "step": 5
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.935483870967742e-05,
      "loss": 2.629,
      "step": 6
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.870967741935484e-05,
      "loss": 2.1875,
      "step": 7
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.806451612903226e-05,
      "loss": 1.9768,
      "step": 8
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.741935483870968e-05,
      "loss": 2.0199,
      "step": 9
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.677419354838711e-05,
      "loss": 2.0266,
      "step": 10
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.612903225806452e-05,
      "loss": 1.7383,
      "step": 11
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.548387096774195e-05,
      "loss": 1.7494,
      "step": 12
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.483870967741936e-05,
      "loss": 1.8509,
      "step": 13
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.419354838709677e-05,
      "loss": 1.8364,
      "step": 14
    },
    {
      "epoch": 1.88,
      "learning_rate": 9.35483870967742e-05,
      "loss": 1.6808,
      "step": 15
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.290322580645162e-05,
      "loss": 1.5642,
      "step": 16
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.225806451612904e-05,
      "loss": 1.6336,
      "step": 17
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.161290322580646e-05,
      "loss": 1.6452,
      "step": 18
    },
    {
      "epoch": 2.38,
      "learning_rate": 9.096774193548387e-05,
      "loss": 1.1714,
      "step": 19
    },
    {
      "epoch": 2.5,
      "learning_rate": 9.096774193548387e-05,
      "loss": 1.9018,
      "step": 20
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.032258064516129e-05,
      "loss": 1.6711,
      "step": 21
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.967741935483871e-05,
      "loss": 1.1646,
      "step": 22
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.903225806451614e-05,
      "loss": 1.0608,
      "step": 23
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.838709677419355e-05,
      "loss": 1.2916,
      "step": 24
    },
    {
      "epoch": 3.12,
      "learning_rate": 8.774193548387098e-05,
      "loss": 0.9801,
      "step": 25
    },
    {
      "epoch": 3.25,
      "learning_rate": 8.709677419354839e-05,
      "loss": 1.0964,
      "step": 26
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.645161290322581e-05,
      "loss": 1.1112,
      "step": 27
    },
    {
      "epoch": 3.5,
      "learning_rate": 8.580645161290323e-05,
      "loss": 1.2538,
      "step": 28
    },
    {
      "epoch": 3.62,
      "learning_rate": 8.516129032258064e-05,
      "loss": 1.0811,
      "step": 29
    },
    {
      "epoch": 3.75,
      "learning_rate": 8.451612903225808e-05,
      "loss": 0.9926,
      "step": 30
    },
    {
      "epoch": 3.88,
      "learning_rate": 8.387096774193549e-05,
      "loss": 1.3067,
      "step": 31
    },
    {
      "epoch": 4.0,
      "learning_rate": 8.32258064516129e-05,
      "loss": 0.9473,
      "step": 32
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.258064516129033e-05,
      "loss": 0.8545,
      "step": 33
    },
    {
      "epoch": 4.25,
      "learning_rate": 8.193548387096774e-05,
      "loss": 0.704,
      "step": 34
    },
    {
      "epoch": 4.38,
      "learning_rate": 8.129032258064517e-05,
      "loss": 0.8111,
      "step": 35
    },
    {
      "epoch": 4.5,
      "learning_rate": 8.064516129032258e-05,
      "loss": 0.9651,
      "step": 36
    },
    {
      "epoch": 4.62,
      "learning_rate": 8e-05,
      "loss": 0.6063,
      "step": 37
    },
    {
      "epoch": 4.75,
      "learning_rate": 7.935483870967743e-05,
      "loss": 0.5452,
      "step": 38
    },
    {
      "epoch": 4.88,
      "learning_rate": 7.870967741935484e-05,
      "loss": 0.8213,
      "step": 39
    },
    {
      "epoch": 5.0,
      "learning_rate": 7.806451612903226e-05,
      "loss": 0.7832,
      "step": 40
    },
    {
      "epoch": 5.12,
      "learning_rate": 7.741935483870968e-05,
      "loss": 0.3778,
      "step": 41
    },
    {
      "epoch": 5.25,
      "learning_rate": 7.67741935483871e-05,
      "loss": 0.3637,
      "step": 42
    },
    {
      "epoch": 5.38,
      "learning_rate": 7.612903225806451e-05,
      "loss": 0.6275,
      "step": 43
    },
    {
      "epoch": 5.5,
      "learning_rate": 7.548387096774195e-05,
      "loss": 0.6516,
      "step": 44
    },
    {
      "epoch": 5.62,
      "learning_rate": 7.483870967741936e-05,
      "loss": 0.3812,
      "step": 45
    },
    {
      "epoch": 5.75,
      "learning_rate": 7.419354838709677e-05,
      "loss": 0.3557,
      "step": 46
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.35483870967742e-05,
      "loss": 0.2963,
      "step": 47
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.290322580645161e-05,
      "loss": 0.2674,
      "step": 48
    },
    {
      "epoch": 6.12,
      "learning_rate": 7.225806451612904e-05,
      "loss": 0.1111,
      "step": 49
    },
    {
      "epoch": 6.25,
      "learning_rate": 7.161290322580646e-05,
      "loss": 0.1296,
      "step": 50
    },
    {
      "epoch": 6.38,
      "learning_rate": 7.096774193548388e-05,
      "loss": 0.1281,
      "step": 51
    },
    {
      "epoch": 6.5,
      "learning_rate": 7.03225806451613e-05,
      "loss": 0.238,
      "step": 52
    },
    {
      "epoch": 6.62,
      "learning_rate": 6.967741935483871e-05,
      "loss": 0.1803,
      "step": 53
    },
    {
      "epoch": 6.75,
      "learning_rate": 6.903225806451613e-05,
      "loss": 0.1911,
      "step": 54
    },
    {
      "epoch": 6.88,
      "learning_rate": 6.838709677419355e-05,
      "loss": 0.2073,
      "step": 55
    },
    {
      "epoch": 7.0,
      "learning_rate": 6.774193548387096e-05,
      "loss": 0.2415,
      "step": 56
    },
    {
      "epoch": 7.12,
      "learning_rate": 6.709677419354839e-05,
      "loss": 0.1104,
      "step": 57
    },
    {
      "epoch": 7.25,
      "learning_rate": 6.645161290322582e-05,
      "loss": 0.1154,
      "step": 58
    },
    {
      "epoch": 7.38,
      "learning_rate": 6.580645161290323e-05,
      "loss": 0.0931,
      "step": 59
    },
    {
      "epoch": 7.5,
      "learning_rate": 6.516129032258065e-05,
      "loss": 0.0671,
      "step": 60
    },
    {
      "epoch": 7.62,
      "learning_rate": 6.451612903225807e-05,
      "loss": 0.1003,
      "step": 61
    },
    {
      "epoch": 7.75,
      "learning_rate": 6.387096774193548e-05,
      "loss": 0.1646,
      "step": 62
    },
    {
      "epoch": 7.88,
      "learning_rate": 6.32258064516129e-05,
      "loss": 0.1203,
      "step": 63
    },
    {
      "epoch": 8.0,
      "learning_rate": 6.258064516129033e-05,
      "loss": 0.1888,
      "step": 64
    },
    {
      "epoch": 8.12,
      "learning_rate": 6.193548387096774e-05,
      "loss": 0.0798,
      "step": 65
    },
    {
      "epoch": 8.25,
      "learning_rate": 6.129032258064517e-05,
      "loss": 0.0921,
      "step": 66
    },
    {
      "epoch": 8.38,
      "learning_rate": 6.064516129032258e-05,
      "loss": 0.1284,
      "step": 67
    },
    {
      "epoch": 8.5,
      "learning_rate": 6e-05,
      "loss": 0.1008,
      "step": 68
    },
    {
      "epoch": 8.62,
      "learning_rate": 5.935483870967742e-05,
      "loss": 0.0732,
      "step": 69
    },
    {
      "epoch": 8.75,
      "learning_rate": 5.870967741935483e-05,
      "loss": 0.0628,
      "step": 70
    },
    {
      "epoch": 8.88,
      "learning_rate": 5.8064516129032266e-05,
      "loss": 0.0694,
      "step": 71
    },
    {
      "epoch": 9.0,
      "learning_rate": 5.7419354838709685e-05,
      "loss": 0.0856,
      "step": 72
    },
    {
      "epoch": 9.12,
      "learning_rate": 5.67741935483871e-05,
      "loss": 0.0698,
      "step": 73
    },
    {
      "epoch": 9.25,
      "learning_rate": 5.612903225806452e-05,
      "loss": 0.0619,
      "step": 74
    },
    {
      "epoch": 9.38,
      "learning_rate": 5.5483870967741936e-05,
      "loss": 0.1077,
      "step": 75
    },
    {
      "epoch": 9.5,
      "learning_rate": 5.4838709677419355e-05,
      "loss": 0.0524,
      "step": 76
    },
    {
      "epoch": 9.62,
      "learning_rate": 5.419354838709678e-05,
      "loss": 0.0747,
      "step": 77
    },
    {
      "epoch": 9.75,
      "learning_rate": 5.35483870967742e-05,
      "loss": 0.0805,
      "step": 78
    },
    {
      "epoch": 9.88,
      "learning_rate": 5.290322580645162e-05,
      "loss": 0.0665,
      "step": 79
    },
    {
      "epoch": 10.0,
      "learning_rate": 5.225806451612903e-05,
      "loss": 0.0688,
      "step": 80
    },
    {
      "epoch": 10.12,
      "learning_rate": 5.161290322580645e-05,
      "loss": 0.0421,
      "step": 81
    },
    {
      "epoch": 10.25,
      "learning_rate": 5.096774193548387e-05,
      "loss": 0.0548,
      "step": 82
    },
    {
      "epoch": 10.38,
      "learning_rate": 5.032258064516129e-05,
      "loss": 0.0552,
      "step": 83
    },
    {
      "epoch": 10.5,
      "learning_rate": 4.967741935483871e-05,
      "loss": 0.0777,
      "step": 84
    },
    {
      "epoch": 10.62,
      "learning_rate": 4.903225806451613e-05,
      "loss": 0.0574,
      "step": 85
    },
    {
      "epoch": 10.75,
      "learning_rate": 4.8387096774193554e-05,
      "loss": 0.1117,
      "step": 86
    },
    {
      "epoch": 10.88,
      "learning_rate": 4.774193548387097e-05,
      "loss": 0.0512,
      "step": 87
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.7096774193548385e-05,
      "loss": 0.07,
      "step": 88
    },
    {
      "epoch": 11.12,
      "learning_rate": 4.645161290322581e-05,
      "loss": 0.0381,
      "step": 89
    },
    {
      "epoch": 11.25,
      "learning_rate": 4.580645161290323e-05,
      "loss": 0.0589,
      "step": 90
    },
    {
      "epoch": 11.38,
      "learning_rate": 4.516129032258064e-05,
      "loss": 0.0407,
      "step": 91
    },
    {
      "epoch": 11.5,
      "learning_rate": 4.451612903225807e-05,
      "loss": 0.0484,
      "step": 92
    },
    {
      "epoch": 11.62,
      "learning_rate": 4.387096774193549e-05,
      "loss": 0.0487,
      "step": 93
    },
    {
      "epoch": 11.75,
      "learning_rate": 4.322580645161291e-05,
      "loss": 0.0616,
      "step": 94
    },
    {
      "epoch": 11.88,
      "learning_rate": 4.258064516129032e-05,
      "loss": 0.0739,
      "step": 95
    },
    {
      "epoch": 12.0,
      "learning_rate": 4.1935483870967746e-05,
      "loss": 0.0951,
      "step": 96
    },
    {
      "epoch": 12.12,
      "learning_rate": 4.1290322580645165e-05,
      "loss": 0.0978,
      "step": 97
    },
    {
      "epoch": 12.25,
      "learning_rate": 4.0645161290322584e-05,
      "loss": 0.0436,
      "step": 98
    },
    {
      "epoch": 12.38,
      "learning_rate": 4e-05,
      "loss": 0.0393,
      "step": 99
    },
    {
      "epoch": 12.5,
      "learning_rate": 3.935483870967742e-05,
      "loss": 0.0601,
      "step": 100
    },
    {
      "epoch": 12.62,
      "learning_rate": 3.870967741935484e-05,
      "loss": 0.0518,
      "step": 101
    },
    {
      "epoch": 12.75,
      "learning_rate": 3.8064516129032254e-05,
      "loss": 0.0464,
      "step": 102
    },
    {
      "epoch": 12.88,
      "learning_rate": 3.741935483870968e-05,
      "loss": 0.0483,
      "step": 103
    },
    {
      "epoch": 13.0,
      "learning_rate": 3.67741935483871e-05,
      "loss": 0.0659,
      "step": 104
    },
    {
      "epoch": 13.12,
      "learning_rate": 3.612903225806452e-05,
      "loss": 0.033,
      "step": 105
    },
    {
      "epoch": 13.25,
      "learning_rate": 3.548387096774194e-05,
      "loss": 0.0395,
      "step": 106
    },
    {
      "epoch": 13.38,
      "learning_rate": 3.483870967741936e-05,
      "loss": 0.0711,
      "step": 107
    },
    {
      "epoch": 13.5,
      "learning_rate": 3.4193548387096776e-05,
      "loss": 0.051,
      "step": 108
    },
    {
      "epoch": 13.62,
      "learning_rate": 3.3548387096774195e-05,
      "loss": 0.0633,
      "step": 109
    },
    {
      "epoch": 13.75,
      "learning_rate": 3.2903225806451614e-05,
      "loss": 0.0406,
      "step": 110
    },
    {
      "epoch": 13.88,
      "learning_rate": 3.2258064516129034e-05,
      "loss": 0.0562,
      "step": 111
    },
    {
      "epoch": 14.0,
      "learning_rate": 3.161290322580645e-05,
      "loss": 0.0406,
      "step": 112
    },
    {
      "epoch": 14.12,
      "learning_rate": 3.096774193548387e-05,
      "loss": 0.0346,
      "step": 113
    },
    {
      "epoch": 14.25,
      "learning_rate": 3.032258064516129e-05,
      "loss": 0.0555,
      "step": 114
    },
    {
      "epoch": 14.38,
      "learning_rate": 2.967741935483871e-05,
      "loss": 0.0435,
      "step": 115
    },
    {
      "epoch": 14.5,
      "learning_rate": 2.9032258064516133e-05,
      "loss": 0.0486,
      "step": 116
    },
    {
      "epoch": 14.62,
      "learning_rate": 2.838709677419355e-05,
      "loss": 0.0334,
      "step": 117
    },
    {
      "epoch": 14.75,
      "learning_rate": 2.7741935483870968e-05,
      "loss": 0.0571,
      "step": 118
    },
    {
      "epoch": 14.88,
      "learning_rate": 2.709677419354839e-05,
      "loss": 0.0437,
      "step": 119
    },
    {
      "epoch": 15.0,
      "learning_rate": 2.645161290322581e-05,
      "loss": 0.0578,
      "step": 120
    },
    {
      "epoch": 15.12,
      "learning_rate": 2.5806451612903226e-05,
      "loss": 0.0373,
      "step": 121
    },
    {
      "epoch": 15.25,
      "learning_rate": 2.5161290322580645e-05,
      "loss": 0.0642,
      "step": 122
    },
    {
      "epoch": 15.38,
      "learning_rate": 2.4516129032258064e-05,
      "loss": 0.0428,
      "step": 123
    },
    {
      "epoch": 15.5,
      "learning_rate": 2.3870967741935486e-05,
      "loss": 0.0523,
      "step": 124
    },
    {
      "epoch": 15.62,
      "learning_rate": 2.3225806451612906e-05,
      "loss": 0.0311,
      "step": 125
    },
    {
      "epoch": 15.75,
      "learning_rate": 2.258064516129032e-05,
      "loss": 0.038,
      "step": 126
    },
    {
      "epoch": 15.88,
      "learning_rate": 2.1935483870967744e-05,
      "loss": 0.0385,
      "step": 127
    },
    {
      "epoch": 16.0,
      "learning_rate": 2.129032258064516e-05,
      "loss": 0.0699,
      "step": 128
    },
    {
      "epoch": 16.12,
      "learning_rate": 2.0645161290322582e-05,
      "loss": 0.061,
      "step": 129
    },
    {
      "epoch": 16.25,
      "learning_rate": 2e-05,
      "loss": 0.0431,
      "step": 130
    },
    {
      "epoch": 16.38,
      "learning_rate": 1.935483870967742e-05,
      "loss": 0.0665,
      "step": 131
    },
    {
      "epoch": 16.5,
      "learning_rate": 1.870967741935484e-05,
      "loss": 0.032,
      "step": 132
    },
    {
      "epoch": 16.62,
      "learning_rate": 1.806451612903226e-05,
      "loss": 0.045,
      "step": 133
    },
    {
      "epoch": 16.75,
      "learning_rate": 1.741935483870968e-05,
      "loss": 0.0533,
      "step": 134
    },
    {
      "epoch": 16.88,
      "learning_rate": 1.6774193548387098e-05,
      "loss": 0.0393,
      "step": 135
    },
    {
      "epoch": 17.0,
      "learning_rate": 1.6129032258064517e-05,
      "loss": 0.0462,
      "step": 136
    },
    {
      "epoch": 17.12,
      "learning_rate": 1.5483870967741936e-05,
      "loss": 0.042,
      "step": 137
    },
    {
      "epoch": 17.25,
      "learning_rate": 1.4838709677419355e-05,
      "loss": 0.0694,
      "step": 138
    },
    {
      "epoch": 17.38,
      "learning_rate": 1.4193548387096774e-05,
      "loss": 0.0253,
      "step": 139
    },
    {
      "epoch": 17.5,
      "learning_rate": 1.3548387096774195e-05,
      "loss": 0.0367,
      "step": 140
    },
    {
      "epoch": 17.62,
      "learning_rate": 1.2903225806451613e-05,
      "loss": 0.0391,
      "step": 141
    },
    {
      "epoch": 17.75,
      "learning_rate": 1.2258064516129032e-05,
      "loss": 0.0508,
      "step": 142
    },
    {
      "epoch": 17.88,
      "learning_rate": 1.1612903225806453e-05,
      "loss": 0.0516,
      "step": 143
    },
    {
      "epoch": 18.0,
      "learning_rate": 1.0967741935483872e-05,
      "loss": 0.0562,
      "step": 144
    },
    {
      "epoch": 18.12,
      "learning_rate": 1.0322580645161291e-05,
      "loss": 0.0404,
      "step": 145
    },
    {
      "epoch": 18.25,
      "learning_rate": 9.67741935483871e-06,
      "loss": 0.0427,
      "step": 146
    },
    {
      "epoch": 18.38,
      "learning_rate": 9.03225806451613e-06,
      "loss": 0.0318,
      "step": 147
    },
    {
      "epoch": 18.5,
      "learning_rate": 8.387096774193549e-06,
      "loss": 0.052,
      "step": 148
    },
    {
      "epoch": 18.62,
      "learning_rate": 7.741935483870968e-06,
      "loss": 0.0278,
      "step": 149
    },
    {
      "epoch": 18.75,
      "learning_rate": 7.096774193548387e-06,
      "loss": 0.095,
      "step": 150
    },
    {
      "epoch": 18.88,
      "learning_rate": 6.451612903225806e-06,
      "loss": 0.0408,
      "step": 151
    },
    {
      "epoch": 19.0,
      "learning_rate": 5.806451612903226e-06,
      "loss": 0.0509,
      "step": 152
    },
    {
      "epoch": 19.12,
      "learning_rate": 5.161290322580646e-06,
      "loss": 0.0297,
      "step": 153
    },
    {
      "epoch": 19.25,
      "learning_rate": 4.516129032258065e-06,
      "loss": 0.0493,
      "step": 154
    },
    {
      "epoch": 19.38,
      "learning_rate": 3.870967741935484e-06,
      "loss": 0.1001,
      "step": 155
    },
    {
      "epoch": 19.5,
      "learning_rate": 3.225806451612903e-06,
      "loss": 0.0401,
      "step": 156
    },
    {
      "epoch": 19.62,
      "learning_rate": 2.580645161290323e-06,
      "loss": 0.0434,
      "step": 157
    },
    {
      "epoch": 19.75,
      "learning_rate": 1.935483870967742e-06,
      "loss": 0.0363,
      "step": 158
    },
    {
      "epoch": 19.88,
      "learning_rate": 1.2903225806451614e-06,
      "loss": 0.0432,
      "step": 159
    },
    {
      "epoch": 20.0,
      "learning_rate": 6.451612903225807e-07,
      "loss": 0.0328,
      "step": 160
    },
    {
      "epoch": 20.0,
      "step": 160,
      "total_flos": 7453507615457280.0,
      "train_loss": 0.4629385274136439,
      "train_runtime": 260.0227,
      "train_samples_per_second": 2.461,
      "train_steps_per_second": 0.615
    }
  ],
  "logging_steps": 1,
  "max_steps": 160,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 7453507615457280.0,
  "trial_name": null,
  "trial_params": null
}