{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 217,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004608294930875576,
      "grad_norm": 0.4140625,
      "learning_rate": 9.953917050691245e-06,
      "loss": 1.7388,
      "step": 1
    },
    {
      "epoch": 0.009216589861751152,
      "grad_norm": 0.412109375,
      "learning_rate": 9.90783410138249e-06,
      "loss": 1.6388,
      "step": 2
    },
    {
      "epoch": 0.013824884792626729,
      "grad_norm": 0.439453125,
      "learning_rate": 9.861751152073733e-06,
      "loss": 1.7512,
      "step": 3
    },
    {
      "epoch": 0.018433179723502304,
      "grad_norm": 0.412109375,
      "learning_rate": 9.815668202764977e-06,
      "loss": 1.694,
      "step": 4
    },
    {
      "epoch": 0.02304147465437788,
      "grad_norm": 0.390625,
      "learning_rate": 9.769585253456221e-06,
      "loss": 1.672,
      "step": 5
    },
    {
      "epoch": 0.027649769585253458,
      "grad_norm": 0.390625,
      "learning_rate": 9.723502304147466e-06,
      "loss": 1.643,
      "step": 6
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 0.396484375,
      "learning_rate": 9.67741935483871e-06,
      "loss": 1.6822,
      "step": 7
    },
    {
      "epoch": 0.03686635944700461,
      "grad_norm": 0.37109375,
      "learning_rate": 9.631336405529955e-06,
      "loss": 1.6898,
      "step": 8
    },
    {
      "epoch": 0.041474654377880185,
      "grad_norm": 0.357421875,
      "learning_rate": 9.5852534562212e-06,
      "loss": 1.6326,
      "step": 9
    },
    {
      "epoch": 0.04608294930875576,
      "grad_norm": 0.55078125,
      "learning_rate": 9.539170506912442e-06,
      "loss": 1.649,
      "step": 10
    },
    {
      "epoch": 0.05069124423963134,
      "grad_norm": 0.388671875,
      "learning_rate": 9.493087557603687e-06,
      "loss": 1.5545,
      "step": 11
    },
    {
      "epoch": 0.055299539170506916,
      "grad_norm": 0.314453125,
      "learning_rate": 9.447004608294931e-06,
      "loss": 1.5782,
      "step": 12
    },
    {
      "epoch": 0.059907834101382486,
      "grad_norm": 0.29296875,
      "learning_rate": 9.400921658986176e-06,
      "loss": 1.52,
      "step": 13
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 0.2890625,
      "learning_rate": 9.35483870967742e-06,
      "loss": 1.4491,
      "step": 14
    },
    {
      "epoch": 0.06912442396313365,
      "grad_norm": 0.306640625,
      "learning_rate": 9.308755760368664e-06,
      "loss": 1.511,
      "step": 15
    },
    {
      "epoch": 0.07373271889400922,
      "grad_norm": 0.294921875,
      "learning_rate": 9.262672811059909e-06,
      "loss": 1.4852,
      "step": 16
    },
    {
      "epoch": 0.07834101382488479,
      "grad_norm": 0.3125,
      "learning_rate": 9.216589861751153e-06,
      "loss": 1.5376,
      "step": 17
    },
    {
      "epoch": 0.08294930875576037,
      "grad_norm": 0.294921875,
      "learning_rate": 9.170506912442398e-06,
      "loss": 1.5211,
      "step": 18
    },
    {
      "epoch": 0.08755760368663594,
      "grad_norm": 0.287109375,
      "learning_rate": 9.124423963133642e-06,
      "loss": 1.4544,
      "step": 19
    },
    {
      "epoch": 0.09216589861751152,
      "grad_norm": 0.30859375,
      "learning_rate": 9.078341013824885e-06,
      "loss": 1.5231,
      "step": 20
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 0.30078125,
      "learning_rate": 9.03225806451613e-06,
      "loss": 1.5041,
      "step": 21
    },
    {
      "epoch": 0.10138248847926268,
      "grad_norm": 0.2470703125,
      "learning_rate": 8.986175115207374e-06,
      "loss": 1.4218,
      "step": 22
    },
    {
      "epoch": 0.10599078341013825,
      "grad_norm": 0.25390625,
      "learning_rate": 8.940092165898619e-06,
      "loss": 1.4255,
      "step": 23
    },
    {
      "epoch": 0.11059907834101383,
      "grad_norm": 0.22265625,
      "learning_rate": 8.894009216589863e-06,
      "loss": 1.3793,
      "step": 24
    },
    {
      "epoch": 0.1152073732718894,
      "grad_norm": 0.2255859375,
      "learning_rate": 8.847926267281107e-06,
      "loss": 1.3534,
      "step": 25
    },
    {
      "epoch": 0.11981566820276497,
      "grad_norm": 0.22265625,
      "learning_rate": 8.80184331797235e-06,
      "loss": 1.4523,
      "step": 26
    },
    {
      "epoch": 0.12442396313364056,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.755760368663595e-06,
      "loss": 1.4478,
      "step": 27
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 0.2216796875,
      "learning_rate": 8.70967741935484e-06,
      "loss": 1.4346,
      "step": 28
    },
    {
      "epoch": 0.1336405529953917,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.663594470046084e-06,
      "loss": 1.4048,
      "step": 29
    },
    {
      "epoch": 0.1382488479262673,
      "grad_norm": 0.19921875,
      "learning_rate": 8.617511520737328e-06,
      "loss": 1.3774,
      "step": 30
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.220703125,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.3979,
      "step": 31
    },
    {
      "epoch": 0.14746543778801843,
      "grad_norm": 0.189453125,
      "learning_rate": 8.525345622119815e-06,
      "loss": 1.3433,
      "step": 32
    },
    {
      "epoch": 0.15207373271889402,
      "grad_norm": 0.1953125,
      "learning_rate": 8.47926267281106e-06,
      "loss": 1.3798,
      "step": 33
    },
    {
      "epoch": 0.15668202764976957,
      "grad_norm": 0.1884765625,
      "learning_rate": 8.433179723502304e-06,
      "loss": 1.3775,
      "step": 34
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 0.193359375,
      "learning_rate": 8.387096774193549e-06,
      "loss": 1.3377,
      "step": 35
    },
    {
      "epoch": 0.16589861751152074,
      "grad_norm": 0.212890625,
      "learning_rate": 8.341013824884793e-06,
      "loss": 1.3534,
      "step": 36
    },
    {
      "epoch": 0.17050691244239632,
      "grad_norm": 0.232421875,
      "learning_rate": 8.294930875576038e-06,
      "loss": 1.3038,
      "step": 37
    },
    {
      "epoch": 0.17511520737327188,
      "grad_norm": 0.16796875,
      "learning_rate": 8.248847926267282e-06,
      "loss": 1.3418,
      "step": 38
    },
    {
      "epoch": 0.17972350230414746,
      "grad_norm": 0.181640625,
      "learning_rate": 8.202764976958527e-06,
      "loss": 1.3258,
      "step": 39
    },
    {
      "epoch": 0.18433179723502305,
      "grad_norm": 0.1962890625,
      "learning_rate": 8.156682027649771e-06,
      "loss": 1.3713,
      "step": 40
    },
    {
      "epoch": 0.1889400921658986,
      "grad_norm": 0.1845703125,
      "learning_rate": 8.110599078341016e-06,
      "loss": 1.3138,
      "step": 41
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 0.1962890625,
      "learning_rate": 8.064516129032258e-06,
      "loss": 1.3456,
      "step": 42
    },
    {
      "epoch": 0.19815668202764977,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.018433179723503e-06,
      "loss": 1.306,
      "step": 43
    },
    {
      "epoch": 0.20276497695852536,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.972350230414747e-06,
      "loss": 1.3187,
      "step": 44
    },
    {
      "epoch": 0.2073732718894009,
      "grad_norm": 0.1533203125,
      "learning_rate": 7.926267281105992e-06,
      "loss": 1.2345,
      "step": 45
    },
    {
      "epoch": 0.2119815668202765,
      "grad_norm": 0.1728515625,
      "learning_rate": 7.880184331797236e-06,
      "loss": 1.2895,
      "step": 46
    },
    {
      "epoch": 0.21658986175115208,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.83410138248848e-06,
      "loss": 1.3031,
      "step": 47
    },
    {
      "epoch": 0.22119815668202766,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.788018433179724e-06,
      "loss": 1.3293,
      "step": 48
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.741935483870968e-06,
      "loss": 1.2938,
      "step": 49
    },
    {
      "epoch": 0.2304147465437788,
      "grad_norm": 0.2373046875,
      "learning_rate": 7.695852534562212e-06,
      "loss": 1.2732,
      "step": 50
    },
    {
      "epoch": 0.2350230414746544,
      "grad_norm": 0.181640625,
      "learning_rate": 7.649769585253457e-06,
      "loss": 1.2642,
      "step": 51
    },
    {
      "epoch": 0.23963133640552994,
      "grad_norm": 0.154296875,
      "learning_rate": 7.603686635944701e-06,
      "loss": 1.3026,
      "step": 52
    },
    {
      "epoch": 0.24423963133640553,
      "grad_norm": 0.1591796875,
      "learning_rate": 7.557603686635945e-06,
      "loss": 1.2532,
      "step": 53
    },
    {
      "epoch": 0.2488479262672811,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.5115207373271895e-06,
      "loss": 1.2107,
      "step": 54
    },
    {
      "epoch": 0.2534562211981567,
      "grad_norm": 0.158203125,
      "learning_rate": 7.465437788018434e-06,
      "loss": 1.2805,
      "step": 55
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.1572265625,
      "learning_rate": 7.4193548387096784e-06,
      "loss": 1.3163,
      "step": 56
    },
    {
      "epoch": 0.2626728110599078,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.373271889400923e-06,
      "loss": 1.2855,
      "step": 57
    },
    {
      "epoch": 0.2672811059907834,
      "grad_norm": 0.189453125,
      "learning_rate": 7.327188940092167e-06,
      "loss": 1.2611,
      "step": 58
    },
    {
      "epoch": 0.271889400921659,
      "grad_norm": 0.1591796875,
      "learning_rate": 7.28110599078341e-06,
      "loss": 1.2804,
      "step": 59
    },
    {
      "epoch": 0.2764976958525346,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.235023041474655e-06,
      "loss": 1.2066,
      "step": 60
    },
    {
      "epoch": 0.28110599078341014,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.188940092165899e-06,
      "loss": 1.2636,
      "step": 61
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.1513671875,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 1.2248,
      "step": 62
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 0.19921875,
      "learning_rate": 7.096774193548388e-06,
      "loss": 1.2694,
      "step": 63
    },
    {
      "epoch": 0.29493087557603687,
      "grad_norm": 0.1533203125,
      "learning_rate": 7.050691244239632e-06,
      "loss": 1.2142,
      "step": 64
    },
    {
      "epoch": 0.2995391705069124,
      "grad_norm": 0.1513671875,
      "learning_rate": 7.004608294930876e-06,
      "loss": 1.2363,
      "step": 65
    },
    {
      "epoch": 0.30414746543778803,
      "grad_norm": 0.15234375,
      "learning_rate": 6.958525345622121e-06,
      "loss": 1.2021,
      "step": 66
    },
    {
      "epoch": 0.3087557603686636,
      "grad_norm": 0.1572265625,
      "learning_rate": 6.912442396313365e-06,
      "loss": 1.2599,
      "step": 67
    },
    {
      "epoch": 0.31336405529953915,
      "grad_norm": 0.240234375,
      "learning_rate": 6.866359447004609e-06,
      "loss": 1.2112,
      "step": 68
    },
    {
      "epoch": 0.31797235023041476,
      "grad_norm": 0.1533203125,
      "learning_rate": 6.820276497695853e-06,
      "loss": 1.2716,
      "step": 69
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 0.15625,
      "learning_rate": 6.774193548387097e-06,
      "loss": 1.206,
      "step": 70
    },
    {
      "epoch": 0.3271889400921659,
      "grad_norm": 0.15625,
      "learning_rate": 6.728110599078341e-06,
      "loss": 1.2148,
      "step": 71
    },
    {
      "epoch": 0.3317972350230415,
      "grad_norm": 0.1494140625,
      "learning_rate": 6.682027649769586e-06,
      "loss": 1.2395,
      "step": 72
    },
    {
      "epoch": 0.33640552995391704,
      "grad_norm": 0.1513671875,
      "learning_rate": 6.63594470046083e-06,
      "loss": 1.2244,
      "step": 73
    },
    {
      "epoch": 0.34101382488479265,
      "grad_norm": 0.1552734375,
      "learning_rate": 6.589861751152075e-06,
      "loss": 1.2592,
      "step": 74
    },
    {
      "epoch": 0.3456221198156682,
      "grad_norm": 0.1513671875,
      "learning_rate": 6.543778801843319e-06,
      "loss": 1.1787,
      "step": 75
    },
    {
      "epoch": 0.35023041474654376,
      "grad_norm": 0.15234375,
      "learning_rate": 6.497695852534563e-06,
      "loss": 1.2108,
      "step": 76
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 0.158203125,
      "learning_rate": 6.451612903225806e-06,
      "loss": 1.2065,
      "step": 77
    },
    {
      "epoch": 0.35944700460829493,
      "grad_norm": 0.1494140625,
      "learning_rate": 6.405529953917051e-06,
      "loss": 1.1616,
      "step": 78
    },
    {
      "epoch": 0.3640552995391705,
      "grad_norm": 0.19140625,
      "learning_rate": 6.359447004608295e-06,
      "loss": 1.2286,
      "step": 79
    },
    {
      "epoch": 0.3686635944700461,
      "grad_norm": 0.1494140625,
      "learning_rate": 6.31336405529954e-06,
      "loss": 1.2252,
      "step": 80
    },
    {
      "epoch": 0.37327188940092165,
      "grad_norm": 0.158203125,
      "learning_rate": 6.267281105990783e-06,
      "loss": 1.1997,
      "step": 81
    },
    {
      "epoch": 0.3778801843317972,
      "grad_norm": 0.15234375,
      "learning_rate": 6.221198156682028e-06,
      "loss": 1.2137,
      "step": 82
    },
    {
      "epoch": 0.3824884792626728,
      "grad_norm": 0.1416015625,
      "learning_rate": 6.175115207373272e-06,
      "loss": 1.21,
      "step": 83
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 0.150390625,
      "learning_rate": 6.129032258064517e-06,
      "loss": 1.2207,
      "step": 84
    },
    {
      "epoch": 0.391705069124424,
      "grad_norm": 0.1455078125,
      "learning_rate": 6.082949308755761e-06,
      "loss": 1.1927,
      "step": 85
    },
    {
      "epoch": 0.39631336405529954,
      "grad_norm": 0.2109375,
      "learning_rate": 6.036866359447006e-06,
      "loss": 1.1553,
      "step": 86
    },
    {
      "epoch": 0.4009216589861751,
      "grad_norm": 0.1689453125,
      "learning_rate": 5.9907834101382485e-06,
      "loss": 1.1945,
      "step": 87
    },
    {
      "epoch": 0.4055299539170507,
      "grad_norm": 0.158203125,
      "learning_rate": 5.944700460829493e-06,
      "loss": 1.2145,
      "step": 88
    },
    {
      "epoch": 0.41013824884792627,
      "grad_norm": 0.16015625,
      "learning_rate": 5.8986175115207375e-06,
      "loss": 1.2411,
      "step": 89
    },
    {
      "epoch": 0.4147465437788018,
      "grad_norm": 0.150390625,
      "learning_rate": 5.852534562211982e-06,
      "loss": 1.1648,
      "step": 90
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 0.140625,
      "learning_rate": 5.806451612903226e-06,
      "loss": 1.2028,
      "step": 91
    },
    {
      "epoch": 0.423963133640553,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.76036866359447e-06,
      "loss": 1.1712,
      "step": 92
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.150390625,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.1534,
      "step": 93
    },
    {
      "epoch": 0.43317972350230416,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.668202764976959e-06,
      "loss": 1.199,
      "step": 94
    },
    {
      "epoch": 0.4377880184331797,
      "grad_norm": 0.13671875,
      "learning_rate": 5.6221198156682035e-06,
      "loss": 1.1339,
      "step": 95
    },
    {
      "epoch": 0.4423963133640553,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.576036866359448e-06,
      "loss": 1.1097,
      "step": 96
    },
    {
      "epoch": 0.4470046082949309,
      "grad_norm": 0.154296875,
      "learning_rate": 5.529953917050692e-06,
      "loss": 1.191,
      "step": 97
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 0.1376953125,
      "learning_rate": 5.483870967741935e-06,
      "loss": 1.1458,
      "step": 98
    },
    {
      "epoch": 0.45622119815668205,
      "grad_norm": 0.1494140625,
      "learning_rate": 5.43778801843318e-06,
      "loss": 1.1599,
      "step": 99
    },
    {
      "epoch": 0.4608294930875576,
      "grad_norm": 0.146484375,
      "learning_rate": 5.391705069124424e-06,
      "loss": 1.2086,
      "step": 100
    },
    {
      "epoch": 0.46543778801843316,
      "grad_norm": 0.1474609375,
      "learning_rate": 5.345622119815669e-06,
      "loss": 1.1756,
      "step": 101
    },
    {
      "epoch": 0.4700460829493088,
      "grad_norm": 0.16015625,
      "learning_rate": 5.299539170506913e-06,
      "loss": 1.1644,
      "step": 102
    },
    {
      "epoch": 0.47465437788018433,
      "grad_norm": 0.1474609375,
      "learning_rate": 5.253456221198157e-06,
      "loss": 1.1534,
      "step": 103
    },
    {
      "epoch": 0.4792626728110599,
      "grad_norm": 0.146484375,
      "learning_rate": 5.207373271889401e-06,
      "loss": 1.1422,
      "step": 104
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 0.34375,
      "learning_rate": 5.161290322580646e-06,
      "loss": 1.1176,
      "step": 105
    },
    {
      "epoch": 0.48847926267281105,
      "grad_norm": 0.146484375,
      "learning_rate": 5.11520737327189e-06,
      "loss": 1.191,
      "step": 106
    },
    {
      "epoch": 0.4930875576036866,
      "grad_norm": 0.13671875,
      "learning_rate": 5.0691244239631346e-06,
      "loss": 1.1478,
      "step": 107
    },
    {
      "epoch": 0.4976958525345622,
      "grad_norm": 0.142578125,
      "learning_rate": 5.023041474654379e-06,
      "loss": 1.1111,
      "step": 108
    },
    {
      "epoch": 0.5023041474654378,
      "grad_norm": 0.154296875,
      "learning_rate": 4.976958525345623e-06,
      "loss": 1.1853,
      "step": 109
    },
    {
      "epoch": 0.5069124423963134,
      "grad_norm": 0.13671875,
      "learning_rate": 4.930875576036866e-06,
      "loss": 1.0536,
      "step": 110
    },
    {
      "epoch": 0.511520737327189,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.884792626728111e-06,
      "loss": 1.2063,
      "step": 111
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 0.1552734375,
      "learning_rate": 4.838709677419355e-06,
      "loss": 1.2118,
      "step": 112
    },
    {
      "epoch": 0.5207373271889401,
      "grad_norm": 0.13671875,
      "learning_rate": 4.7926267281106e-06,
      "loss": 1.1711,
      "step": 113
    },
    {
      "epoch": 0.5253456221198156,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.746543778801843e-06,
      "loss": 1.0994,
      "step": 114
    },
    {
      "epoch": 0.5299539170506913,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.700460829493088e-06,
      "loss": 1.2298,
      "step": 115
    },
    {
      "epoch": 0.5345622119815668,
      "grad_norm": 0.15625,
      "learning_rate": 4.654377880184332e-06,
      "loss": 1.1533,
      "step": 116
    },
    {
      "epoch": 0.5391705069124424,
      "grad_norm": 0.1396484375,
      "learning_rate": 4.608294930875577e-06,
      "loss": 1.1314,
      "step": 117
    },
    {
      "epoch": 0.543778801843318,
      "grad_norm": 0.1328125,
      "learning_rate": 4.562211981566821e-06,
      "loss": 1.0946,
      "step": 118
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 0.150390625,
      "learning_rate": 4.516129032258065e-06,
      "loss": 1.1598,
      "step": 119
    },
    {
      "epoch": 0.5529953917050692,
      "grad_norm": 0.15625,
      "learning_rate": 4.470046082949309e-06,
      "loss": 1.205,
      "step": 120
    },
    {
      "epoch": 0.5576036866359447,
      "grad_norm": 0.158203125,
      "learning_rate": 4.423963133640554e-06,
      "loss": 1.1472,
      "step": 121
    },
    {
      "epoch": 0.5622119815668203,
      "grad_norm": 0.140625,
      "learning_rate": 4.377880184331797e-06,
      "loss": 1.161,
      "step": 122
    },
    {
      "epoch": 0.5668202764976958,
      "grad_norm": 0.1455078125,
      "learning_rate": 4.331797235023042e-06,
      "loss": 1.1368,
      "step": 123
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.181640625,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 1.1498,
      "step": 124
    },
    {
      "epoch": 0.576036866359447,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.23963133640553e-06,
      "loss": 1.1394,
      "step": 125
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 0.189453125,
      "learning_rate": 4.193548387096774e-06,
      "loss": 1.118,
      "step": 126
    },
    {
      "epoch": 0.5852534562211982,
      "grad_norm": 0.140625,
      "learning_rate": 4.147465437788019e-06,
      "loss": 1.1132,
      "step": 127
    },
    {
      "epoch": 0.5898617511520737,
      "grad_norm": 0.1416015625,
      "learning_rate": 4.101382488479263e-06,
      "loss": 1.1317,
      "step": 128
    },
    {
      "epoch": 0.5944700460829493,
      "grad_norm": 0.162109375,
      "learning_rate": 4.055299539170508e-06,
      "loss": 1.1931,
      "step": 129
    },
    {
      "epoch": 0.5990783410138248,
      "grad_norm": 0.15234375,
      "learning_rate": 4.0092165898617514e-06,
      "loss": 1.1573,
      "step": 130
    },
    {
      "epoch": 0.6036866359447005,
      "grad_norm": 0.15234375,
      "learning_rate": 3.963133640552996e-06,
      "loss": 1.1818,
      "step": 131
    },
    {
      "epoch": 0.6082949308755761,
      "grad_norm": 0.171875,
      "learning_rate": 3.91705069124424e-06,
      "loss": 1.1848,
      "step": 132
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 0.1484375,
      "learning_rate": 3.870967741935484e-06,
      "loss": 1.2125,
      "step": 133
    },
    {
      "epoch": 0.6175115207373272,
      "grad_norm": 0.1455078125,
      "learning_rate": 3.8248847926267285e-06,
      "loss": 1.0897,
      "step": 134
    },
    {
      "epoch": 0.6221198156682027,
      "grad_norm": 0.1494140625,
      "learning_rate": 3.7788018433179725e-06,
      "loss": 1.1279,
      "step": 135
    },
    {
      "epoch": 0.6267281105990783,
      "grad_norm": 0.169921875,
      "learning_rate": 3.732718894009217e-06,
      "loss": 1.1931,
      "step": 136
    },
    {
      "epoch": 0.631336405529954,
      "grad_norm": 0.1806640625,
      "learning_rate": 3.6866359447004615e-06,
      "loss": 1.1233,
      "step": 137
    },
    {
      "epoch": 0.6359447004608295,
      "grad_norm": 0.1650390625,
      "learning_rate": 3.640552995391705e-06,
      "loss": 1.1553,
      "step": 138
    },
    {
      "epoch": 0.6405529953917051,
      "grad_norm": 0.15234375,
      "learning_rate": 3.5944700460829495e-06,
      "loss": 1.1352,
      "step": 139
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 0.162109375,
      "learning_rate": 3.548387096774194e-06,
      "loss": 1.1494,
      "step": 140
    },
    {
      "epoch": 0.6497695852534562,
      "grad_norm": 0.1728515625,
      "learning_rate": 3.502304147465438e-06,
      "loss": 1.1649,
      "step": 141
    },
    {
      "epoch": 0.6543778801843319,
      "grad_norm": 0.1689453125,
      "learning_rate": 3.4562211981566825e-06,
      "loss": 1.1116,
      "step": 142
    },
    {
      "epoch": 0.6589861751152074,
      "grad_norm": 0.1640625,
      "learning_rate": 3.4101382488479266e-06,
      "loss": 1.1897,
      "step": 143
    },
    {
      "epoch": 0.663594470046083,
      "grad_norm": 0.1591796875,
      "learning_rate": 3.3640552995391706e-06,
      "loss": 1.1898,
      "step": 144
    },
    {
      "epoch": 0.6682027649769585,
      "grad_norm": 0.146484375,
      "learning_rate": 3.317972350230415e-06,
      "loss": 1.125,
      "step": 145
    },
    {
      "epoch": 0.6728110599078341,
      "grad_norm": 0.1591796875,
      "learning_rate": 3.2718894009216596e-06,
      "loss": 1.133,
      "step": 146
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 0.16015625,
      "learning_rate": 3.225806451612903e-06,
      "loss": 1.1272,
      "step": 147
    },
    {
      "epoch": 0.6820276497695853,
      "grad_norm": 0.1396484375,
      "learning_rate": 3.1797235023041477e-06,
      "loss": 1.1471,
      "step": 148
    },
    {
      "epoch": 0.6866359447004609,
      "grad_norm": 0.1513671875,
      "learning_rate": 3.1336405529953917e-06,
      "loss": 1.1574,
      "step": 149
    },
    {
      "epoch": 0.6912442396313364,
      "grad_norm": 0.158203125,
      "learning_rate": 3.087557603686636e-06,
      "loss": 1.1739,
      "step": 150
    },
    {
      "epoch": 0.695852534562212,
      "grad_norm": 0.1552734375,
      "learning_rate": 3.0414746543778806e-06,
      "loss": 1.1835,
      "step": 151
    },
    {
      "epoch": 0.7004608294930875,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.9953917050691243e-06,
      "loss": 1.1064,
      "step": 152
    },
    {
      "epoch": 0.7050691244239631,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.9493087557603687e-06,
      "loss": 1.1405,
      "step": 153
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.903225806451613e-06,
      "loss": 1.1237,
      "step": 154
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.150390625,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 1.067,
      "step": 155
    },
    {
      "epoch": 0.7188940092165899,
      "grad_norm": 0.16015625,
      "learning_rate": 2.8110599078341017e-06,
      "loss": 1.0972,
      "step": 156
    },
    {
      "epoch": 0.7235023041474654,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.764976958525346e-06,
      "loss": 1.1545,
      "step": 157
    },
    {
      "epoch": 0.728110599078341,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.71889400921659e-06,
      "loss": 1.1446,
      "step": 158
    },
    {
      "epoch": 0.7327188940092166,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.6728110599078343e-06,
      "loss": 1.1437,
      "step": 159
    },
    {
      "epoch": 0.7373271889400922,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.6267281105990783e-06,
      "loss": 1.0699,
      "step": 160
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 0.166015625,
      "learning_rate": 2.580645161290323e-06,
      "loss": 1.1981,
      "step": 161
    },
    {
      "epoch": 0.7465437788018433,
      "grad_norm": 0.150390625,
      "learning_rate": 2.5345622119815673e-06,
      "loss": 1.1747,
      "step": 162
    },
    {
      "epoch": 0.7511520737327189,
      "grad_norm": 0.154296875,
      "learning_rate": 2.4884792626728113e-06,
      "loss": 1.1183,
      "step": 163
    },
    {
      "epoch": 0.7557603686635944,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.4423963133640554e-06,
      "loss": 1.1191,
      "step": 164
    },
    {
      "epoch": 0.7603686635944701,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3963133640553e-06,
      "loss": 1.1296,
      "step": 165
    },
    {
      "epoch": 0.7649769585253456,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.350230414746544e-06,
      "loss": 1.1151,
      "step": 166
    },
    {
      "epoch": 0.7695852534562212,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3041474654377884e-06,
      "loss": 1.1226,
      "step": 167
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.2580645161290324e-06,
      "loss": 1.1928,
      "step": 168
    },
    {
      "epoch": 0.7788018433179723,
      "grad_norm": 0.15625,
      "learning_rate": 2.211981566820277e-06,
      "loss": 1.1284,
      "step": 169
    },
    {
      "epoch": 0.783410138248848,
      "grad_norm": 0.15625,
      "learning_rate": 2.165898617511521e-06,
      "loss": 1.1221,
      "step": 170
    },
    {
      "epoch": 0.7880184331797235,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.119815668202765e-06,
      "loss": 1.0839,
      "step": 171
    },
    {
      "epoch": 0.7926267281105991,
      "grad_norm": 0.162109375,
      "learning_rate": 2.0737327188940094e-06,
      "loss": 1.1434,
      "step": 172
    },
    {
      "epoch": 0.7972350230414746,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.027649769585254e-06,
      "loss": 1.0912,
      "step": 173
    },
    {
      "epoch": 0.8018433179723502,
      "grad_norm": 0.16015625,
      "learning_rate": 1.981566820276498e-06,
      "loss": 1.1158,
      "step": 174
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 0.1650390625,
      "learning_rate": 1.935483870967742e-06,
      "loss": 1.1269,
      "step": 175
    },
    {
      "epoch": 0.8110599078341014,
      "grad_norm": 0.140625,
      "learning_rate": 1.8894009216589863e-06,
      "loss": 1.0894,
      "step": 176
    },
    {
      "epoch": 0.815668202764977,
      "grad_norm": 0.197265625,
      "learning_rate": 1.8433179723502307e-06,
      "loss": 1.126,
      "step": 177
    },
    {
      "epoch": 0.8202764976958525,
      "grad_norm": 0.15234375,
      "learning_rate": 1.7972350230414748e-06,
      "loss": 1.1025,
      "step": 178
    },
    {
      "epoch": 0.8248847926267281,
      "grad_norm": 0.1650390625,
      "learning_rate": 1.751152073732719e-06,
      "loss": 1.1413,
      "step": 179
    },
    {
      "epoch": 0.8294930875576036,
      "grad_norm": 0.1796875,
      "learning_rate": 1.7050691244239633e-06,
      "loss": 1.1502,
      "step": 180
    },
    {
      "epoch": 0.8341013824884793,
      "grad_norm": 0.1611328125,
      "learning_rate": 1.6589861751152075e-06,
      "loss": 1.1985,
      "step": 181
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 0.1630859375,
      "learning_rate": 1.6129032258064516e-06,
      "loss": 1.1576,
      "step": 182
    },
    {
      "epoch": 0.8433179723502304,
      "grad_norm": 0.146484375,
      "learning_rate": 1.5668202764976959e-06,
      "loss": 1.1345,
      "step": 183
    },
    {
      "epoch": 0.847926267281106,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.5207373271889403e-06,
      "loss": 1.0959,
      "step": 184
    },
    {
      "epoch": 0.8525345622119815,
      "grad_norm": 0.1630859375,
      "learning_rate": 1.4746543778801844e-06,
      "loss": 1.0911,
      "step": 185
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.16015625,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 1.1769,
      "step": 186
    },
    {
      "epoch": 0.8617511520737328,
      "grad_norm": 0.154296875,
      "learning_rate": 1.382488479262673e-06,
      "loss": 1.1501,
      "step": 187
    },
    {
      "epoch": 0.8663594470046083,
      "grad_norm": 0.166015625,
      "learning_rate": 1.3364055299539171e-06,
      "loss": 1.1175,
      "step": 188
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 0.14453125,
      "learning_rate": 1.2903225806451614e-06,
      "loss": 1.1311,
      "step": 189
    },
    {
      "epoch": 0.8755760368663594,
      "grad_norm": 0.1689453125,
      "learning_rate": 1.2442396313364057e-06,
      "loss": 1.1673,
      "step": 190
    },
    {
      "epoch": 0.880184331797235,
      "grad_norm": 0.154296875,
      "learning_rate": 1.19815668202765e-06,
      "loss": 1.1911,
      "step": 191
    },
    {
      "epoch": 0.8847926267281107,
      "grad_norm": 0.1484375,
      "learning_rate": 1.1520737327188942e-06,
      "loss": 1.1032,
      "step": 192
    },
    {
      "epoch": 0.8894009216589862,
      "grad_norm": 0.1748046875,
      "learning_rate": 1.1059907834101384e-06,
      "loss": 1.1286,
      "step": 193
    },
    {
      "epoch": 0.8940092165898618,
      "grad_norm": 0.1572265625,
      "learning_rate": 1.0599078341013825e-06,
      "loss": 1.1544,
      "step": 194
    },
    {
      "epoch": 0.8986175115207373,
      "grad_norm": 0.1572265625,
      "learning_rate": 1.013824884792627e-06,
      "loss": 1.1807,
      "step": 195
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 0.1875,
      "learning_rate": 9.67741935483871e-07,
      "loss": 1.1156,
      "step": 196
    },
    {
      "epoch": 0.9078341013824884,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.216589861751154e-07,
      "loss": 1.1991,
      "step": 197
    },
    {
      "epoch": 0.9124423963133641,
      "grad_norm": 0.1640625,
      "learning_rate": 8.755760368663595e-07,
      "loss": 1.156,
      "step": 198
    },
    {
      "epoch": 0.9170506912442397,
      "grad_norm": 0.158203125,
      "learning_rate": 8.294930875576038e-07,
      "loss": 1.2011,
      "step": 199
    },
    {
      "epoch": 0.9216589861751152,
      "grad_norm": 0.1572265625,
      "learning_rate": 7.834101382488479e-07,
      "loss": 1.1454,
      "step": 200
    },
    {
      "epoch": 0.9262672811059908,
      "grad_norm": 0.171875,
      "learning_rate": 7.373271889400922e-07,
      "loss": 1.1033,
      "step": 201
    },
    {
      "epoch": 0.9308755760368663,
      "grad_norm": 0.1484375,
      "learning_rate": 6.912442396313365e-07,
      "loss": 1.1421,
      "step": 202
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 0.1484375,
      "learning_rate": 6.451612903225807e-07,
      "loss": 1.1448,
      "step": 203
    },
    {
      "epoch": 0.9400921658986175,
      "grad_norm": 0.1669921875,
      "learning_rate": 5.99078341013825e-07,
      "loss": 1.0805,
      "step": 204
    },
    {
      "epoch": 0.9447004608294931,
      "grad_norm": 0.146484375,
      "learning_rate": 5.529953917050692e-07,
      "loss": 1.0988,
      "step": 205
    },
    {
      "epoch": 0.9493087557603687,
      "grad_norm": 0.1484375,
      "learning_rate": 5.069124423963135e-07,
      "loss": 1.1227,
      "step": 206
    },
    {
      "epoch": 0.9539170506912442,
      "grad_norm": 0.16796875,
      "learning_rate": 4.608294930875577e-07,
      "loss": 1.1434,
      "step": 207
    },
    {
      "epoch": 0.9585253456221198,
      "grad_norm": 0.166015625,
      "learning_rate": 4.147465437788019e-07,
      "loss": 1.1174,
      "step": 208
    },
    {
      "epoch": 0.9631336405529954,
      "grad_norm": 0.150390625,
      "learning_rate": 3.686635944700461e-07,
      "loss": 1.1378,
      "step": 209
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 0.1552734375,
      "learning_rate": 3.2258064516129035e-07,
      "loss": 1.1046,
      "step": 210
    },
    {
      "epoch": 0.9723502304147466,
      "grad_norm": 0.158203125,
      "learning_rate": 2.764976958525346e-07,
      "loss": 1.1156,
      "step": 211
    },
    {
      "epoch": 0.9769585253456221,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.3041474654377884e-07,
      "loss": 1.0785,
      "step": 212
    },
    {
      "epoch": 0.9815668202764977,
      "grad_norm": 0.1611328125,
      "learning_rate": 1.8433179723502305e-07,
      "loss": 1.1603,
      "step": 213
    },
    {
      "epoch": 0.9861751152073732,
      "grad_norm": 0.1630859375,
      "learning_rate": 1.382488479262673e-07,
      "loss": 1.1627,
      "step": 214
    },
    {
      "epoch": 0.9907834101382489,
      "grad_norm": 0.1572265625,
      "learning_rate": 9.216589861751152e-08,
      "loss": 1.1749,
      "step": 215
    },
    {
      "epoch": 0.9953917050691244,
      "grad_norm": 0.158203125,
      "learning_rate": 4.608294930875576e-08,
      "loss": 1.1352,
      "step": 216
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0,
      "loss": 1.176,
      "step": 217
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 217,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.833535671169188e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}