{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 1000,
  "global_step": 186,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.016129032258064516,
      "grad_norm": 25.0,
      "learning_rate": 4.973118279569893e-05,
      "loss": 50.0,
      "step": 1
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 23.625,
      "learning_rate": 4.9462365591397855e-05,
      "loss": 40.75,
      "step": 2
    },
    {
      "epoch": 0.04838709677419355,
      "grad_norm": 19.125,
      "learning_rate": 4.9193548387096775e-05,
      "loss": 30.25,
      "step": 3
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 7.8125,
      "learning_rate": 4.89247311827957e-05,
      "loss": 22.0,
      "step": 4
    },
    {
      "epoch": 0.08064516129032258,
      "grad_norm": 338.0,
      "learning_rate": 4.865591397849463e-05,
      "loss": 24.0,
      "step": 5
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 14.375,
      "learning_rate": 4.8387096774193554e-05,
      "loss": 22.125,
      "step": 6
    },
    {
      "epoch": 0.11290322580645161,
      "grad_norm": 14.5,
      "learning_rate": 4.811827956989248e-05,
      "loss": 19.125,
      "step": 7
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 8.0,
      "learning_rate": 4.78494623655914e-05,
      "loss": 18.625,
      "step": 8
    },
    {
      "epoch": 0.14516129032258066,
      "grad_norm": 9.6875,
      "learning_rate": 4.7580645161290326e-05,
      "loss": 18.5,
      "step": 9
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 8.375,
      "learning_rate": 4.731182795698925e-05,
      "loss": 17.125,
      "step": 10
    },
    {
      "epoch": 0.1774193548387097,
      "grad_norm": 6.71875,
      "learning_rate": 4.704301075268818e-05,
      "loss": 15.6875,
      "step": 11
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 10.375,
      "learning_rate": 4.67741935483871e-05,
      "loss": 15.625,
      "step": 12
    },
    {
      "epoch": 0.20967741935483872,
      "grad_norm": 13.6875,
      "learning_rate": 4.650537634408602e-05,
      "loss": 15.375,
      "step": 13
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 15.5,
      "learning_rate": 4.6236559139784944e-05,
      "loss": 11.8125,
      "step": 14
    },
    {
      "epoch": 0.24193548387096775,
      "grad_norm": 17.0,
      "learning_rate": 4.596774193548387e-05,
      "loss": 12.0625,
      "step": 15
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 15.625,
      "learning_rate": 4.56989247311828e-05,
      "loss": 11.0,
      "step": 16
    },
    {
      "epoch": 0.27419354838709675,
      "grad_norm": 18.5,
      "learning_rate": 4.543010752688172e-05,
      "loss": 11.0,
      "step": 17
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 15.25,
      "learning_rate": 4.516129032258064e-05,
      "loss": 8.6875,
      "step": 18
    },
    {
      "epoch": 0.3064516129032258,
      "grad_norm": 16.875,
      "learning_rate": 4.489247311827957e-05,
      "loss": 9.875,
      "step": 19
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 16.0,
      "learning_rate": 4.4623655913978496e-05,
      "loss": 10.25,
      "step": 20
    },
    {
      "epoch": 0.3387096774193548,
      "grad_norm": 16.625,
      "learning_rate": 4.435483870967742e-05,
      "loss": 9.1875,
      "step": 21
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 12.75,
      "learning_rate": 4.408602150537635e-05,
      "loss": 9.5625,
      "step": 22
    },
    {
      "epoch": 0.3709677419354839,
      "grad_norm": 14.75,
      "learning_rate": 4.381720430107527e-05,
      "loss": 9.125,
      "step": 23
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 13.375,
      "learning_rate": 4.3548387096774194e-05,
      "loss": 6.3438,
      "step": 24
    },
    {
      "epoch": 0.4032258064516129,
      "grad_norm": 12.6875,
      "learning_rate": 4.327956989247312e-05,
      "loss": 7.4062,
      "step": 25
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 13.25,
      "learning_rate": 4.301075268817205e-05,
      "loss": 6.6875,
      "step": 26
    },
    {
      "epoch": 0.43548387096774194,
      "grad_norm": 12.0,
      "learning_rate": 4.2741935483870973e-05,
      "loss": 5.8125,
      "step": 27
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 14.3125,
      "learning_rate": 4.247311827956989e-05,
      "loss": 6.4688,
      "step": 28
    },
    {
      "epoch": 0.46774193548387094,
      "grad_norm": 11.1875,
      "learning_rate": 4.220430107526882e-05,
      "loss": 5.5312,
      "step": 29
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 14.0,
      "learning_rate": 4.1935483870967746e-05,
      "loss": 6.0625,
      "step": 30
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.875,
      "learning_rate": 4.166666666666667e-05,
      "loss": 4.5,
      "step": 31
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 11.0,
      "learning_rate": 4.13978494623656e-05,
      "loss": 6.75,
      "step": 32
    },
    {
      "epoch": 0.532258064516129,
      "grad_norm": 11.8125,
      "learning_rate": 4.112903225806452e-05,
      "loss": 6.625,
      "step": 33
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 11.5625,
      "learning_rate": 4.0860215053763444e-05,
      "loss": 4.2812,
      "step": 34
    },
    {
      "epoch": 0.5645161290322581,
      "grad_norm": 8.9375,
      "learning_rate": 4.0591397849462364e-05,
      "loss": 4.5938,
      "step": 35
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 10.75,
      "learning_rate": 4.032258064516129e-05,
      "loss": 6.125,
      "step": 36
    },
    {
      "epoch": 0.5967741935483871,
      "grad_norm": 9.5625,
      "learning_rate": 4.005376344086022e-05,
      "loss": 5.4062,
      "step": 37
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 9.4375,
      "learning_rate": 3.978494623655914e-05,
      "loss": 6.0,
      "step": 38
    },
    {
      "epoch": 0.6290322580645161,
      "grad_norm": 8.5625,
      "learning_rate": 3.951612903225806e-05,
      "loss": 3.7969,
      "step": 39
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 9.625,
      "learning_rate": 3.924731182795699e-05,
      "loss": 5.0,
      "step": 40
    },
    {
      "epoch": 0.6612903225806451,
      "grad_norm": 12.625,
      "learning_rate": 3.8978494623655915e-05,
      "loss": 7.0,
      "step": 41
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 8.5625,
      "learning_rate": 3.870967741935484e-05,
      "loss": 6.0938,
      "step": 42
    },
    {
      "epoch": 0.6935483870967742,
      "grad_norm": 11.625,
      "learning_rate": 3.844086021505376e-05,
      "loss": 6.0625,
      "step": 43
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 10.8125,
      "learning_rate": 3.817204301075269e-05,
      "loss": 4.2188,
      "step": 44
    },
    {
      "epoch": 0.7258064516129032,
      "grad_norm": 13.125,
      "learning_rate": 3.7903225806451614e-05,
      "loss": 5.6875,
      "step": 45
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 9.875,
      "learning_rate": 3.763440860215054e-05,
      "loss": 4.9688,
      "step": 46
    },
    {
      "epoch": 0.7580645161290323,
      "grad_norm": 8.75,
      "learning_rate": 3.736559139784947e-05,
      "loss": 4.0625,
      "step": 47
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 13.125,
      "learning_rate": 3.7096774193548386e-05,
      "loss": 6.2812,
      "step": 48
    },
    {
      "epoch": 0.7903225806451613,
      "grad_norm": 12.1875,
      "learning_rate": 3.682795698924731e-05,
      "loss": 5.125,
      "step": 49
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 9.5,
      "learning_rate": 3.655913978494624e-05,
      "loss": 4.5938,
      "step": 50
    },
    {
      "epoch": 0.8225806451612904,
      "grad_norm": 9.125,
      "learning_rate": 3.6290322580645165e-05,
      "loss": 3.4375,
      "step": 51
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 8.8125,
      "learning_rate": 3.602150537634409e-05,
      "loss": 5.1562,
      "step": 52
    },
    {
      "epoch": 0.8548387096774194,
      "grad_norm": 12.625,
      "learning_rate": 3.575268817204301e-05,
      "loss": 5.3125,
      "step": 53
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 10.8125,
      "learning_rate": 3.548387096774194e-05,
      "loss": 6.0312,
      "step": 54
    },
    {
      "epoch": 0.8870967741935484,
      "grad_norm": 10.0,
      "learning_rate": 3.5215053763440864e-05,
      "loss": 3.9844,
      "step": 55
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 7.15625,
      "learning_rate": 3.494623655913979e-05,
      "loss": 2.1562,
      "step": 56
    },
    {
      "epoch": 0.9193548387096774,
      "grad_norm": 9.1875,
      "learning_rate": 3.467741935483872e-05,
      "loss": 3.8438,
      "step": 57
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 9.1875,
      "learning_rate": 3.4408602150537636e-05,
      "loss": 4.4062,
      "step": 58
    },
    {
      "epoch": 0.9516129032258065,
      "grad_norm": 8.6875,
      "learning_rate": 3.4139784946236556e-05,
      "loss": 3.7969,
      "step": 59
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 7.84375,
      "learning_rate": 3.387096774193548e-05,
      "loss": 4.3125,
      "step": 60
    },
    {
      "epoch": 0.9838709677419355,
      "grad_norm": 8.375,
      "learning_rate": 3.360215053763441e-05,
      "loss": 3.2812,
      "step": 61
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.5625,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 4.5625,
      "step": 62
    },
    {
      "epoch": 1.0161290322580645,
      "grad_norm": 4.53125,
      "learning_rate": 3.306451612903226e-05,
      "loss": 1.6016,
      "step": 63
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 5.5625,
      "learning_rate": 3.279569892473118e-05,
      "loss": 1.8984,
      "step": 64
    },
    {
      "epoch": 1.0483870967741935,
      "grad_norm": 5.625,
      "learning_rate": 3.252688172043011e-05,
      "loss": 1.3438,
      "step": 65
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 6.03125,
      "learning_rate": 3.2258064516129034e-05,
      "loss": 1.7109,
      "step": 66
    },
    {
      "epoch": 1.0806451612903225,
      "grad_norm": 6.40625,
      "learning_rate": 3.198924731182796e-05,
      "loss": 2.4219,
      "step": 67
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 7.84375,
      "learning_rate": 3.172043010752688e-05,
      "loss": 2.6094,
      "step": 68
    },
    {
      "epoch": 1.1129032258064515,
      "grad_norm": 4.90625,
      "learning_rate": 3.1451612903225806e-05,
      "loss": 1.2031,
      "step": 69
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 7.34375,
      "learning_rate": 3.118279569892473e-05,
      "loss": 1.5703,
      "step": 70
    },
    {
      "epoch": 1.1451612903225807,
      "grad_norm": 9.1875,
      "learning_rate": 3.091397849462366e-05,
      "loss": 2.0469,
      "step": 71
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 6.46875,
      "learning_rate": 3.0645161290322585e-05,
      "loss": 0.9648,
      "step": 72
    },
    {
      "epoch": 1.1774193548387097,
      "grad_norm": 9.375,
      "learning_rate": 3.0376344086021508e-05,
      "loss": 2.2031,
      "step": 73
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 10.375,
      "learning_rate": 3.010752688172043e-05,
      "loss": 2.5938,
      "step": 74
    },
    {
      "epoch": 1.2096774193548387,
      "grad_norm": 4.625,
      "learning_rate": 2.9838709677419357e-05,
      "loss": 1.125,
      "step": 75
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 5.09375,
      "learning_rate": 2.9569892473118284e-05,
      "loss": 1.0625,
      "step": 76
    },
    {
      "epoch": 1.2419354838709677,
      "grad_norm": 6.71875,
      "learning_rate": 2.9301075268817207e-05,
      "loss": 1.8672,
      "step": 77
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 7.53125,
      "learning_rate": 2.9032258064516133e-05,
      "loss": 2.2656,
      "step": 78
    },
    {
      "epoch": 1.2741935483870968,
      "grad_norm": 6.5,
      "learning_rate": 2.8763440860215056e-05,
      "loss": 2.0625,
      "step": 79
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 3.5,
      "learning_rate": 2.8494623655913982e-05,
      "loss": 0.8086,
      "step": 80
    },
    {
      "epoch": 1.3064516129032258,
      "grad_norm": 4.5625,
      "learning_rate": 2.822580645161291e-05,
      "loss": 1.6797,
      "step": 81
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 5.0625,
      "learning_rate": 2.7956989247311828e-05,
      "loss": 1.5391,
      "step": 82
    },
    {
      "epoch": 1.3387096774193548,
      "grad_norm": 6.96875,
      "learning_rate": 2.768817204301075e-05,
      "loss": 1.8438,
      "step": 83
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 4.375,
      "learning_rate": 2.7419354838709678e-05,
      "loss": 1.2812,
      "step": 84
    },
    {
      "epoch": 1.370967741935484,
      "grad_norm": 5.8125,
      "learning_rate": 2.71505376344086e-05,
      "loss": 1.3359,
      "step": 85
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 4.6875,
      "learning_rate": 2.6881720430107527e-05,
      "loss": 0.9766,
      "step": 86
    },
    {
      "epoch": 1.403225806451613,
      "grad_norm": 8.6875,
      "learning_rate": 2.661290322580645e-05,
      "loss": 2.4219,
      "step": 87
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 6.6875,
      "learning_rate": 2.6344086021505376e-05,
      "loss": 1.3672,
      "step": 88
    },
    {
      "epoch": 1.435483870967742,
      "grad_norm": 8.375,
      "learning_rate": 2.6075268817204303e-05,
      "loss": 1.8828,
      "step": 89
    },
    {
      "epoch": 1.4516129032258065,
      "grad_norm": 5.25,
      "learning_rate": 2.5806451612903226e-05,
      "loss": 1.3984,
      "step": 90
    },
    {
      "epoch": 1.467741935483871,
      "grad_norm": 4.625,
      "learning_rate": 2.5537634408602152e-05,
      "loss": 0.8516,
      "step": 91
    },
    {
      "epoch": 1.4838709677419355,
      "grad_norm": 7.8125,
      "learning_rate": 2.5268817204301075e-05,
      "loss": 2.4844,
      "step": 92
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.25,
      "learning_rate": 2.5e-05,
      "loss": 0.8594,
      "step": 93
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 5.75,
      "learning_rate": 2.4731182795698928e-05,
      "loss": 1.5078,
      "step": 94
    },
    {
      "epoch": 1.532258064516129,
      "grad_norm": 5.625,
      "learning_rate": 2.446236559139785e-05,
      "loss": 1.625,
      "step": 95
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 5.0,
      "learning_rate": 2.4193548387096777e-05,
      "loss": 1.1016,
      "step": 96
    },
    {
      "epoch": 1.564516129032258,
      "grad_norm": 4.5,
      "learning_rate": 2.39247311827957e-05,
      "loss": 2.0,
      "step": 97
    },
    {
      "epoch": 1.5806451612903225,
      "grad_norm": 5.53125,
      "learning_rate": 2.3655913978494626e-05,
      "loss": 1.4766,
      "step": 98
    },
    {
      "epoch": 1.596774193548387,
      "grad_norm": 4.15625,
      "learning_rate": 2.338709677419355e-05,
      "loss": 0.7227,
      "step": 99
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 5.15625,
      "learning_rate": 2.3118279569892472e-05,
      "loss": 1.0781,
      "step": 100
    },
    {
      "epoch": 1.629032258064516,
      "grad_norm": 2.96875,
      "learning_rate": 2.28494623655914e-05,
      "loss": 0.5977,
      "step": 101
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 7.78125,
      "learning_rate": 2.258064516129032e-05,
      "loss": 2.1562,
      "step": 102
    },
    {
      "epoch": 1.661290322580645,
      "grad_norm": 5.0625,
      "learning_rate": 2.2311827956989248e-05,
      "loss": 1.1172,
      "step": 103
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 4.4375,
      "learning_rate": 2.2043010752688174e-05,
      "loss": 0.5586,
      "step": 104
    },
    {
      "epoch": 1.6935483870967742,
      "grad_norm": 6.21875,
      "learning_rate": 2.1774193548387097e-05,
      "loss": 1.5469,
      "step": 105
    },
    {
      "epoch": 1.7096774193548387,
      "grad_norm": 6.40625,
      "learning_rate": 2.1505376344086024e-05,
      "loss": 1.2891,
      "step": 106
    },
    {
      "epoch": 1.7258064516129032,
      "grad_norm": 6.0625,
      "learning_rate": 2.1236559139784946e-05,
      "loss": 2.6719,
      "step": 107
    },
    {
      "epoch": 1.7419354838709677,
      "grad_norm": 10.5625,
      "learning_rate": 2.0967741935483873e-05,
      "loss": 2.2031,
      "step": 108
    },
    {
      "epoch": 1.7580645161290323,
      "grad_norm": 10.0625,
      "learning_rate": 2.06989247311828e-05,
      "loss": 2.1094,
      "step": 109
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 7.53125,
      "learning_rate": 2.0430107526881722e-05,
      "loss": 2.4375,
      "step": 110
    },
    {
      "epoch": 1.7903225806451613,
      "grad_norm": 6.5625,
      "learning_rate": 2.0161290322580645e-05,
      "loss": 1.4922,
      "step": 111
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 9.3125,
      "learning_rate": 1.989247311827957e-05,
      "loss": 2.6406,
      "step": 112
    },
    {
      "epoch": 1.8225806451612905,
      "grad_norm": 5.625,
      "learning_rate": 1.9623655913978494e-05,
      "loss": 0.8398,
      "step": 113
    },
    {
      "epoch": 1.838709677419355,
      "grad_norm": 2.828125,
      "learning_rate": 1.935483870967742e-05,
      "loss": 0.4219,
      "step": 114
    },
    {
      "epoch": 1.8548387096774195,
      "grad_norm": 6.03125,
      "learning_rate": 1.9086021505376344e-05,
      "loss": 0.9766,
      "step": 115
    },
    {
      "epoch": 1.870967741935484,
      "grad_norm": 4.59375,
      "learning_rate": 1.881720430107527e-05,
      "loss": 0.7578,
      "step": 116
    },
    {
      "epoch": 1.8870967741935485,
      "grad_norm": 1.40625,
      "learning_rate": 1.8548387096774193e-05,
      "loss": 0.2539,
      "step": 117
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 4.40625,
      "learning_rate": 1.827956989247312e-05,
      "loss": 0.9102,
      "step": 118
    },
    {
      "epoch": 1.9193548387096775,
      "grad_norm": 6.03125,
      "learning_rate": 1.8010752688172046e-05,
      "loss": 1.5469,
      "step": 119
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 8.125,
      "learning_rate": 1.774193548387097e-05,
      "loss": 1.6172,
      "step": 120
    },
    {
      "epoch": 1.9516129032258065,
      "grad_norm": 4.53125,
      "learning_rate": 1.7473118279569895e-05,
      "loss": 0.5195,
      "step": 121
    },
    {
      "epoch": 1.967741935483871,
      "grad_norm": 6.53125,
      "learning_rate": 1.7204301075268818e-05,
      "loss": 1.1797,
      "step": 122
    },
    {
      "epoch": 1.9838709677419355,
      "grad_norm": 5.84375,
      "learning_rate": 1.693548387096774e-05,
      "loss": 1.3125,
      "step": 123
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.09375,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.5352,
      "step": 124
    },
    {
      "epoch": 2.0161290322580645,
      "grad_norm": 2.546875,
      "learning_rate": 1.639784946236559e-05,
      "loss": 0.3926,
      "step": 125
    },
    {
      "epoch": 2.032258064516129,
      "grad_norm": 3.90625,
      "learning_rate": 1.6129032258064517e-05,
      "loss": 0.2715,
      "step": 126
    },
    {
      "epoch": 2.0483870967741935,
      "grad_norm": 3.375,
      "learning_rate": 1.586021505376344e-05,
      "loss": 0.7422,
      "step": 127
    },
    {
      "epoch": 2.064516129032258,
      "grad_norm": 2.453125,
      "learning_rate": 1.5591397849462366e-05,
      "loss": 0.332,
      "step": 128
    },
    {
      "epoch": 2.0806451612903225,
      "grad_norm": 1.1484375,
      "learning_rate": 1.5322580645161292e-05,
      "loss": 0.1719,
      "step": 129
    },
    {
      "epoch": 2.096774193548387,
      "grad_norm": 1.0390625,
      "learning_rate": 1.5053763440860215e-05,
      "loss": 0.1475,
      "step": 130
    },
    {
      "epoch": 2.1129032258064515,
      "grad_norm": 2.625,
      "learning_rate": 1.4784946236559142e-05,
      "loss": 0.4004,
      "step": 131
    },
    {
      "epoch": 2.129032258064516,
      "grad_norm": 4.34375,
      "learning_rate": 1.4516129032258066e-05,
      "loss": 0.2656,
      "step": 132
    },
    {
      "epoch": 2.1451612903225805,
      "grad_norm": 0.984375,
      "learning_rate": 1.4247311827956991e-05,
      "loss": 0.1357,
      "step": 133
    },
    {
      "epoch": 2.161290322580645,
      "grad_norm": 1.40625,
      "learning_rate": 1.3978494623655914e-05,
      "loss": 0.1582,
      "step": 134
    },
    {
      "epoch": 2.1774193548387095,
      "grad_norm": 2.625,
      "learning_rate": 1.3709677419354839e-05,
      "loss": 0.4531,
      "step": 135
    },
    {
      "epoch": 2.193548387096774,
      "grad_norm": 4.3125,
      "learning_rate": 1.3440860215053763e-05,
      "loss": 0.3574,
      "step": 136
    },
    {
      "epoch": 2.2096774193548385,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3172043010752688e-05,
      "loss": 0.1387,
      "step": 137
    },
    {
      "epoch": 2.225806451612903,
      "grad_norm": 0.921875,
      "learning_rate": 1.2903225806451613e-05,
      "loss": 0.2852,
      "step": 138
    },
    {
      "epoch": 2.241935483870968,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2634408602150537e-05,
      "loss": 0.1357,
      "step": 139
    },
    {
      "epoch": 2.258064516129032,
      "grad_norm": 0.75,
      "learning_rate": 1.2365591397849464e-05,
      "loss": 0.0781,
      "step": 140
    },
    {
      "epoch": 2.274193548387097,
      "grad_norm": 4.90625,
      "learning_rate": 1.2096774193548388e-05,
      "loss": 1.4297,
      "step": 141
    },
    {
      "epoch": 2.2903225806451615,
      "grad_norm": 1.015625,
      "learning_rate": 1.1827956989247313e-05,
      "loss": 0.1338,
      "step": 142
    },
    {
      "epoch": 2.306451612903226,
      "grad_norm": 0.67578125,
      "learning_rate": 1.1559139784946236e-05,
      "loss": 0.0981,
      "step": 143
    },
    {
      "epoch": 2.3225806451612905,
      "grad_norm": 2.703125,
      "learning_rate": 1.129032258064516e-05,
      "loss": 0.4043,
      "step": 144
    },
    {
      "epoch": 2.338709677419355,
      "grad_norm": 1.953125,
      "learning_rate": 1.1021505376344087e-05,
      "loss": 0.1592,
      "step": 145
    },
    {
      "epoch": 2.3548387096774195,
      "grad_norm": 2.875,
      "learning_rate": 1.0752688172043012e-05,
      "loss": 0.5117,
      "step": 146
    },
    {
      "epoch": 2.370967741935484,
      "grad_norm": 1.0703125,
      "learning_rate": 1.0483870967741936e-05,
      "loss": 0.1611,
      "step": 147
    },
    {
      "epoch": 2.3870967741935485,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0215053763440861e-05,
      "loss": 0.1182,
      "step": 148
    },
    {
      "epoch": 2.403225806451613,
      "grad_norm": 1.2265625,
      "learning_rate": 9.946236559139786e-06,
      "loss": 0.1445,
      "step": 149
    },
    {
      "epoch": 2.4193548387096775,
      "grad_norm": 1.484375,
      "learning_rate": 9.67741935483871e-06,
      "loss": 0.1816,
      "step": 150
    },
    {
      "epoch": 2.435483870967742,
      "grad_norm": 0.56640625,
      "learning_rate": 9.408602150537635e-06,
      "loss": 0.0811,
      "step": 151
    },
    {
      "epoch": 2.4516129032258065,
      "grad_norm": 1.625,
      "learning_rate": 9.13978494623656e-06,
      "loss": 0.5469,
      "step": 152
    },
    {
      "epoch": 2.467741935483871,
      "grad_norm": 2.109375,
      "learning_rate": 8.870967741935484e-06,
      "loss": 0.2285,
      "step": 153
    },
    {
      "epoch": 2.4838709677419355,
      "grad_norm": 2.96875,
      "learning_rate": 8.602150537634409e-06,
      "loss": 0.3691,
      "step": 154
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0703125,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0923,
      "step": 155
    },
    {
      "epoch": 2.5161290322580645,
      "grad_norm": 4.09375,
      "learning_rate": 8.064516129032258e-06,
      "loss": 0.3828,
      "step": 156
    },
    {
      "epoch": 2.532258064516129,
      "grad_norm": 1.8515625,
      "learning_rate": 7.795698924731183e-06,
      "loss": 0.2441,
      "step": 157
    },
    {
      "epoch": 2.5483870967741935,
      "grad_norm": 2.921875,
      "learning_rate": 7.526881720430108e-06,
      "loss": 0.5,
      "step": 158
    },
    {
      "epoch": 2.564516129032258,
      "grad_norm": 1.171875,
      "learning_rate": 7.258064516129033e-06,
      "loss": 0.1611,
      "step": 159
    },
    {
      "epoch": 2.5806451612903225,
      "grad_norm": 1.53125,
      "learning_rate": 6.989247311827957e-06,
      "loss": 0.1914,
      "step": 160
    },
    {
      "epoch": 2.596774193548387,
      "grad_norm": 2.421875,
      "learning_rate": 6.720430107526882e-06,
      "loss": 0.2041,
      "step": 161
    },
    {
      "epoch": 2.6129032258064515,
      "grad_norm": 2.015625,
      "learning_rate": 6.451612903225806e-06,
      "loss": 0.2793,
      "step": 162
    },
    {
      "epoch": 2.629032258064516,
      "grad_norm": 1.875,
      "learning_rate": 6.182795698924732e-06,
      "loss": 0.4238,
      "step": 163
    },
    {
      "epoch": 2.6451612903225805,
      "grad_norm": 1.7578125,
      "learning_rate": 5.9139784946236566e-06,
      "loss": 0.3535,
      "step": 164
    },
    {
      "epoch": 2.661290322580645,
      "grad_norm": 3.984375,
      "learning_rate": 5.64516129032258e-06,
      "loss": 0.2988,
      "step": 165
    },
    {
      "epoch": 2.6774193548387095,
      "grad_norm": 3.90625,
      "learning_rate": 5.376344086021506e-06,
      "loss": 0.6875,
      "step": 166
    },
    {
      "epoch": 2.693548387096774,
      "grad_norm": 2.375,
      "learning_rate": 5.1075268817204305e-06,
      "loss": 0.2949,
      "step": 167
    },
    {
      "epoch": 2.709677419354839,
      "grad_norm": 0.640625,
      "learning_rate": 4.838709677419355e-06,
      "loss": 0.084,
      "step": 168
    },
    {
      "epoch": 2.725806451612903,
      "grad_norm": 3.25,
      "learning_rate": 4.56989247311828e-06,
      "loss": 0.2334,
      "step": 169
    },
    {
      "epoch": 2.741935483870968,
      "grad_norm": 2.265625,
      "learning_rate": 4.3010752688172045e-06,
      "loss": 0.2158,
      "step": 170
    },
    {
      "epoch": 2.758064516129032,
      "grad_norm": 2.3125,
      "learning_rate": 4.032258064516129e-06,
      "loss": 0.2471,
      "step": 171
    },
    {
      "epoch": 2.774193548387097,
      "grad_norm": 1.2734375,
      "learning_rate": 3.763440860215054e-06,
      "loss": 0.1377,
      "step": 172
    },
    {
      "epoch": 2.790322580645161,
      "grad_norm": 1.3828125,
      "learning_rate": 3.4946236559139785e-06,
      "loss": 0.1885,
      "step": 173
    },
    {
      "epoch": 2.806451612903226,
      "grad_norm": 2.28125,
      "learning_rate": 3.225806451612903e-06,
      "loss": 0.209,
      "step": 174
    },
    {
      "epoch": 2.8225806451612905,
      "grad_norm": 1.90625,
      "learning_rate": 2.9569892473118283e-06,
      "loss": 0.3027,
      "step": 175
    },
    {
      "epoch": 2.838709677419355,
      "grad_norm": 2.96875,
      "learning_rate": 2.688172043010753e-06,
      "loss": 0.2295,
      "step": 176
    },
    {
      "epoch": 2.8548387096774195,
      "grad_norm": 4.03125,
      "learning_rate": 2.4193548387096776e-06,
      "loss": 0.5352,
      "step": 177
    },
    {
      "epoch": 2.870967741935484,
      "grad_norm": 1.2421875,
      "learning_rate": 2.1505376344086023e-06,
      "loss": 0.1216,
      "step": 178
    },
    {
      "epoch": 2.8870967741935485,
      "grad_norm": 2.28125,
      "learning_rate": 1.881720430107527e-06,
      "loss": 0.3145,
      "step": 179
    },
    {
      "epoch": 2.903225806451613,
      "grad_norm": 4.625,
      "learning_rate": 1.6129032258064516e-06,
      "loss": 0.8047,
      "step": 180
    },
    {
      "epoch": 2.9193548387096775,
      "grad_norm": 1.0,
      "learning_rate": 1.3440860215053765e-06,
      "loss": 0.1157,
      "step": 181
    },
    {
      "epoch": 2.935483870967742,
      "grad_norm": 3.3125,
      "learning_rate": 1.0752688172043011e-06,
      "loss": 0.5586,
      "step": 182
    },
    {
      "epoch": 2.9516129032258065,
      "grad_norm": 1.5078125,
      "learning_rate": 8.064516129032258e-07,
      "loss": 0.1484,
      "step": 183
    },
    {
      "epoch": 2.967741935483871,
      "grad_norm": 0.91796875,
      "learning_rate": 5.376344086021506e-07,
      "loss": 0.0986,
      "step": 184
    },
    {
      "epoch": 2.9838709677419355,
      "grad_norm": 2.78125,
      "learning_rate": 2.688172043010753e-07,
      "loss": 0.4648,
      "step": 185
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8125,
      "learning_rate": 0.0,
      "loss": 0.2148,
      "step": 186
    }
  ],
  "logging_steps": 1,
  "max_steps": 186,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}