{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 0.0,
      "loss": 0.4507,
      "step": 1
    },
    {
      "epoch": 0.07,
      "learning_rate": 6.309297535714572e-07,
      "loss": 0.4332,
      "step": 2
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-06,
      "loss": 0.4095,
      "step": 3
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.2618595071429145e-06,
      "loss": 0.3777,
      "step": 4
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.4649735207179267e-06,
      "loss": 0.3387,
      "step": 5
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6309297535714573e-06,
      "loss": 0.2963,
      "step": 6
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.771243749161422e-06,
      "loss": 0.2802,
      "step": 7
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.8927892607143718e-06,
      "loss": 0.2886,
      "step": 8
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-06,
      "loss": 0.2751,
      "step": 9
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-06,
      "loss": 0.2763,
      "step": 10
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-06,
      "loss": 0.2835,
      "step": 11
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-06,
      "loss": 0.2989,
      "step": 12
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-06,
      "loss": 0.3137,
      "step": 13
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-06,
      "loss": 0.3081,
      "step": 14
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-06,
      "loss": 0.2929,
      "step": 15
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-06,
      "loss": 0.2676,
      "step": 16
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-06,
      "loss": 0.2813,
      "step": 17
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-06,
      "loss": 0.2857,
      "step": 18
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-06,
      "loss": 0.282,
      "step": 19
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-06,
      "loss": 0.2907,
      "step": 20
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-06,
      "loss": 0.291,
      "step": 21
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-06,
      "loss": 0.2951,
      "step": 22
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-06,
      "loss": 0.3005,
      "step": 23
    },
    {
      "epoch": 0.8,
      "learning_rate": 2e-06,
      "loss": 0.2923,
      "step": 24
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-06,
      "loss": 0.2706,
      "step": 25
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-06,
      "loss": 0.2516,
      "step": 26
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-06,
      "loss": 0.2789,
      "step": 27
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-06,
      "loss": 0.3096,
      "step": 28
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-06,
      "loss": 0.3061,
      "step": 29
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-06,
      "loss": 0.273,
      "step": 30
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-06,
      "loss": 0.2905,
      "step": 31
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-06,
      "loss": 0.2352,
      "step": 32
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-06,
      "loss": 0.2464,
      "step": 33
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-06,
      "loss": 0.2439,
      "step": 34
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-06,
      "loss": 0.2085,
      "step": 35
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-06,
      "loss": 0.2413,
      "step": 36
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-06,
      "loss": 0.2326,
      "step": 37
    },
    {
      "epoch": 1.27,
      "learning_rate": 2e-06,
      "loss": 0.1857,
      "step": 38
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-06,
      "loss": 0.2265,
      "step": 39
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-06,
      "loss": 0.2026,
      "step": 40
    },
    {
      "epoch": 1.37,
      "learning_rate": 2e-06,
      "loss": 0.2049,
      "step": 41
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-06,
      "loss": 0.2017,
      "step": 42
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-06,
      "loss": 0.2633,
      "step": 43
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-06,
      "loss": 0.2236,
      "step": 44
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-06,
      "loss": 0.2229,
      "step": 45
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-06,
      "loss": 0.2513,
      "step": 46
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-06,
      "loss": 0.2306,
      "step": 47
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-06,
      "loss": 0.1931,
      "step": 48
    },
    {
      "epoch": 1.63,
      "learning_rate": 2e-06,
      "loss": 0.2128,
      "step": 49
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-06,
      "loss": 0.2793,
      "step": 50
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-06,
      "loss": 0.2577,
      "step": 51
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-06,
      "loss": 0.2338,
      "step": 52
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-06,
      "loss": 0.2527,
      "step": 53
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-06,
      "loss": 0.2056,
      "step": 54
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-06,
      "loss": 0.2402,
      "step": 55
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-06,
      "loss": 0.2702,
      "step": 56
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-06,
      "loss": 0.2408,
      "step": 57
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-06,
      "loss": 0.2624,
      "step": 58
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-06,
      "loss": 0.2506,
      "step": 59
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-06,
      "loss": 0.2207,
      "step": 60
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-06,
      "loss": 0.1908,
      "step": 61
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-06,
      "loss": 0.1972,
      "step": 62
    },
    {
      "epoch": 2.1,
      "learning_rate": 2e-06,
      "loss": 0.1733,
      "step": 63
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-06,
      "loss": 0.1911,
      "step": 64
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-06,
      "loss": 0.1844,
      "step": 65
    },
    {
      "epoch": 2.2,
      "learning_rate": 2e-06,
      "loss": 0.1906,
      "step": 66
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-06,
      "loss": 0.1579,
      "step": 67
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-06,
      "loss": 0.1442,
      "step": 68
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-06,
      "loss": 0.1304,
      "step": 69
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-06,
      "loss": 0.1577,
      "step": 70
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-06,
      "loss": 0.1483,
      "step": 71
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-06,
      "loss": 0.1303,
      "step": 72
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-06,
      "loss": 0.1581,
      "step": 73
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-06,
      "loss": 0.1285,
      "step": 74
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-06,
      "loss": 0.1392,
      "step": 75
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-06,
      "loss": 0.1,
      "step": 76
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-06,
      "loss": 0.1034,
      "step": 77
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-06,
      "loss": 0.16,
      "step": 78
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-06,
      "loss": 0.155,
      "step": 79
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-06,
      "loss": 0.17,
      "step": 80
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-06,
      "loss": 0.1425,
      "step": 81
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-06,
      "loss": 0.133,
      "step": 82
    },
    {
      "epoch": 2.77,
      "learning_rate": 2e-06,
      "loss": 0.1195,
      "step": 83
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-06,
      "loss": 0.1645,
      "step": 84
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-06,
      "loss": 0.1297,
      "step": 85
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-06,
      "loss": 0.1537,
      "step": 86
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-06,
      "loss": 0.2221,
      "step": 87
    },
    {
      "epoch": 2.93,
      "learning_rate": 2e-06,
      "loss": 0.1513,
      "step": 88
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-06,
      "loss": 0.1769,
      "step": 89
    },
    {
      "epoch": 3.0,
      "learning_rate": 2e-06,
      "loss": 0.1087,
      "step": 90
    },
    {
      "epoch": 3.03,
      "learning_rate": 2e-06,
      "loss": 0.1059,
      "step": 91
    },
    {
      "epoch": 3.07,
      "learning_rate": 2e-06,
      "loss": 0.1128,
      "step": 92
    },
    {
      "epoch": 3.1,
      "learning_rate": 2e-06,
      "loss": 0.0981,
      "step": 93
    },
    {
      "epoch": 3.13,
      "learning_rate": 2e-06,
      "loss": 0.1217,
      "step": 94
    },
    {
      "epoch": 3.17,
      "learning_rate": 2e-06,
      "loss": 0.0764,
      "step": 95
    },
    {
      "epoch": 3.2,
      "learning_rate": 2e-06,
      "loss": 0.1044,
      "step": 96
    },
    {
      "epoch": 3.23,
      "learning_rate": 2e-06,
      "loss": 0.0737,
      "step": 97
    },
    {
      "epoch": 3.27,
      "learning_rate": 2e-06,
      "loss": 0.0824,
      "step": 98
    },
    {
      "epoch": 3.3,
      "learning_rate": 2e-06,
      "loss": 0.1528,
      "step": 99
    },
    {
      "epoch": 3.33,
      "learning_rate": 2e-06,
      "loss": 0.0676,
      "step": 100
    },
    {
      "epoch": 3.37,
      "learning_rate": 2e-06,
      "loss": 0.0771,
      "step": 101
    },
    {
      "epoch": 3.4,
      "learning_rate": 2e-06,
      "loss": 0.082,
      "step": 102
    },
    {
      "epoch": 3.43,
      "learning_rate": 2e-06,
      "loss": 0.0945,
      "step": 103
    },
    {
      "epoch": 3.47,
      "learning_rate": 2e-06,
      "loss": 0.044,
      "step": 104
    },
    {
      "epoch": 3.5,
      "learning_rate": 2e-06,
      "loss": 0.0504,
      "step": 105
    },
    {
      "epoch": 3.53,
      "learning_rate": 2e-06,
      "loss": 0.0386,
      "step": 106
    },
    {
      "epoch": 3.57,
      "learning_rate": 2e-06,
      "loss": 0.1174,
      "step": 107
    },
    {
      "epoch": 3.6,
      "learning_rate": 2e-06,
      "loss": 0.0509,
      "step": 108
    },
    {
      "epoch": 3.63,
      "learning_rate": 2e-06,
      "loss": 0.0503,
      "step": 109
    },
    {
      "epoch": 3.67,
      "learning_rate": 2e-06,
      "loss": 0.0833,
      "step": 110
    },
    {
      "epoch": 3.7,
      "learning_rate": 2e-06,
      "loss": 0.0789,
      "step": 111
    },
    {
      "epoch": 3.73,
      "learning_rate": 2e-06,
      "loss": 0.1442,
      "step": 112
    },
    {
      "epoch": 3.77,
      "learning_rate": 2e-06,
      "loss": 0.1368,
      "step": 113
    },
    {
      "epoch": 3.8,
      "learning_rate": 2e-06,
      "loss": 0.0607,
      "step": 114
    },
    {
      "epoch": 3.83,
      "learning_rate": 2e-06,
      "loss": 0.1192,
      "step": 115
    },
    {
      "epoch": 3.87,
      "learning_rate": 2e-06,
      "loss": 0.0617,
      "step": 116
    },
    {
      "epoch": 3.9,
      "learning_rate": 2e-06,
      "loss": 0.0811,
      "step": 117
    },
    {
      "epoch": 3.93,
      "learning_rate": 2e-06,
      "loss": 0.1615,
      "step": 118
    },
    {
      "epoch": 3.97,
      "learning_rate": 2e-06,
      "loss": 0.0693,
      "step": 119
    },
    {
      "epoch": 4.0,
      "learning_rate": 2e-06,
      "loss": 0.057,
      "step": 120
    },
    {
      "epoch": 4.03,
      "learning_rate": 2e-06,
      "loss": 0.0336,
      "step": 121
    },
    {
      "epoch": 4.07,
      "learning_rate": 2e-06,
      "loss": 0.0383,
      "step": 122
    },
    {
      "epoch": 4.1,
      "learning_rate": 2e-06,
      "loss": 0.0421,
      "step": 123
    },
    {
      "epoch": 4.13,
      "learning_rate": 2e-06,
      "loss": 0.0472,
      "step": 124
    },
    {
      "epoch": 4.17,
      "learning_rate": 2e-06,
      "loss": 0.0466,
      "step": 125
    },
    {
      "epoch": 4.2,
      "learning_rate": 2e-06,
      "loss": 0.0351,
      "step": 126
    },
    {
      "epoch": 4.23,
      "learning_rate": 2e-06,
      "loss": 0.0467,
      "step": 127
    },
    {
      "epoch": 4.27,
      "learning_rate": 2e-06,
      "loss": 0.046,
      "step": 128
    },
    {
      "epoch": 4.3,
      "learning_rate": 2e-06,
      "loss": 0.0703,
      "step": 129
    },
    {
      "epoch": 4.33,
      "learning_rate": 2e-06,
      "loss": 0.0197,
      "step": 130
    },
    {
      "epoch": 4.37,
      "learning_rate": 2e-06,
      "loss": 0.0138,
      "step": 131
    },
    {
      "epoch": 4.4,
      "learning_rate": 2e-06,
      "loss": 0.0707,
      "step": 132
    },
    {
      "epoch": 4.43,
      "learning_rate": 2e-06,
      "loss": 0.1097,
      "step": 133
    },
    {
      "epoch": 4.47,
      "learning_rate": 2e-06,
      "loss": 0.024,
      "step": 134
    },
    {
      "epoch": 4.5,
      "learning_rate": 2e-06,
      "loss": 0.0831,
      "step": 135
    },
    {
      "epoch": 4.53,
      "learning_rate": 2e-06,
      "loss": 0.0557,
      "step": 136
    },
    {
      "epoch": 4.57,
      "learning_rate": 2e-06,
      "loss": 0.0252,
      "step": 137
    },
    {
      "epoch": 4.6,
      "learning_rate": 2e-06,
      "loss": 0.0631,
      "step": 138
    },
    {
      "epoch": 4.63,
      "learning_rate": 2e-06,
      "loss": 0.0354,
      "step": 139
    },
    {
      "epoch": 4.67,
      "learning_rate": 2e-06,
      "loss": 0.1027,
      "step": 140
    },
    {
      "epoch": 4.7,
      "learning_rate": 2e-06,
      "loss": 0.0393,
      "step": 141
    },
    {
      "epoch": 4.73,
      "learning_rate": 2e-06,
      "loss": 0.0306,
      "step": 142
    },
    {
      "epoch": 4.77,
      "learning_rate": 2e-06,
      "loss": 0.0478,
      "step": 143
    },
    {
      "epoch": 4.8,
      "learning_rate": 2e-06,
      "loss": 0.0425,
      "step": 144
    },
    {
      "epoch": 4.83,
      "learning_rate": 2e-06,
      "loss": 0.0452,
      "step": 145
    },
    {
      "epoch": 4.87,
      "learning_rate": 2e-06,
      "loss": 0.0722,
      "step": 146
    },
    {
      "epoch": 4.9,
      "learning_rate": 2e-06,
      "loss": 0.0519,
      "step": 147
    },
    {
      "epoch": 4.93,
      "learning_rate": 2e-06,
      "loss": 0.0784,
      "step": 148
    },
    {
      "epoch": 4.97,
      "learning_rate": 2e-06,
      "loss": 0.0693,
      "step": 149
    },
    {
      "epoch": 5.0,
      "learning_rate": 2e-06,
      "loss": 0.0736,
      "step": 150
    },
    {
      "epoch": 5.03,
      "learning_rate": 2e-06,
      "loss": 0.015,
      "step": 151
    },
    {
      "epoch": 5.07,
      "learning_rate": 2e-06,
      "loss": 0.01,
      "step": 152
    },
    {
      "epoch": 5.1,
      "learning_rate": 2e-06,
      "loss": 0.015,
      "step": 153
    },
    {
      "epoch": 5.13,
      "learning_rate": 2e-06,
      "loss": 0.0122,
      "step": 154
    },
    {
      "epoch": 5.17,
      "learning_rate": 2e-06,
      "loss": 0.0349,
      "step": 155
    },
    {
      "epoch": 5.2,
      "learning_rate": 2e-06,
      "loss": 0.0164,
      "step": 156
    },
    {
      "epoch": 5.23,
      "learning_rate": 2e-06,
      "loss": 0.0321,
      "step": 157
    },
    {
      "epoch": 5.27,
      "learning_rate": 2e-06,
      "loss": 0.0351,
      "step": 158
    },
    {
      "epoch": 5.3,
      "learning_rate": 2e-06,
      "loss": 0.0134,
      "step": 159
    },
    {
      "epoch": 5.33,
      "learning_rate": 2e-06,
      "loss": 0.0488,
      "step": 160
    },
    {
      "epoch": 5.37,
      "learning_rate": 2e-06,
      "loss": 0.0303,
      "step": 161
    },
    {
      "epoch": 5.4,
      "learning_rate": 2e-06,
      "loss": 0.0243,
      "step": 162
    },
    {
      "epoch": 5.43,
      "learning_rate": 2e-06,
      "loss": 0.0131,
      "step": 163
    },
    {
      "epoch": 5.47,
      "learning_rate": 2e-06,
      "loss": 0.0411,
      "step": 164
    },
    {
      "epoch": 5.5,
      "learning_rate": 2e-06,
      "loss": 0.0293,
      "step": 165
    },
    {
      "epoch": 5.53,
      "learning_rate": 2e-06,
      "loss": 0.0132,
      "step": 166
    },
    {
      "epoch": 5.57,
      "learning_rate": 2e-06,
      "loss": 0.0116,
      "step": 167
    },
    {
      "epoch": 5.6,
      "learning_rate": 2e-06,
      "loss": 0.0876,
      "step": 168
    },
    {
      "epoch": 5.63,
      "learning_rate": 2e-06,
      "loss": 0.0471,
      "step": 169
    },
    {
      "epoch": 5.67,
      "learning_rate": 2e-06,
      "loss": 0.0356,
      "step": 170
    },
    {
      "epoch": 5.7,
      "learning_rate": 2e-06,
      "loss": 0.0374,
      "step": 171
    },
    {
      "epoch": 5.73,
      "learning_rate": 2e-06,
      "loss": 0.0214,
      "step": 172
    },
    {
      "epoch": 5.77,
      "learning_rate": 2e-06,
      "loss": 0.015,
      "step": 173
    },
    {
      "epoch": 5.8,
      "learning_rate": 2e-06,
      "loss": 0.02,
      "step": 174
    },
    {
      "epoch": 5.83,
      "learning_rate": 2e-06,
      "loss": 0.051,
      "step": 175
    },
    {
      "epoch": 5.87,
      "learning_rate": 2e-06,
      "loss": 0.0256,
      "step": 176
    },
    {
      "epoch": 5.9,
      "learning_rate": 2e-06,
      "loss": 0.0405,
      "step": 177
    },
    {
      "epoch": 5.93,
      "learning_rate": 2e-06,
      "loss": 0.0162,
      "step": 178
    },
    {
      "epoch": 5.97,
      "learning_rate": 2e-06,
      "loss": 0.0582,
      "step": 179
    },
    {
      "epoch": 6.0,
      "learning_rate": 2e-06,
      "loss": 0.054,
      "step": 180
    },
    {
      "epoch": 6.03,
      "learning_rate": 2e-06,
      "loss": 0.0122,
      "step": 181
    },
    {
      "epoch": 6.07,
      "learning_rate": 2e-06,
      "loss": 0.0318,
      "step": 182
    },
    {
      "epoch": 6.1,
      "learning_rate": 2e-06,
      "loss": 0.0068,
      "step": 183
    },
    {
      "epoch": 6.13,
      "learning_rate": 2e-06,
      "loss": 0.031,
      "step": 184
    },
    {
      "epoch": 6.17,
      "learning_rate": 2e-06,
      "loss": 0.0115,
      "step": 185
    },
    {
      "epoch": 6.2,
      "learning_rate": 2e-06,
      "loss": 0.0295,
      "step": 186
    },
    {
      "epoch": 6.23,
      "learning_rate": 2e-06,
      "loss": 0.0128,
      "step": 187
    },
    {
      "epoch": 6.27,
      "learning_rate": 2e-06,
      "loss": 0.0123,
      "step": 188
    },
    {
      "epoch": 6.3,
      "learning_rate": 2e-06,
      "loss": 0.0128,
      "step": 189
    },
    {
      "epoch": 6.33,
      "learning_rate": 2e-06,
      "loss": 0.0052,
      "step": 190
    },
    {
      "epoch": 6.37,
      "learning_rate": 2e-06,
      "loss": 0.0127,
      "step": 191
    },
    {
      "epoch": 6.4,
      "learning_rate": 2e-06,
      "loss": 0.0036,
      "step": 192
    },
    {
      "epoch": 6.43,
      "learning_rate": 2e-06,
      "loss": 0.0159,
      "step": 193
    },
    {
      "epoch": 6.47,
      "learning_rate": 2e-06,
      "loss": 0.008,
      "step": 194
    },
    {
      "epoch": 6.5,
      "learning_rate": 2e-06,
      "loss": 0.0038,
      "step": 195
    },
    {
      "epoch": 6.53,
      "learning_rate": 2e-06,
      "loss": 0.0428,
      "step": 196
    },
    {
      "epoch": 6.57,
      "learning_rate": 2e-06,
      "loss": 0.0281,
      "step": 197
    },
    {
      "epoch": 6.6,
      "learning_rate": 2e-06,
      "loss": 0.014,
      "step": 198
    },
    {
      "epoch": 6.63,
      "learning_rate": 2e-06,
      "loss": 0.0514,
      "step": 199
    },
    {
      "epoch": 6.67,
      "learning_rate": 2e-06,
      "loss": 0.0832,
      "step": 200
    },
    {
      "epoch": 6.7,
      "learning_rate": 2e-06,
      "loss": 0.0285,
      "step": 201
    },
    {
      "epoch": 6.73,
      "learning_rate": 2e-06,
      "loss": 0.0095,
      "step": 202
    },
    {
      "epoch": 6.77,
      "learning_rate": 2e-06,
      "loss": 0.0488,
      "step": 203
    },
    {
      "epoch": 6.8,
      "learning_rate": 2e-06,
      "loss": 0.0438,
      "step": 204
    },
    {
      "epoch": 6.83,
      "learning_rate": 2e-06,
      "loss": 0.0163,
      "step": 205
    },
    {
      "epoch": 6.87,
      "learning_rate": 2e-06,
      "loss": 0.0419,
      "step": 206
    },
    {
      "epoch": 6.9,
      "learning_rate": 2e-06,
      "loss": 0.0099,
      "step": 207
    },
    {
      "epoch": 6.93,
      "learning_rate": 2e-06,
      "loss": 0.0621,
      "step": 208
    },
    {
      "epoch": 6.97,
      "learning_rate": 2e-06,
      "loss": 0.0579,
      "step": 209
    },
    {
      "epoch": 7.0,
      "learning_rate": 2e-06,
      "loss": 0.0467,
      "step": 210
    },
    {
      "epoch": 7.03,
      "learning_rate": 2e-06,
      "loss": 0.0031,
      "step": 211
    },
    {
      "epoch": 7.07,
      "learning_rate": 2e-06,
      "loss": 0.0068,
      "step": 212
    },
    {
      "epoch": 7.1,
      "learning_rate": 2e-06,
      "loss": 0.0083,
      "step": 213
    },
    {
      "epoch": 7.13,
      "learning_rate": 2e-06,
      "loss": 0.0551,
      "step": 214
    },
    {
      "epoch": 7.17,
      "learning_rate": 2e-06,
      "loss": 0.0211,
      "step": 215
    },
    {
      "epoch": 7.2,
      "learning_rate": 2e-06,
      "loss": 0.0153,
      "step": 216
    },
    {
      "epoch": 7.23,
      "learning_rate": 2e-06,
      "loss": 0.0035,
      "step": 217
    },
    {
      "epoch": 7.27,
      "learning_rate": 2e-06,
      "loss": 0.0025,
      "step": 218
    },
    {
      "epoch": 7.3,
      "learning_rate": 2e-06,
      "loss": 0.02,
      "step": 219
    },
    {
      "epoch": 7.33,
      "learning_rate": 2e-06,
      "loss": 0.005,
      "step": 220
    },
    {
      "epoch": 7.37,
      "learning_rate": 2e-06,
      "loss": 0.0145,
      "step": 221
    },
    {
      "epoch": 7.4,
      "learning_rate": 2e-06,
      "loss": 0.0026,
      "step": 222
    },
    {
      "epoch": 7.43,
      "learning_rate": 2e-06,
      "loss": 0.0257,
      "step": 223
    },
    {
      "epoch": 7.47,
      "learning_rate": 2e-06,
      "loss": 0.0278,
      "step": 224
    },
    {
      "epoch": 7.5,
      "learning_rate": 2e-06,
      "loss": 0.0428,
      "step": 225
    },
    {
      "epoch": 7.53,
      "learning_rate": 2e-06,
      "loss": 0.0036,
      "step": 226
    },
    {
      "epoch": 7.57,
      "learning_rate": 2e-06,
      "loss": 0.0604,
      "step": 227
    },
    {
      "epoch": 7.6,
      "learning_rate": 2e-06,
      "loss": 0.0048,
      "step": 228
    },
    {
      "epoch": 7.63,
      "learning_rate": 2e-06,
      "loss": 0.0314,
      "step": 229
    },
    {
      "epoch": 7.67,
      "learning_rate": 2e-06,
      "loss": 0.0254,
      "step": 230
    },
    {
      "epoch": 7.7,
      "learning_rate": 2e-06,
      "loss": 0.015,
      "step": 231
    },
    {
      "epoch": 7.73,
      "learning_rate": 2e-06,
      "loss": 0.0417,
      "step": 232
    },
    {
      "epoch": 7.77,
      "learning_rate": 2e-06,
      "loss": 0.0044,
      "step": 233
    },
    {
      "epoch": 7.8,
      "learning_rate": 2e-06,
      "loss": 0.0134,
      "step": 234
    },
    {
      "epoch": 7.83,
      "learning_rate": 2e-06,
      "loss": 0.0353,
      "step": 235
    },
    {
      "epoch": 7.87,
      "learning_rate": 2e-06,
      "loss": 0.0168,
      "step": 236
    },
    {
      "epoch": 7.9,
      "learning_rate": 2e-06,
      "loss": 0.0449,
      "step": 237
    },
    {
      "epoch": 7.93,
      "learning_rate": 2e-06,
      "loss": 0.0204,
      "step": 238
    },
    {
      "epoch": 7.97,
      "learning_rate": 2e-06,
      "loss": 0.0271,
      "step": 239
    },
    {
      "epoch": 8.0,
      "learning_rate": 2e-06,
      "loss": 0.0034,
      "step": 240
    },
    {
      "epoch": 8.03,
      "learning_rate": 2e-06,
      "loss": 0.0051,
      "step": 241
    },
    {
      "epoch": 8.07,
      "learning_rate": 2e-06,
      "loss": 0.0018,
      "step": 242
    },
    {
      "epoch": 8.1,
      "learning_rate": 2e-06,
      "loss": 0.0016,
      "step": 243
    },
    {
      "epoch": 8.13,
      "learning_rate": 2e-06,
      "loss": 0.0205,
      "step": 244
    },
    {
      "epoch": 8.17,
      "learning_rate": 2e-06,
      "loss": 0.0006,
      "step": 245
    },
    {
      "epoch": 8.2,
      "learning_rate": 2e-06,
      "loss": 0.0095,
      "step": 246
    },
    {
      "epoch": 8.23,
      "learning_rate": 2e-06,
      "loss": 0.0529,
      "step": 247
    },
    {
      "epoch": 8.27,
      "learning_rate": 2e-06,
      "loss": 0.0021,
      "step": 248
    },
    {
      "epoch": 8.3,
      "learning_rate": 2e-06,
      "loss": 0.0619,
      "step": 249
    },
    {
      "epoch": 8.33,
      "learning_rate": 2e-06,
      "loss": 0.006,
      "step": 250
    },
    {
      "epoch": 8.37,
      "learning_rate": 2e-06,
      "loss": 0.0124,
      "step": 251
    },
    {
      "epoch": 8.4,
      "learning_rate": 2e-06,
      "loss": 0.003,
      "step": 252
    },
    {
      "epoch": 8.43,
      "learning_rate": 2e-06,
      "loss": 0.0176,
      "step": 253
    },
    {
      "epoch": 8.47,
      "learning_rate": 2e-06,
      "loss": 0.011,
      "step": 254
    },
    {
      "epoch": 8.5,
      "learning_rate": 2e-06,
      "loss": 0.0031,
      "step": 255
    },
    {
      "epoch": 8.53,
      "learning_rate": 2e-06,
      "loss": 0.0007,
      "step": 256
    },
    {
      "epoch": 8.57,
      "learning_rate": 2e-06,
      "loss": 0.002,
      "step": 257
    },
    {
      "epoch": 8.6,
      "learning_rate": 2e-06,
      "loss": 0.0225,
      "step": 258
    },
    {
      "epoch": 8.63,
      "learning_rate": 2e-06,
      "loss": 0.0065,
      "step": 259
    },
    {
      "epoch": 8.67,
      "learning_rate": 2e-06,
      "loss": 0.0004,
      "step": 260
    },
    {
      "epoch": 8.7,
      "learning_rate": 2e-06,
      "loss": 0.0302,
      "step": 261
    },
    {
      "epoch": 8.73,
      "learning_rate": 2e-06,
      "loss": 0.0459,
      "step": 262
    },
    {
      "epoch": 8.77,
      "learning_rate": 2e-06,
      "loss": 0.0263,
      "step": 263
    },
    {
      "epoch": 8.8,
      "learning_rate": 2e-06,
      "loss": 0.0059,
      "step": 264
    },
    {
      "epoch": 8.83,
      "learning_rate": 2e-06,
      "loss": 0.0303,
      "step": 265
    },
    {
      "epoch": 8.87,
      "learning_rate": 2e-06,
      "loss": 0.0022,
      "step": 266
    },
    {
      "epoch": 8.9,
      "learning_rate": 2e-06,
      "loss": 0.0185,
      "step": 267
    },
    {
      "epoch": 8.93,
      "learning_rate": 2e-06,
      "loss": 0.0267,
      "step": 268
    },
    {
      "epoch": 8.97,
      "learning_rate": 2e-06,
      "loss": 0.0063,
      "step": 269
    },
    {
      "epoch": 9.0,
      "learning_rate": 2e-06,
      "loss": 0.0347,
      "step": 270
    },
    {
      "epoch": 9.03,
      "learning_rate": 2e-06,
      "loss": 0.0018,
      "step": 271
    },
    {
      "epoch": 9.07,
      "learning_rate": 2e-06,
      "loss": 0.0202,
      "step": 272
    },
    {
      "epoch": 9.1,
      "learning_rate": 2e-06,
      "loss": 0.0022,
      "step": 273
    },
    {
      "epoch": 9.13,
      "learning_rate": 2e-06,
      "loss": 0.0014,
      "step": 274
    },
    {
      "epoch": 9.17,
      "learning_rate": 2e-06,
      "loss": 0.002,
      "step": 275
    },
    {
      "epoch": 9.2,
      "learning_rate": 2e-06,
      "loss": 0.0016,
      "step": 276
    },
    {
      "epoch": 9.23,
      "learning_rate": 2e-06,
      "loss": 0.0336,
      "step": 277
    },
    {
      "epoch": 9.27,
      "learning_rate": 2e-06,
      "loss": 0.0175,
      "step": 278
    },
    {
      "epoch": 9.3,
      "learning_rate": 2e-06,
      "loss": 0.0026,
      "step": 279
    },
    {
      "epoch": 9.33,
      "learning_rate": 2e-06,
      "loss": 0.0061,
      "step": 280
    },
    {
      "epoch": 9.37,
      "learning_rate": 2e-06,
      "loss": 0.0213,
      "step": 281
    },
    {
      "epoch": 9.4,
      "learning_rate": 2e-06,
      "loss": 0.0066,
      "step": 282
    },
    {
      "epoch": 9.43,
      "learning_rate": 2e-06,
      "loss": 0.0326,
      "step": 283
    },
    {
      "epoch": 9.47,
      "learning_rate": 2e-06,
      "loss": 0.0192,
      "step": 284
    },
    {
      "epoch": 9.5,
      "learning_rate": 2e-06,
      "loss": 0.0037,
      "step": 285
    },
    {
      "epoch": 9.53,
      "learning_rate": 2e-06,
      "loss": 0.0032,
      "step": 286
    },
    {
      "epoch": 9.57,
      "learning_rate": 2e-06,
      "loss": 0.0016,
      "step": 287
    },
    {
      "epoch": 9.6,
      "learning_rate": 2e-06,
      "loss": 0.0037,
      "step": 288
    },
    {
      "epoch": 9.63,
      "learning_rate": 2e-06,
      "loss": 0.0039,
      "step": 289
    },
    {
      "epoch": 9.67,
      "learning_rate": 2e-06,
      "loss": 0.008,
      "step": 290
    },
    {
      "epoch": 9.7,
      "learning_rate": 2e-06,
      "loss": 0.006,
      "step": 291
    },
    {
      "epoch": 9.73,
      "learning_rate": 2e-06,
      "loss": 0.0013,
      "step": 292
    },
    {
      "epoch": 9.77,
      "learning_rate": 2e-06,
      "loss": 0.0076,
      "step": 293
    },
    {
      "epoch": 9.8,
      "learning_rate": 2e-06,
      "loss": 0.0012,
      "step": 294
    },
    {
      "epoch": 9.83,
      "learning_rate": 2e-06,
      "loss": 0.0019,
      "step": 295
    },
    {
      "epoch": 9.87,
      "learning_rate": 2e-06,
      "loss": 0.0125,
      "step": 296
    },
    {
      "epoch": 9.9,
      "learning_rate": 2e-06,
      "loss": 0.0622,
      "step": 297
    },
    {
      "epoch": 9.93,
      "learning_rate": 2e-06,
      "loss": 0.0105,
      "step": 298
    },
    {
      "epoch": 9.97,
      "learning_rate": 2e-06,
      "loss": 0.0012,
      "step": 299
    },
    {
      "epoch": 10.0,
      "learning_rate": 2e-06,
      "loss": 0.0005,
      "step": 300
    },
    {
      "epoch": 10.0,
      "step": 300,
      "total_flos": 14299209400320.0,
      "train_loss": 0.09375451525129999,
      "train_runtime": 3510.756,
      "train_samples_per_second": 10.824,
      "train_steps_per_second": 0.085
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 300,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 50000,
  "total_flos": 14299209400320.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}