{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.26737967914438504,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005347593582887701,
      "grad_norm": 10.25,
      "learning_rate": 0.0002,
      "loss": 5.5244,
      "step": 1
    },
    {
      "epoch": 0.0010695187165775401,
      "grad_norm": 5.65625,
      "learning_rate": 0.0002,
      "loss": 1.5152,
      "step": 2
    },
    {
      "epoch": 0.0016042780748663102,
      "grad_norm": 5.0,
      "learning_rate": 0.0002,
      "loss": 0.9678,
      "step": 3
    },
    {
      "epoch": 0.0021390374331550803,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002,
      "loss": 0.2586,
      "step": 4
    },
    {
      "epoch": 0.00267379679144385,
      "grad_norm": 2.21875,
      "learning_rate": 0.0002,
      "loss": 0.8375,
      "step": 5
    },
    {
      "epoch": 0.0032085561497326204,
      "grad_norm": 2.359375,
      "learning_rate": 0.0002,
      "loss": 0.8124,
      "step": 6
    },
    {
      "epoch": 0.0037433155080213902,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002,
      "loss": 0.4274,
      "step": 7
    },
    {
      "epoch": 0.0042780748663101605,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0002,
      "loss": 0.6398,
      "step": 8
    },
    {
      "epoch": 0.004812834224598931,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002,
      "loss": 0.4485,
      "step": 9
    },
    {
      "epoch": 0.0053475935828877,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002,
      "loss": 0.3639,
      "step": 10
    },
    {
      "epoch": 0.0058823529411764705,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 0.3139,
      "step": 11
    },
    {
      "epoch": 0.006417112299465241,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002,
      "loss": 0.1684,
      "step": 12
    },
    {
      "epoch": 0.006951871657754011,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002,
      "loss": 0.4753,
      "step": 13
    },
    {
      "epoch": 0.0074866310160427805,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 0.0907,
      "step": 14
    },
    {
      "epoch": 0.008021390374331552,
      "grad_norm": 1.9375,
      "learning_rate": 0.0002,
      "loss": 0.3021,
      "step": 15
    },
    {
      "epoch": 0.008556149732620321,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002,
      "loss": 0.292,
      "step": 16
    },
    {
      "epoch": 0.00909090909090909,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002,
      "loss": 0.2465,
      "step": 17
    },
    {
      "epoch": 0.009625668449197862,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002,
      "loss": 0.2335,
      "step": 18
    },
    {
      "epoch": 0.010160427807486631,
      "grad_norm": 4.84375,
      "learning_rate": 0.0002,
      "loss": 0.2976,
      "step": 19
    },
    {
      "epoch": 0.0106951871657754,
      "grad_norm": 4.84375,
      "learning_rate": 0.0002,
      "loss": 0.6077,
      "step": 20
    },
    {
      "epoch": 0.011229946524064172,
      "grad_norm": 2.25,
      "learning_rate": 0.0002,
      "loss": 0.3396,
      "step": 21
    },
    {
      "epoch": 0.011764705882352941,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002,
      "loss": 0.218,
      "step": 22
    },
    {
      "epoch": 0.01229946524064171,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002,
      "loss": 0.0862,
      "step": 23
    },
    {
      "epoch": 0.012834224598930482,
      "grad_norm": 2.65625,
      "learning_rate": 0.0002,
      "loss": 0.4492,
      "step": 24
    },
    {
      "epoch": 0.013368983957219251,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002,
      "loss": 0.1986,
      "step": 25
    },
    {
      "epoch": 0.013903743315508022,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002,
      "loss": 0.2902,
      "step": 26
    },
    {
      "epoch": 0.014438502673796792,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002,
      "loss": 0.6038,
      "step": 27
    },
    {
      "epoch": 0.014973262032085561,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002,
      "loss": 0.2869,
      "step": 28
    },
    {
      "epoch": 0.015508021390374332,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 0.2646,
      "step": 29
    },
    {
      "epoch": 0.016042780748663103,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 0.1653,
      "step": 30
    },
    {
      "epoch": 0.016577540106951873,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 0.2535,
      "step": 31
    },
    {
      "epoch": 0.017112299465240642,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002,
      "loss": 0.2485,
      "step": 32
    },
    {
      "epoch": 0.01764705882352941,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002,
      "loss": 0.2425,
      "step": 33
    },
    {
      "epoch": 0.01818181818181818,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 0.0913,
      "step": 34
    },
    {
      "epoch": 0.01871657754010695,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 0.1082,
      "step": 35
    },
    {
      "epoch": 0.019251336898395723,
      "grad_norm": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.4729,
      "step": 36
    },
    {
      "epoch": 0.019786096256684493,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002,
      "loss": 0.3255,
      "step": 37
    },
    {
      "epoch": 0.020320855614973262,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002,
      "loss": 0.4583,
      "step": 38
    },
    {
      "epoch": 0.02085561497326203,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002,
      "loss": 0.1099,
      "step": 39
    },
    {
      "epoch": 0.0213903743315508,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002,
      "loss": 0.1941,
      "step": 40
    },
    {
      "epoch": 0.021925133689839574,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002,
      "loss": 0.278,
      "step": 41
    },
    {
      "epoch": 0.022459893048128343,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 0.4062,
      "step": 42
    },
    {
      "epoch": 0.022994652406417113,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002,
      "loss": 0.4801,
      "step": 43
    },
    {
      "epoch": 0.023529411764705882,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 0.2161,
      "step": 44
    },
    {
      "epoch": 0.02406417112299465,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002,
      "loss": 0.2423,
      "step": 45
    },
    {
      "epoch": 0.02459893048128342,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002,
      "loss": 0.208,
      "step": 46
    },
    {
      "epoch": 0.025133689839572194,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0002,
      "loss": 0.2149,
      "step": 47
    },
    {
      "epoch": 0.025668449197860963,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 0.2528,
      "step": 48
    },
    {
      "epoch": 0.026203208556149733,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 0.2094,
      "step": 49
    },
    {
      "epoch": 0.026737967914438502,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0002,
      "loss": 0.1544,
      "step": 50
    },
    {
      "epoch": 0.02727272727272727,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 0.2201,
      "step": 51
    },
    {
      "epoch": 0.027807486631016044,
      "grad_norm": 2.859375,
      "learning_rate": 0.0002,
      "loss": 0.3119,
      "step": 52
    },
    {
      "epoch": 0.028342245989304814,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 0.2003,
      "step": 53
    },
    {
      "epoch": 0.028877005347593583,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 0.2708,
      "step": 54
    },
    {
      "epoch": 0.029411764705882353,
      "grad_norm": 1.515625,
      "learning_rate": 0.0002,
      "loss": 0.2502,
      "step": 55
    },
    {
      "epoch": 0.029946524064171122,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 0.2379,
      "step": 56
    },
    {
      "epoch": 0.03048128342245989,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 0.1,
      "step": 57
    },
    {
      "epoch": 0.031016042780748664,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 0.0705,
      "step": 58
    },
    {
      "epoch": 0.03155080213903743,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 0.2252,
      "step": 59
    },
    {
      "epoch": 0.03208556149732621,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0002,
      "loss": 0.2479,
      "step": 60
    },
    {
      "epoch": 0.032620320855614976,
      "grad_norm": 1.25,
      "learning_rate": 0.0002,
      "loss": 0.3067,
      "step": 61
    },
    {
      "epoch": 0.033155080213903745,
      "grad_norm": 1.390625,
      "learning_rate": 0.0002,
      "loss": 0.2026,
      "step": 62
    },
    {
      "epoch": 0.033689839572192515,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 0.1771,
      "step": 63
    },
    {
      "epoch": 0.034224598930481284,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002,
      "loss": 0.1886,
      "step": 64
    },
    {
      "epoch": 0.034759358288770054,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 0.1603,
      "step": 65
    },
    {
      "epoch": 0.03529411764705882,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 0.1577,
      "step": 66
    },
    {
      "epoch": 0.03582887700534759,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002,
      "loss": 0.2385,
      "step": 67
    },
    {
      "epoch": 0.03636363636363636,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 0.2457,
      "step": 68
    },
    {
      "epoch": 0.03689839572192513,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002,
      "loss": 0.2403,
      "step": 69
    },
    {
      "epoch": 0.0374331550802139,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002,
      "loss": 0.2959,
      "step": 70
    },
    {
      "epoch": 0.03796791443850268,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002,
      "loss": 0.2408,
      "step": 71
    },
    {
      "epoch": 0.038502673796791446,
      "grad_norm": 2.0625,
      "learning_rate": 0.0002,
      "loss": 0.5305,
      "step": 72
    },
    {
      "epoch": 0.039037433155080216,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002,
      "loss": 0.1871,
      "step": 73
    },
    {
      "epoch": 0.039572192513368985,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002,
      "loss": 0.2837,
      "step": 74
    },
    {
      "epoch": 0.040106951871657755,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 0.133,
      "step": 75
    },
    {
      "epoch": 0.040641711229946524,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002,
      "loss": 0.2151,
      "step": 76
    },
    {
      "epoch": 0.041176470588235294,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 0.1379,
      "step": 77
    },
    {
      "epoch": 0.04171122994652406,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002,
      "loss": 0.1798,
      "step": 78
    },
    {
      "epoch": 0.04224598930481283,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 0.1735,
      "step": 79
    },
    {
      "epoch": 0.0427807486631016,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 0.2279,
      "step": 80
    },
    {
      "epoch": 0.04331550802139037,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 0.2083,
      "step": 81
    },
    {
      "epoch": 0.04385026737967915,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 0.2038,
      "step": 82
    },
    {
      "epoch": 0.04438502673796792,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 0.2185,
      "step": 83
    },
    {
      "epoch": 0.044919786096256686,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0002,
      "loss": 0.1184,
      "step": 84
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 0.1767,
      "step": 85
    },
    {
      "epoch": 0.045989304812834225,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.1461,
      "step": 86
    },
    {
      "epoch": 0.046524064171122995,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 0.1723,
      "step": 87
    },
    {
      "epoch": 0.047058823529411764,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0002,
      "loss": 0.1796,
      "step": 88
    },
    {
      "epoch": 0.04759358288770053,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002,
      "loss": 0.1747,
      "step": 89
    },
    {
      "epoch": 0.0481283422459893,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 0.1606,
      "step": 90
    },
    {
      "epoch": 0.04866310160427807,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0002,
      "loss": 0.1675,
      "step": 91
    },
    {
      "epoch": 0.04919786096256684,
      "grad_norm": 2.203125,
      "learning_rate": 0.0002,
      "loss": 0.3444,
      "step": 92
    },
    {
      "epoch": 0.04973262032085562,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 0.2217,
      "step": 93
    },
    {
      "epoch": 0.05026737967914439,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0002,
      "loss": 0.2684,
      "step": 94
    },
    {
      "epoch": 0.05080213903743316,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0002,
      "loss": 0.2776,
      "step": 95
    },
    {
      "epoch": 0.051336898395721926,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0002,
      "loss": 0.1626,
      "step": 96
    },
    {
      "epoch": 0.051871657754010696,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002,
      "loss": 0.1733,
      "step": 97
    },
    {
      "epoch": 0.052406417112299465,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002,
      "loss": 0.2631,
      "step": 98
    },
    {
      "epoch": 0.052941176470588235,
      "grad_norm": 5.78125,
      "learning_rate": 0.0002,
      "loss": 0.3128,
      "step": 99
    },
    {
      "epoch": 0.053475935828877004,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0002,
      "loss": 0.3108,
      "step": 100
    },
    {
      "epoch": 0.05401069518716577,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002,
      "loss": 0.4231,
      "step": 101
    },
    {
      "epoch": 0.05454545454545454,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002,
      "loss": 0.1636,
      "step": 102
    },
    {
      "epoch": 0.05508021390374331,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002,
      "loss": 0.4548,
      "step": 103
    },
    {
      "epoch": 0.05561497326203209,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002,
      "loss": 0.1718,
      "step": 104
    },
    {
      "epoch": 0.05614973262032086,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 0.1438,
      "step": 105
    },
    {
      "epoch": 0.05668449197860963,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002,
      "loss": 0.2534,
      "step": 106
    },
    {
      "epoch": 0.0572192513368984,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0002,
      "loss": 0.2303,
      "step": 107
    },
    {
      "epoch": 0.057754010695187166,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0002,
      "loss": 0.2488,
      "step": 108
    },
    {
      "epoch": 0.058288770053475936,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002,
      "loss": 0.2962,
      "step": 109
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 0.2266,
      "step": 110
    },
    {
      "epoch": 0.059358288770053474,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 0.2208,
      "step": 111
    },
    {
      "epoch": 0.059893048128342244,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002,
      "loss": 0.2489,
      "step": 112
    },
    {
      "epoch": 0.06042780748663101,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 0.3115,
      "step": 113
    },
    {
      "epoch": 0.06096256684491978,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002,
      "loss": 0.2222,
      "step": 114
    },
    {
      "epoch": 0.06149732620320856,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002,
      "loss": 0.389,
      "step": 115
    },
    {
      "epoch": 0.06203208556149733,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 0.2148,
      "step": 116
    },
    {
      "epoch": 0.06256684491978609,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0002,
      "loss": 0.2527,
      "step": 117
    },
    {
      "epoch": 0.06310160427807486,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0002,
      "loss": 0.2395,
      "step": 118
    },
    {
      "epoch": 0.06363636363636363,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002,
      "loss": 0.3902,
      "step": 119
    },
    {
      "epoch": 0.06417112299465241,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 0.2973,
      "step": 120
    },
    {
      "epoch": 0.06470588235294118,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 0.2476,
      "step": 121
    },
    {
      "epoch": 0.06524064171122995,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002,
      "loss": 0.186,
      "step": 122
    },
    {
      "epoch": 0.06577540106951872,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0002,
      "loss": 0.165,
      "step": 123
    },
    {
      "epoch": 0.06631016042780749,
      "grad_norm": 1.546875,
      "learning_rate": 0.0002,
      "loss": 0.5423,
      "step": 124
    },
    {
      "epoch": 0.06684491978609626,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002,
      "loss": 0.4157,
      "step": 125
    },
    {
      "epoch": 0.06737967914438503,
      "grad_norm": 3.40625,
      "learning_rate": 0.0002,
      "loss": 0.5118,
      "step": 126
    },
    {
      "epoch": 0.0679144385026738,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002,
      "loss": 0.1738,
      "step": 127
    },
    {
      "epoch": 0.06844919786096257,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.0812,
      "step": 128
    },
    {
      "epoch": 0.06898395721925134,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 0.1073,
      "step": 129
    },
    {
      "epoch": 0.06951871657754011,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 0.1118,
      "step": 130
    },
    {
      "epoch": 0.07005347593582888,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002,
      "loss": 0.2042,
      "step": 131
    },
    {
      "epoch": 0.07058823529411765,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0002,
      "loss": 0.1226,
      "step": 132
    },
    {
      "epoch": 0.07112299465240642,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002,
      "loss": 0.1684,
      "step": 133
    },
    {
      "epoch": 0.07165775401069518,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0002,
      "loss": 0.1795,
      "step": 134
    },
    {
      "epoch": 0.07219251336898395,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002,
      "loss": 0.2594,
      "step": 135
    },
    {
      "epoch": 0.07272727272727272,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0002,
      "loss": 0.0331,
      "step": 136
    },
    {
      "epoch": 0.0732620320855615,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 0.1815,
      "step": 137
    },
    {
      "epoch": 0.07379679144385026,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002,
      "loss": 0.458,
      "step": 138
    },
    {
      "epoch": 0.07433155080213903,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 0.266,
      "step": 139
    },
    {
      "epoch": 0.0748663101604278,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 0.2798,
      "step": 140
    },
    {
      "epoch": 0.07540106951871657,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 0.2428,
      "step": 141
    },
    {
      "epoch": 0.07593582887700535,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 0.1931,
      "step": 142
    },
    {
      "epoch": 0.07647058823529412,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 0.2107,
      "step": 143
    },
    {
      "epoch": 0.07700534759358289,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 0.1649,
      "step": 144
    },
    {
      "epoch": 0.07754010695187166,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0002,
      "loss": 0.1408,
      "step": 145
    },
    {
      "epoch": 0.07807486631016043,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0002,
      "loss": 0.3663,
      "step": 146
    },
    {
      "epoch": 0.0786096256684492,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 0.2539,
      "step": 147
    },
    {
      "epoch": 0.07914438502673797,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 0.2954,
      "step": 148
    },
    {
      "epoch": 0.07967914438502674,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0002,
      "loss": 0.3071,
      "step": 149
    },
    {
      "epoch": 0.08021390374331551,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 0.1121,
      "step": 150
    },
    {
      "epoch": 0.08074866310160428,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 0.1059,
      "step": 151
    },
    {
      "epoch": 0.08128342245989305,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002,
      "loss": 0.2265,
      "step": 152
    },
    {
      "epoch": 0.08181818181818182,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 0.1875,
      "step": 153
    },
    {
      "epoch": 0.08235294117647059,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 0.2944,
      "step": 154
    },
    {
      "epoch": 0.08288770053475936,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002,
      "loss": 0.2663,
      "step": 155
    },
    {
      "epoch": 0.08342245989304813,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 0.1909,
      "step": 156
    },
    {
      "epoch": 0.0839572192513369,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0002,
      "loss": 0.1816,
      "step": 157
    },
    {
      "epoch": 0.08449197860962566,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 0.0903,
      "step": 158
    },
    {
      "epoch": 0.08502673796791443,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002,
      "loss": 0.1757,
      "step": 159
    },
    {
      "epoch": 0.0855614973262032,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 0.1636,
      "step": 160
    },
    {
      "epoch": 0.08609625668449197,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002,
      "loss": 0.1363,
      "step": 161
    },
    {
      "epoch": 0.08663101604278074,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 0.1262,
      "step": 162
    },
    {
      "epoch": 0.08716577540106951,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.2162,
      "step": 163
    },
    {
      "epoch": 0.0877005347593583,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0002,
      "loss": 0.0586,
      "step": 164
    },
    {
      "epoch": 0.08823529411764706,
      "grad_norm": 0.0771484375,
      "learning_rate": 0.0002,
      "loss": 0.0908,
      "step": 165
    },
    {
      "epoch": 0.08877005347593583,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002,
      "loss": 0.109,
      "step": 166
    },
    {
      "epoch": 0.0893048128342246,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002,
      "loss": 0.1283,
      "step": 167
    },
    {
      "epoch": 0.08983957219251337,
      "grad_norm": 0.0023651123046875,
      "learning_rate": 0.0002,
      "loss": 0.0003,
      "step": 168
    },
    {
      "epoch": 0.09037433155080214,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002,
      "loss": 0.0709,
      "step": 169
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0002,
      "loss": 0.1616,
      "step": 170
    },
    {
      "epoch": 0.09144385026737968,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002,
      "loss": 0.1496,
      "step": 171
    },
    {
      "epoch": 0.09197860962566845,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002,
      "loss": 0.3648,
      "step": 172
    },
    {
      "epoch": 0.09251336898395722,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002,
      "loss": 0.4267,
      "step": 173
    },
    {
      "epoch": 0.09304812834224599,
      "grad_norm": 1.71875,
      "learning_rate": 0.0002,
      "loss": 0.1812,
      "step": 174
    },
    {
      "epoch": 0.09358288770053476,
      "grad_norm": 3.34375,
      "learning_rate": 0.0002,
      "loss": 0.3108,
      "step": 175
    },
    {
      "epoch": 0.09411764705882353,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002,
      "loss": 0.2502,
      "step": 176
    },
    {
      "epoch": 0.0946524064171123,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002,
      "loss": 0.3469,
      "step": 177
    },
    {
      "epoch": 0.09518716577540107,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 0.1502,
      "step": 178
    },
    {
      "epoch": 0.09572192513368984,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002,
      "loss": 0.3546,
      "step": 179
    },
    {
      "epoch": 0.0962566844919786,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0002,
      "loss": 0.2236,
      "step": 180
    },
    {
      "epoch": 0.09679144385026738,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002,
      "loss": 0.0742,
      "step": 181
    },
    {
      "epoch": 0.09732620320855614,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 0.2046,
      "step": 182
    },
    {
      "epoch": 0.09786096256684491,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 0.1355,
      "step": 183
    },
    {
      "epoch": 0.09839572192513368,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0002,
      "loss": 0.2228,
      "step": 184
    },
    {
      "epoch": 0.09893048128342247,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002,
      "loss": 0.2756,
      "step": 185
    },
    {
      "epoch": 0.09946524064171124,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 0.228,
      "step": 186
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002,
      "loss": 0.122,
      "step": 187
    },
    {
      "epoch": 0.10053475935828877,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 0.1864,
      "step": 188
    },
    {
      "epoch": 0.10106951871657754,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 0.1586,
      "step": 189
    },
    {
      "epoch": 0.10160427807486631,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0002,
      "loss": 0.1432,
      "step": 190
    },
    {
      "epoch": 0.10213903743315508,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002,
      "loss": 0.1971,
      "step": 191
    },
    {
      "epoch": 0.10267379679144385,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 0.2811,
      "step": 192
    },
    {
      "epoch": 0.10320855614973262,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 0.1839,
      "step": 193
    },
    {
      "epoch": 0.10374331550802139,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 0.2092,
      "step": 194
    },
    {
      "epoch": 0.10427807486631016,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 0.2743,
      "step": 195
    },
    {
      "epoch": 0.10481283422459893,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 0.2752,
      "step": 196
    },
    {
      "epoch": 0.1053475935828877,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002,
      "loss": 0.1735,
      "step": 197
    },
    {
      "epoch": 0.10588235294117647,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0002,
      "loss": 0.137,
      "step": 198
    },
    {
      "epoch": 0.10641711229946524,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002,
      "loss": 0.2498,
      "step": 199
    },
    {
      "epoch": 0.10695187165775401,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 0.1308,
      "step": 200
    },
    {
      "epoch": 0.10748663101604278,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002,
      "loss": 0.2849,
      "step": 201
    },
    {
      "epoch": 0.10802139037433155,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 0.1342,
      "step": 202
    },
    {
      "epoch": 0.10855614973262032,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 0.105,
      "step": 203
    },
    {
      "epoch": 0.10909090909090909,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 0.1172,
      "step": 204
    },
    {
      "epoch": 0.10962566844919786,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 0.0579,
      "step": 205
    },
    {
      "epoch": 0.11016042780748662,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002,
      "loss": 0.1235,
      "step": 206
    },
    {
      "epoch": 0.11069518716577541,
      "grad_norm": 1.375,
      "learning_rate": 0.0002,
      "loss": 0.5436,
      "step": 207
    },
    {
      "epoch": 0.11122994652406418,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 0.2197,
      "step": 208
    },
    {
      "epoch": 0.11176470588235295,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 0.2319,
      "step": 209
    },
    {
      "epoch": 0.11229946524064172,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0002,
      "loss": 0.1091,
      "step": 210
    },
    {
      "epoch": 0.11283422459893049,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0002,
      "loss": 0.1932,
      "step": 211
    },
    {
      "epoch": 0.11336898395721925,
      "grad_norm": 0.10009765625,
      "learning_rate": 0.0002,
      "loss": 0.1829,
      "step": 212
    },
    {
      "epoch": 0.11390374331550802,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0002,
      "loss": 0.1738,
      "step": 213
    },
    {
      "epoch": 0.1144385026737968,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002,
      "loss": 0.1746,
      "step": 214
    },
    {
      "epoch": 0.11497326203208556,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002,
      "loss": 0.1831,
      "step": 215
    },
    {
      "epoch": 0.11550802139037433,
      "grad_norm": 1.609375,
      "learning_rate": 0.0002,
      "loss": 0.4379,
      "step": 216
    },
    {
      "epoch": 0.1160427807486631,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 0.1707,
      "step": 217
    },
    {
      "epoch": 0.11657754010695187,
      "grad_norm": 1.5625,
      "learning_rate": 0.0002,
      "loss": 0.1901,
      "step": 218
    },
    {
      "epoch": 0.11711229946524064,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002,
      "loss": 0.2282,
      "step": 219
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002,
      "loss": 0.1478,
      "step": 220
    },
    {
      "epoch": 0.11818181818181818,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0002,
      "loss": 0.2262,
      "step": 221
    },
    {
      "epoch": 0.11871657754010695,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 0.1373,
      "step": 222
    },
    {
      "epoch": 0.11925133689839572,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 0.123,
      "step": 223
    },
    {
      "epoch": 0.11978609625668449,
      "grad_norm": 0.08837890625,
      "learning_rate": 0.0002,
      "loss": 0.1712,
      "step": 224
    },
    {
      "epoch": 0.12032085561497326,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 0.1837,
      "step": 225
    },
    {
      "epoch": 0.12085561497326203,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002,
      "loss": 0.3184,
      "step": 226
    },
    {
      "epoch": 0.1213903743315508,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 0.1646,
      "step": 227
    },
    {
      "epoch": 0.12192513368983957,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002,
      "loss": 0.1659,
      "step": 228
    },
    {
      "epoch": 0.12245989304812835,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0002,
      "loss": 0.151,
      "step": 229
    },
    {
      "epoch": 0.12299465240641712,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.1841,
      "step": 230
    },
    {
      "epoch": 0.12352941176470589,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002,
      "loss": 0.0834,
      "step": 231
    },
    {
      "epoch": 0.12406417112299466,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002,
      "loss": 0.1417,
      "step": 232
    },
    {
      "epoch": 0.12459893048128343,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.1227,
      "step": 233
    },
    {
      "epoch": 0.12513368983957218,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002,
      "loss": 0.1835,
      "step": 234
    },
    {
      "epoch": 0.12566844919786097,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002,
      "loss": 0.3382,
      "step": 235
    },
    {
      "epoch": 0.12620320855614972,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0002,
      "loss": 0.4302,
      "step": 236
    },
    {
      "epoch": 0.1267379679144385,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002,
      "loss": 0.1895,
      "step": 237
    },
    {
      "epoch": 0.12727272727272726,
      "grad_norm": 2.5,
      "learning_rate": 0.0002,
      "loss": 0.3654,
      "step": 238
    },
    {
      "epoch": 0.12780748663101604,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002,
      "loss": 0.4818,
      "step": 239
    },
    {
      "epoch": 0.12834224598930483,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002,
      "loss": 0.3575,
      "step": 240
    },
    {
      "epoch": 0.12887700534759358,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.0789,
      "step": 241
    },
    {
      "epoch": 0.12941176470588237,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002,
      "loss": 0.2638,
      "step": 242
    },
    {
      "epoch": 0.12994652406417112,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002,
      "loss": 0.1225,
      "step": 243
    },
    {
      "epoch": 0.1304812834224599,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002,
      "loss": 0.0733,
      "step": 244
    },
    {
      "epoch": 0.13101604278074866,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 0.0766,
      "step": 245
    },
    {
      "epoch": 0.13155080213903744,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002,
      "loss": 0.3096,
      "step": 246
    },
    {
      "epoch": 0.1320855614973262,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002,
      "loss": 0.2276,
      "step": 247
    },
    {
      "epoch": 0.13262032085561498,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0002,
      "loss": 0.2357,
      "step": 248
    },
    {
      "epoch": 0.13315508021390374,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.1144,
      "step": 249
    },
    {
      "epoch": 0.13368983957219252,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002,
      "loss": 0.1118,
      "step": 250
    },
    {
      "epoch": 0.13422459893048128,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0002,
      "loss": 0.1682,
      "step": 251
    },
    {
      "epoch": 0.13475935828877006,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 0.0453,
      "step": 252
    },
    {
      "epoch": 0.13529411764705881,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 0.1895,
      "step": 253
    },
    {
      "epoch": 0.1358288770053476,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002,
      "loss": 0.4819,
      "step": 254
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002,
      "loss": 0.1763,
      "step": 255
    },
    {
      "epoch": 0.13689839572192514,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 0.268,
      "step": 256
    },
    {
      "epoch": 0.1374331550802139,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 0.1276,
      "step": 257
    },
    {
      "epoch": 0.13796791443850268,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 0.1819,
      "step": 258
    },
    {
      "epoch": 0.13850267379679143,
      "grad_norm": 0.076171875,
      "learning_rate": 0.0002,
      "loss": 0.0893,
      "step": 259
    },
    {
      "epoch": 0.13903743315508021,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002,
      "loss": 0.2154,
      "step": 260
    },
    {
      "epoch": 0.139572192513369,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002,
      "loss": 0.3956,
      "step": 261
    },
    {
      "epoch": 0.14010695187165775,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0002,
      "loss": 0.1585,
      "step": 262
    },
    {
      "epoch": 0.14064171122994654,
      "grad_norm": 0.125,
      "learning_rate": 0.0002,
      "loss": 0.0876,
      "step": 263
    },
    {
      "epoch": 0.1411764705882353,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002,
      "loss": 0.0673,
      "step": 264
    },
    {
      "epoch": 0.14171122994652408,
      "grad_norm": 4.0625,
      "learning_rate": 0.0002,
      "loss": 0.184,
      "step": 265
    },
    {
      "epoch": 0.14224598930481283,
      "grad_norm": 0.0830078125,
      "learning_rate": 0.0002,
      "loss": 0.1651,
      "step": 266
    },
    {
      "epoch": 0.14278074866310161,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 0.0828,
      "step": 267
    },
    {
      "epoch": 0.14331550802139037,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002,
      "loss": 0.1545,
      "step": 268
    },
    {
      "epoch": 0.14385026737967915,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0002,
      "loss": 0.1641,
      "step": 269
    },
    {
      "epoch": 0.1443850267379679,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 0.3245,
      "step": 270
    },
    {
      "epoch": 0.1449197860962567,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 0.1592,
      "step": 271
    },
    {
      "epoch": 0.14545454545454545,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.1406,
      "step": 272
    },
    {
      "epoch": 0.14598930481283423,
      "grad_norm": 0.40625,
      "learning_rate": 0.0002,
      "loss": 0.1832,
      "step": 273
    },
    {
      "epoch": 0.146524064171123,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0002,
      "loss": 0.091,
      "step": 274
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0002,
      "loss": 0.0943,
      "step": 275
    },
    {
      "epoch": 0.14759358288770053,
      "grad_norm": 2.359375,
      "learning_rate": 0.0002,
      "loss": 0.3355,
      "step": 276
    },
    {
      "epoch": 0.1481283422459893,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002,
      "loss": 0.1074,
      "step": 277
    },
    {
      "epoch": 0.14866310160427806,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0002,
      "loss": 0.157,
      "step": 278
    },
    {
      "epoch": 0.14919786096256685,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0002,
      "loss": 0.149,
      "step": 279
    },
    {
      "epoch": 0.1497326203208556,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002,
      "loss": 0.2129,
      "step": 280
    },
    {
      "epoch": 0.15026737967914439,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002,
      "loss": 0.3051,
      "step": 281
    },
    {
      "epoch": 0.15080213903743314,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002,
      "loss": 0.147,
      "step": 282
    },
    {
      "epoch": 0.15133689839572192,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002,
      "loss": 0.1934,
      "step": 283
    },
    {
      "epoch": 0.1518716577540107,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 0.0722,
      "step": 284
    },
    {
      "epoch": 0.15240641711229946,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 0.062,
      "step": 285
    },
    {
      "epoch": 0.15294117647058825,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002,
      "loss": 0.2352,
      "step": 286
    },
    {
      "epoch": 0.153475935828877,
      "grad_norm": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.2244,
      "step": 287
    },
    {
      "epoch": 0.15401069518716579,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 0.1342,
      "step": 288
    },
    {
      "epoch": 0.15454545454545454,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002,
      "loss": 0.206,
      "step": 289
    },
    {
      "epoch": 0.15508021390374332,
      "grad_norm": 1.375,
      "learning_rate": 0.0002,
      "loss": 0.3573,
      "step": 290
    },
    {
      "epoch": 0.15561497326203208,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002,
      "loss": 0.1673,
      "step": 291
    },
    {
      "epoch": 0.15614973262032086,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 0.0897,
      "step": 292
    },
    {
      "epoch": 0.15668449197860962,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0002,
      "loss": 0.0698,
      "step": 293
    },
    {
      "epoch": 0.1572192513368984,
      "grad_norm": 0.0849609375,
      "learning_rate": 0.0002,
      "loss": 0.0981,
      "step": 294
    },
    {
      "epoch": 0.15775401069518716,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0002,
      "loss": 0.0483,
      "step": 295
    },
    {
      "epoch": 0.15828877005347594,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 0.1486,
      "step": 296
    },
    {
      "epoch": 0.1588235294117647,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.0981,
      "step": 297
    },
    {
      "epoch": 0.15935828877005348,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002,
      "loss": 0.282,
      "step": 298
    },
    {
      "epoch": 0.15989304812834224,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 0.1445,
      "step": 299
    },
    {
      "epoch": 0.16042780748663102,
      "grad_norm": 0.25,
      "learning_rate": 0.0002,
      "loss": 0.1235,
      "step": 300
    },
    {
      "epoch": 0.16096256684491977,
      "grad_norm": 0.10693359375,
      "learning_rate": 0.0002,
      "loss": 0.0835,
      "step": 301
    },
    {
      "epoch": 0.16149732620320856,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 0.0934,
      "step": 302
    },
    {
      "epoch": 0.1620320855614973,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 0.3041,
      "step": 303
    },
    {
      "epoch": 0.1625668449197861,
      "grad_norm": 0.06640625,
      "learning_rate": 0.0002,
      "loss": 0.1435,
      "step": 304
    },
    {
      "epoch": 0.16310160427807488,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002,
      "loss": 0.1957,
      "step": 305
    },
    {
      "epoch": 0.16363636363636364,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002,
      "loss": 0.3342,
      "step": 306
    },
    {
      "epoch": 0.16417112299465242,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002,
      "loss": 0.3299,
      "step": 307
    },
    {
      "epoch": 0.16470588235294117,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002,
      "loss": 0.1717,
      "step": 308
    },
    {
      "epoch": 0.16524064171122996,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 0.2635,
      "step": 309
    },
    {
      "epoch": 0.1657754010695187,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.0933,
      "step": 310
    },
    {
      "epoch": 0.1663101604278075,
      "grad_norm": 0.08642578125,
      "learning_rate": 0.0002,
      "loss": 0.0547,
      "step": 311
    },
    {
      "epoch": 0.16684491978609625,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002,
      "loss": 0.2123,
      "step": 312
    },
    {
      "epoch": 0.16737967914438504,
      "grad_norm": 0.095703125,
      "learning_rate": 0.0002,
      "loss": 0.1559,
      "step": 313
    },
    {
      "epoch": 0.1679144385026738,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002,
      "loss": 0.243,
      "step": 314
    },
    {
      "epoch": 0.16844919786096257,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 0.2638,
      "step": 315
    },
    {
      "epoch": 0.16898395721925133,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0002,
      "loss": 0.1185,
      "step": 316
    },
    {
      "epoch": 0.1695187165775401,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002,
      "loss": 0.2719,
      "step": 317
    },
    {
      "epoch": 0.17005347593582887,
      "grad_norm": 1.375,
      "learning_rate": 0.0002,
      "loss": 0.4247,
      "step": 318
    },
    {
      "epoch": 0.17058823529411765,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0002,
      "loss": 0.3772,
      "step": 319
    },
    {
      "epoch": 0.1711229946524064,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.212,
      "step": 320
    },
    {
      "epoch": 0.1716577540106952,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0002,
      "loss": 0.1685,
      "step": 321
    },
    {
      "epoch": 0.17219251336898395,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 0.2688,
      "step": 322
    },
    {
      "epoch": 0.17272727272727273,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0002,
      "loss": 0.2073,
      "step": 323
    },
    {
      "epoch": 0.17326203208556148,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 0.2158,
      "step": 324
    },
    {
      "epoch": 0.17379679144385027,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002,
      "loss": 0.3103,
      "step": 325
    },
    {
      "epoch": 0.17433155080213902,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 0.2458,
      "step": 326
    },
    {
      "epoch": 0.1748663101604278,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 0.2203,
      "step": 327
    },
    {
      "epoch": 0.1754010695187166,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0002,
      "loss": 0.2319,
      "step": 328
    },
    {
      "epoch": 0.17593582887700535,
      "grad_norm": 0.625,
      "learning_rate": 0.0002,
      "loss": 0.2094,
      "step": 329
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0002,
      "loss": 0.2354,
      "step": 330
    },
    {
      "epoch": 0.17700534759358288,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 0.3425,
      "step": 331
    },
    {
      "epoch": 0.17754010695187167,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0002,
      "loss": 0.1983,
      "step": 332
    },
    {
      "epoch": 0.17807486631016042,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0002,
      "loss": 0.1281,
      "step": 333
    },
    {
      "epoch": 0.1786096256684492,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.1543,
      "step": 334
    },
    {
      "epoch": 0.17914438502673796,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002,
      "loss": 0.2649,
      "step": 335
    },
    {
      "epoch": 0.17967914438502675,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0002,
      "loss": 0.1132,
      "step": 336
    },
    {
      "epoch": 0.1802139037433155,
      "grad_norm": 1.0,
      "learning_rate": 0.0002,
      "loss": 0.428,
      "step": 337
    },
    {
      "epoch": 0.18074866310160428,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002,
      "loss": 0.2577,
      "step": 338
    },
    {
      "epoch": 0.18128342245989304,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.1888,
      "step": 339
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002,
      "loss": 0.2504,
      "step": 340
    },
    {
      "epoch": 0.18235294117647058,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002,
      "loss": 0.1937,
      "step": 341
    },
    {
      "epoch": 0.18288770053475936,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0002,
      "loss": 0.1778,
      "step": 342
    },
    {
      "epoch": 0.18342245989304812,
      "grad_norm": 0.06298828125,
      "learning_rate": 0.0002,
      "loss": 0.0472,
      "step": 343
    },
    {
      "epoch": 0.1839572192513369,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0002,
      "loss": 0.1782,
      "step": 344
    },
    {
      "epoch": 0.18449197860962566,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 0.2222,
      "step": 345
    },
    {
      "epoch": 0.18502673796791444,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 0.1891,
      "step": 346
    },
    {
      "epoch": 0.1855614973262032,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002,
      "loss": 0.3089,
      "step": 347
    },
    {
      "epoch": 0.18609625668449198,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002,
      "loss": 0.4823,
      "step": 348
    },
    {
      "epoch": 0.18663101604278076,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002,
      "loss": 0.1879,
      "step": 349
    },
    {
      "epoch": 0.18716577540106952,
      "grad_norm": 0.10546875,
      "learning_rate": 0.0002,
      "loss": 0.1862,
      "step": 350
    },
    {
      "epoch": 0.1877005347593583,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002,
      "loss": 0.2416,
      "step": 351
    },
    {
      "epoch": 0.18823529411764706,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0002,
      "loss": 0.2126,
      "step": 352
    },
    {
      "epoch": 0.18877005347593584,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002,
      "loss": 0.1726,
      "step": 353
    },
    {
      "epoch": 0.1893048128342246,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002,
      "loss": 0.2085,
      "step": 354
    },
    {
      "epoch": 0.18983957219251338,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002,
      "loss": 0.1478,
      "step": 355
    },
    {
      "epoch": 0.19037433155080213,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 0.1796,
      "step": 356
    },
    {
      "epoch": 0.19090909090909092,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002,
      "loss": 0.3885,
      "step": 357
    },
    {
      "epoch": 0.19144385026737967,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0002,
      "loss": 0.1843,
      "step": 358
    },
    {
      "epoch": 0.19197860962566846,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0002,
      "loss": 0.1804,
      "step": 359
    },
    {
      "epoch": 0.1925133689839572,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002,
      "loss": 0.192,
      "step": 360
    },
    {
      "epoch": 0.193048128342246,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0002,
      "loss": 0.0813,
      "step": 361
    },
    {
      "epoch": 0.19358288770053475,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0002,
      "loss": 0.0699,
      "step": 362
    },
    {
      "epoch": 0.19411764705882353,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 0.2945,
      "step": 363
    },
    {
      "epoch": 0.1946524064171123,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0002,
      "loss": 0.1997,
      "step": 364
    },
    {
      "epoch": 0.19518716577540107,
      "grad_norm": 0.28125,
      "learning_rate": 0.0002,
      "loss": 0.2366,
      "step": 365
    },
    {
      "epoch": 0.19572192513368983,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 0.2421,
      "step": 366
    },
    {
      "epoch": 0.1962566844919786,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0002,
      "loss": 0.1364,
      "step": 367
    },
    {
      "epoch": 0.19679144385026737,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002,
      "loss": 0.1706,
      "step": 368
    },
    {
      "epoch": 0.19732620320855615,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0002,
      "loss": 0.1628,
      "step": 369
    },
    {
      "epoch": 0.19786096256684493,
      "grad_norm": 0.0927734375,
      "learning_rate": 0.0002,
      "loss": 0.1654,
      "step": 370
    },
    {
      "epoch": 0.1983957219251337,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0002,
      "loss": 0.0769,
      "step": 371
    },
    {
      "epoch": 0.19893048128342247,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0002,
      "loss": 0.2066,
      "step": 372
    },
    {
      "epoch": 0.19946524064171123,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0002,
      "loss": 0.12,
      "step": 373
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002,
      "loss": 0.164,
      "step": 374
    },
    {
      "epoch": 0.20053475935828877,
      "grad_norm": 0.375,
      "learning_rate": 0.0002,
      "loss": 0.251,
      "step": 375
    },
    {
      "epoch": 0.20106951871657755,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002,
      "loss": 0.1263,
      "step": 376
    },
    {
      "epoch": 0.2016042780748663,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 0.2442,
      "step": 377
    },
    {
      "epoch": 0.2021390374331551,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002,
      "loss": 0.0631,
      "step": 378
    },
    {
      "epoch": 0.20267379679144384,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0002,
      "loss": 0.2238,
      "step": 379
    },
    {
      "epoch": 0.20320855614973263,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 0.319,
      "step": 380
    },
    {
      "epoch": 0.20374331550802138,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0002,
      "loss": 0.1655,
      "step": 381
    },
    {
      "epoch": 0.20427807486631017,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002,
      "loss": 0.1586,
      "step": 382
    },
    {
      "epoch": 0.20481283422459892,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0002,
      "loss": 0.2062,
      "step": 383
    },
    {
      "epoch": 0.2053475935828877,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0002,
      "loss": 0.2637,
      "step": 384
    },
    {
      "epoch": 0.20588235294117646,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 0.203,
      "step": 385
    },
    {
      "epoch": 0.20641711229946524,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002,
      "loss": 0.3365,
      "step": 386
    },
    {
      "epoch": 0.206951871657754,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0002,
      "loss": 0.2558,
      "step": 387
    },
    {
      "epoch": 0.20748663101604278,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 0.1074,
      "step": 388
    },
    {
      "epoch": 0.20802139037433154,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002,
      "loss": 0.1507,
      "step": 389
    },
    {
      "epoch": 0.20855614973262032,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0002,
      "loss": 0.2006,
      "step": 390
    },
    {
      "epoch": 0.20909090909090908,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002,
      "loss": 0.2195,
      "step": 391
    },
    {
      "epoch": 0.20962566844919786,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002,
      "loss": 0.1366,
      "step": 392
    },
    {
      "epoch": 0.21016042780748664,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0002,
      "loss": 0.2126,
      "step": 393
    },
    {
      "epoch": 0.2106951871657754,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002,
      "loss": 0.0699,
      "step": 394
    },
    {
      "epoch": 0.21122994652406418,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0002,
      "loss": 0.1594,
      "step": 395
    },
    {
      "epoch": 0.21176470588235294,
      "grad_norm": 0.265625,
      "learning_rate": 0.0002,
      "loss": 0.1362,
      "step": 396
    },
    {
      "epoch": 0.21229946524064172,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0002,
      "loss": 0.1458,
      "step": 397
    },
    {
      "epoch": 0.21283422459893048,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002,
      "loss": 0.2377,
      "step": 398
    },
    {
      "epoch": 0.21336898395721926,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002,
      "loss": 0.4138,
      "step": 399
    },
    {
      "epoch": 0.21390374331550802,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 0.1331,
      "step": 400
    },
    {
      "epoch": 0.2144385026737968,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002,
      "loss": 0.3063,
      "step": 401
    },
    {
      "epoch": 0.21497326203208555,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002,
      "loss": 0.15,
      "step": 402
    },
    {
      "epoch": 0.21550802139037434,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 0.1727,
      "step": 403
    },
    {
      "epoch": 0.2160427807486631,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 0.1833,
      "step": 404
    },
    {
      "epoch": 0.21657754010695188,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.0712,
      "step": 405
    },
    {
      "epoch": 0.21711229946524063,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0002,
      "loss": 0.1766,
      "step": 406
    },
    {
      "epoch": 0.21764705882352942,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0002,
      "loss": 0.2309,
      "step": 407
    },
    {
      "epoch": 0.21818181818181817,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0002,
      "loss": 0.1819,
      "step": 408
    },
    {
      "epoch": 0.21871657754010695,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0002,
      "loss": 0.1536,
      "step": 409
    },
    {
      "epoch": 0.2192513368983957,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0002,
      "loss": 0.1115,
      "step": 410
    },
    {
      "epoch": 0.2197860962566845,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002,
      "loss": 0.1675,
      "step": 411
    },
    {
      "epoch": 0.22032085561497325,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002,
      "loss": 0.1975,
      "step": 412
    },
    {
      "epoch": 0.22085561497326203,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002,
      "loss": 0.2091,
      "step": 413
    },
    {
      "epoch": 0.22139037433155082,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002,
      "loss": 0.2046,
      "step": 414
    },
    {
      "epoch": 0.22192513368983957,
      "grad_norm": 0.11474609375,
      "learning_rate": 0.0002,
      "loss": 0.1062,
      "step": 415
    },
    {
      "epoch": 0.22245989304812835,
      "grad_norm": 0.328125,
      "learning_rate": 0.0002,
      "loss": 0.252,
      "step": 416
    },
    {
      "epoch": 0.2229946524064171,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0002,
      "loss": 0.2299,
      "step": 417
    },
    {
      "epoch": 0.2235294117647059,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002,
      "loss": 0.1805,
      "step": 418
    },
    {
      "epoch": 0.22406417112299465,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0002,
      "loss": 0.2925,
      "step": 419
    },
    {
      "epoch": 0.22459893048128343,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002,
      "loss": 0.2782,
      "step": 420
    },
    {
      "epoch": 0.2251336898395722,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0002,
      "loss": 0.1329,
      "step": 421
    },
    {
      "epoch": 0.22566844919786097,
      "grad_norm": 0.421875,
      "learning_rate": 0.0002,
      "loss": 0.2275,
      "step": 422
    },
    {
      "epoch": 0.22620320855614973,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002,
      "loss": 0.1434,
      "step": 423
    },
    {
      "epoch": 0.2267379679144385,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002,
      "loss": 0.1863,
      "step": 424
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002,
      "loss": 0.2988,
      "step": 425
    },
    {
      "epoch": 0.22780748663101605,
      "grad_norm": 0.5,
      "learning_rate": 0.0002,
      "loss": 0.2001,
      "step": 426
    },
    {
      "epoch": 0.2283422459893048,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0002,
      "loss": 0.2114,
      "step": 427
    },
    {
      "epoch": 0.2288770053475936,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002,
      "loss": 0.1191,
      "step": 428
    },
    {
      "epoch": 0.22941176470588234,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002,
      "loss": 0.2393,
      "step": 429
    },
    {
      "epoch": 0.22994652406417113,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 0.1093,
      "step": 430
    },
    {
      "epoch": 0.23048128342245988,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 0.2828,
      "step": 431
    },
    {
      "epoch": 0.23101604278074866,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0002,
      "loss": 0.2073,
      "step": 432
    },
    {
      "epoch": 0.23155080213903742,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002,
      "loss": 0.3156,
      "step": 433
    },
    {
      "epoch": 0.2320855614973262,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0002,
      "loss": 0.1862,
      "step": 434
    },
    {
      "epoch": 0.232620320855615,
      "grad_norm": 0.10791015625,
      "learning_rate": 0.0002,
      "loss": 0.1061,
      "step": 435
    },
    {
      "epoch": 0.23315508021390374,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002,
      "loss": 0.2721,
      "step": 436
    },
    {
      "epoch": 0.23368983957219253,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0002,
      "loss": 0.0834,
      "step": 437
    },
    {
      "epoch": 0.23422459893048128,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0002,
      "loss": 0.1377,
      "step": 438
    },
    {
      "epoch": 0.23475935828877006,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002,
      "loss": 0.21,
      "step": 439
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002,
      "loss": 0.1004,
      "step": 440
    },
    {
      "epoch": 0.2358288770053476,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0002,
      "loss": 0.1424,
      "step": 441
    },
    {
      "epoch": 0.23636363636363636,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0002,
      "loss": 0.12,
      "step": 442
    },
    {
      "epoch": 0.23689839572192514,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0002,
      "loss": 0.155,
      "step": 443
    },
    {
      "epoch": 0.2374331550802139,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002,
      "loss": 0.1647,
      "step": 444
    },
    {
      "epoch": 0.23796791443850268,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0002,
      "loss": 0.1718,
      "step": 445
    },
    {
      "epoch": 0.23850267379679144,
      "grad_norm": 0.103515625,
      "learning_rate": 0.0002,
      "loss": 0.0817,
      "step": 446
    },
    {
      "epoch": 0.23903743315508022,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 0.1277,
      "step": 447
    },
    {
      "epoch": 0.23957219251336898,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0002,
      "loss": 0.1198,
      "step": 448
    },
    {
      "epoch": 0.24010695187165776,
      "grad_norm": 0.06787109375,
      "learning_rate": 0.0002,
      "loss": 0.0802,
      "step": 449
    },
    {
      "epoch": 0.24064171122994651,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0002,
      "loss": 0.0657,
      "step": 450
    },
    {
      "epoch": 0.2411764705882353,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.1076,
      "step": 451
    },
    {
      "epoch": 0.24171122994652405,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0002,
      "loss": 0.0829,
      "step": 452
    },
    {
      "epoch": 0.24224598930481284,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 0.1498,
      "step": 453
    },
    {
      "epoch": 0.2427807486631016,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.0773,
      "step": 454
    },
    {
      "epoch": 0.24331550802139038,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0002,
      "loss": 0.0283,
      "step": 455
    },
    {
      "epoch": 0.24385026737967913,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002,
      "loss": 0.1534,
      "step": 456
    },
    {
      "epoch": 0.24438502673796791,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0002,
      "loss": 0.2006,
      "step": 457
    },
    {
      "epoch": 0.2449197860962567,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 0.0201,
      "step": 458
    },
    {
      "epoch": 0.24545454545454545,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 0.3034,
      "step": 459
    },
    {
      "epoch": 0.24598930481283424,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002,
      "loss": 0.3538,
      "step": 460
    },
    {
      "epoch": 0.246524064171123,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002,
      "loss": 0.138,
      "step": 461
    },
    {
      "epoch": 0.24705882352941178,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002,
      "loss": 0.2199,
      "step": 462
    },
    {
      "epoch": 0.24759358288770053,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002,
      "loss": 0.3079,
      "step": 463
    },
    {
      "epoch": 0.24812834224598931,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 0.36,
      "step": 464
    },
    {
      "epoch": 0.24866310160427807,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002,
      "loss": 0.3007,
      "step": 465
    },
    {
      "epoch": 0.24919786096256685,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002,
      "loss": 0.2952,
      "step": 466
    },
    {
      "epoch": 0.2497326203208556,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002,
      "loss": 0.0584,
      "step": 467
    },
    {
      "epoch": 0.25026737967914436,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002,
      "loss": 0.0912,
      "step": 468
    },
    {
      "epoch": 0.25080213903743315,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002,
      "loss": 0.1412,
      "step": 469
    },
    {
      "epoch": 0.25133689839572193,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002,
      "loss": 0.3287,
      "step": 470
    },
    {
      "epoch": 0.2518716577540107,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0002,
      "loss": 0.2102,
      "step": 471
    },
    {
      "epoch": 0.25240641711229944,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002,
      "loss": 0.2511,
      "step": 472
    },
    {
      "epoch": 0.2529411764705882,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0002,
      "loss": 0.2573,
      "step": 473
    },
    {
      "epoch": 0.253475935828877,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0002,
      "loss": 0.1181,
      "step": 474
    },
    {
      "epoch": 0.2540106951871658,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0002,
      "loss": 0.1183,
      "step": 475
    },
    {
      "epoch": 0.2545454545454545,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0002,
      "loss": 0.1341,
      "step": 476
    },
    {
      "epoch": 0.2550802139037433,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0002,
      "loss": 0.16,
      "step": 477
    },
    {
      "epoch": 0.2556149732620321,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0002,
      "loss": 0.2133,
      "step": 478
    },
    {
      "epoch": 0.25614973262032087,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0002,
      "loss": 0.2048,
      "step": 479
    },
    {
      "epoch": 0.25668449197860965,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0002,
      "loss": 0.2218,
      "step": 480
    },
    {
      "epoch": 0.2572192513368984,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0002,
      "loss": 0.1959,
      "step": 481
    },
    {
      "epoch": 0.25775401069518716,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0002,
      "loss": 0.1816,
      "step": 482
    },
    {
      "epoch": 0.25828877005347595,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002,
      "loss": 0.1479,
      "step": 483
    },
    {
      "epoch": 0.25882352941176473,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0002,
      "loss": 0.189,
      "step": 484
    },
    {
      "epoch": 0.25935828877005346,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002,
      "loss": 0.2001,
      "step": 485
    },
    {
      "epoch": 0.25989304812834224,
      "grad_norm": 0.06591796875,
      "learning_rate": 0.0002,
      "loss": 0.0943,
      "step": 486
    },
    {
      "epoch": 0.260427807486631,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0002,
      "loss": 0.1934,
      "step": 487
    },
    {
      "epoch": 0.2609625668449198,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002,
      "loss": 0.2581,
      "step": 488
    },
    {
      "epoch": 0.26149732620320854,
      "grad_norm": 0.15625,
      "learning_rate": 0.0002,
      "loss": 0.0733,
      "step": 489
    },
    {
      "epoch": 0.2620320855614973,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002,
      "loss": 0.2736,
      "step": 490
    },
    {
      "epoch": 0.2625668449197861,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0002,
      "loss": 0.1819,
      "step": 491
    },
    {
      "epoch": 0.2631016042780749,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0002,
      "loss": 0.2011,
      "step": 492
    },
    {
      "epoch": 0.2636363636363636,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0002,
      "loss": 0.1678,
      "step": 493
    },
    {
      "epoch": 0.2641711229946524,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002,
      "loss": 0.1992,
      "step": 494
    },
    {
      "epoch": 0.2647058823529412,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002,
      "loss": 0.3075,
      "step": 495
    },
    {
      "epoch": 0.26524064171122996,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002,
      "loss": 0.1961,
      "step": 496
    },
    {
      "epoch": 0.2657754010695187,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002,
      "loss": 0.1905,
      "step": 497
    },
    {
      "epoch": 0.2663101604278075,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0002,
      "loss": 0.1238,
      "step": 498
    },
    {
      "epoch": 0.26684491978609626,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002,
      "loss": 0.122,
      "step": 499
    },
    {
      "epoch": 0.26737967914438504,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002,
      "loss": 0.192,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.729911211451351e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}