{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 664,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0015060240963855422,
      "grad_norm": 0.40234375,
      "learning_rate": 9.984939759036144e-06,
      "loss": 1.6558,
      "step": 1
    },
    {
      "epoch": 0.0030120481927710845,
      "grad_norm": 0.62109375,
      "learning_rate": 9.96987951807229e-06,
      "loss": 1.615,
      "step": 2
    },
    {
      "epoch": 0.004518072289156626,
      "grad_norm": 0.375,
      "learning_rate": 9.954819277108435e-06,
      "loss": 1.6012,
      "step": 3
    },
    {
      "epoch": 0.006024096385542169,
      "grad_norm": 0.37890625,
      "learning_rate": 9.93975903614458e-06,
      "loss": 1.6058,
      "step": 4
    },
    {
      "epoch": 0.007530120481927711,
      "grad_norm": 0.43359375,
      "learning_rate": 9.924698795180724e-06,
      "loss": 1.5814,
      "step": 5
    },
    {
      "epoch": 0.009036144578313253,
      "grad_norm": 0.3515625,
      "learning_rate": 9.909638554216869e-06,
      "loss": 1.5468,
      "step": 6
    },
    {
      "epoch": 0.010542168674698794,
      "grad_norm": 0.33203125,
      "learning_rate": 9.894578313253013e-06,
      "loss": 1.5938,
      "step": 7
    },
    {
      "epoch": 0.012048192771084338,
      "grad_norm": 0.326171875,
      "learning_rate": 9.879518072289156e-06,
      "loss": 1.5721,
      "step": 8
    },
    {
      "epoch": 0.01355421686746988,
      "grad_norm": 0.326171875,
      "learning_rate": 9.864457831325303e-06,
      "loss": 1.5725,
      "step": 9
    },
    {
      "epoch": 0.015060240963855422,
      "grad_norm": 0.294921875,
      "learning_rate": 9.849397590361447e-06,
      "loss": 1.5319,
      "step": 10
    },
    {
      "epoch": 0.016566265060240965,
      "grad_norm": 0.283203125,
      "learning_rate": 9.834337349397592e-06,
      "loss": 1.5313,
      "step": 11
    },
    {
      "epoch": 0.018072289156626505,
      "grad_norm": 0.30078125,
      "learning_rate": 9.819277108433736e-06,
      "loss": 1.4891,
      "step": 12
    },
    {
      "epoch": 0.01957831325301205,
      "grad_norm": 0.259765625,
      "learning_rate": 9.804216867469881e-06,
      "loss": 1.4545,
      "step": 13
    },
    {
      "epoch": 0.02108433734939759,
      "grad_norm": 0.27734375,
      "learning_rate": 9.789156626506024e-06,
      "loss": 1.49,
      "step": 14
    },
    {
      "epoch": 0.022590361445783132,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.77409638554217e-06,
      "loss": 1.372,
      "step": 15
    },
    {
      "epoch": 0.024096385542168676,
      "grad_norm": 0.255859375,
      "learning_rate": 9.759036144578315e-06,
      "loss": 1.4413,
      "step": 16
    },
    {
      "epoch": 0.025602409638554216,
      "grad_norm": 0.2578125,
      "learning_rate": 9.743975903614459e-06,
      "loss": 1.5121,
      "step": 17
    },
    {
      "epoch": 0.02710843373493976,
      "grad_norm": 0.265625,
      "learning_rate": 9.728915662650604e-06,
      "loss": 1.4431,
      "step": 18
    },
    {
      "epoch": 0.0286144578313253,
      "grad_norm": 0.25,
      "learning_rate": 9.713855421686747e-06,
      "loss": 1.4002,
      "step": 19
    },
    {
      "epoch": 0.030120481927710843,
      "grad_norm": 0.275390625,
      "learning_rate": 9.698795180722893e-06,
      "loss": 1.3377,
      "step": 20
    },
    {
      "epoch": 0.03162650602409638,
      "grad_norm": 0.2265625,
      "learning_rate": 9.683734939759038e-06,
      "loss": 1.3335,
      "step": 21
    },
    {
      "epoch": 0.03313253012048193,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.668674698795182e-06,
      "loss": 1.3533,
      "step": 22
    },
    {
      "epoch": 0.03463855421686747,
      "grad_norm": 0.2138671875,
      "learning_rate": 9.653614457831327e-06,
      "loss": 1.3345,
      "step": 23
    },
    {
      "epoch": 0.03614457831325301,
      "grad_norm": 0.294921875,
      "learning_rate": 9.63855421686747e-06,
      "loss": 1.3724,
      "step": 24
    },
    {
      "epoch": 0.03765060240963856,
      "grad_norm": 0.28125,
      "learning_rate": 9.623493975903616e-06,
      "loss": 1.34,
      "step": 25
    },
    {
      "epoch": 0.0391566265060241,
      "grad_norm": 0.2041015625,
      "learning_rate": 9.60843373493976e-06,
      "loss": 1.3619,
      "step": 26
    },
    {
      "epoch": 0.04066265060240964,
      "grad_norm": 0.197265625,
      "learning_rate": 9.593373493975905e-06,
      "loss": 1.3004,
      "step": 27
    },
    {
      "epoch": 0.04216867469879518,
      "grad_norm": 0.1806640625,
      "learning_rate": 9.57831325301205e-06,
      "loss": 1.317,
      "step": 28
    },
    {
      "epoch": 0.043674698795180725,
      "grad_norm": 0.1982421875,
      "learning_rate": 9.563253012048193e-06,
      "loss": 1.2305,
      "step": 29
    },
    {
      "epoch": 0.045180722891566265,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.548192771084339e-06,
      "loss": 1.2588,
      "step": 30
    },
    {
      "epoch": 0.046686746987951805,
      "grad_norm": 0.248046875,
      "learning_rate": 9.533132530120482e-06,
      "loss": 1.2765,
      "step": 31
    },
    {
      "epoch": 0.04819277108433735,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.518072289156628e-06,
      "loss": 1.2708,
      "step": 32
    },
    {
      "epoch": 0.04969879518072289,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.503012048192773e-06,
      "loss": 1.3056,
      "step": 33
    },
    {
      "epoch": 0.05120481927710843,
      "grad_norm": 0.1923828125,
      "learning_rate": 9.487951807228916e-06,
      "loss": 1.2709,
      "step": 34
    },
    {
      "epoch": 0.05271084337349398,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.472891566265062e-06,
      "loss": 1.2502,
      "step": 35
    },
    {
      "epoch": 0.05421686746987952,
      "grad_norm": 0.25,
      "learning_rate": 9.457831325301205e-06,
      "loss": 1.2777,
      "step": 36
    },
    {
      "epoch": 0.05572289156626506,
      "grad_norm": 0.181640625,
      "learning_rate": 9.44277108433735e-06,
      "loss": 1.2395,
      "step": 37
    },
    {
      "epoch": 0.0572289156626506,
      "grad_norm": 0.162109375,
      "learning_rate": 9.427710843373494e-06,
      "loss": 1.2302,
      "step": 38
    },
    {
      "epoch": 0.058734939759036146,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.41265060240964e-06,
      "loss": 1.2535,
      "step": 39
    },
    {
      "epoch": 0.060240963855421686,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.397590361445785e-06,
      "loss": 1.2582,
      "step": 40
    },
    {
      "epoch": 0.061746987951807226,
      "grad_norm": 0.1875,
      "learning_rate": 9.382530120481928e-06,
      "loss": 1.2125,
      "step": 41
    },
    {
      "epoch": 0.06325301204819277,
      "grad_norm": 0.169921875,
      "learning_rate": 9.367469879518074e-06,
      "loss": 1.2059,
      "step": 42
    },
    {
      "epoch": 0.06475903614457831,
      "grad_norm": 0.1708984375,
      "learning_rate": 9.352409638554217e-06,
      "loss": 1.2282,
      "step": 43
    },
    {
      "epoch": 0.06626506024096386,
      "grad_norm": 0.546875,
      "learning_rate": 9.337349397590362e-06,
      "loss": 1.2547,
      "step": 44
    },
    {
      "epoch": 0.0677710843373494,
      "grad_norm": 0.1650390625,
      "learning_rate": 9.322289156626508e-06,
      "loss": 1.2055,
      "step": 45
    },
    {
      "epoch": 0.06927710843373494,
      "grad_norm": 0.1650390625,
      "learning_rate": 9.307228915662651e-06,
      "loss": 1.2073,
      "step": 46
    },
    {
      "epoch": 0.07078313253012049,
      "grad_norm": 0.16015625,
      "learning_rate": 9.292168674698796e-06,
      "loss": 1.1911,
      "step": 47
    },
    {
      "epoch": 0.07228915662650602,
      "grad_norm": 0.1669921875,
      "learning_rate": 9.27710843373494e-06,
      "loss": 1.1471,
      "step": 48
    },
    {
      "epoch": 0.07379518072289157,
      "grad_norm": 0.16015625,
      "learning_rate": 9.262048192771085e-06,
      "loss": 1.1377,
      "step": 49
    },
    {
      "epoch": 0.07530120481927711,
      "grad_norm": 0.2080078125,
      "learning_rate": 9.246987951807229e-06,
      "loss": 1.1816,
      "step": 50
    },
    {
      "epoch": 0.07680722891566265,
      "grad_norm": 0.1572265625,
      "learning_rate": 9.231927710843374e-06,
      "loss": 1.2112,
      "step": 51
    },
    {
      "epoch": 0.0783132530120482,
      "grad_norm": 0.1484375,
      "learning_rate": 9.21686746987952e-06,
      "loss": 1.1948,
      "step": 52
    },
    {
      "epoch": 0.07981927710843373,
      "grad_norm": 0.16015625,
      "learning_rate": 9.201807228915663e-06,
      "loss": 1.149,
      "step": 53
    },
    {
      "epoch": 0.08132530120481928,
      "grad_norm": 0.150390625,
      "learning_rate": 9.186746987951808e-06,
      "loss": 1.1705,
      "step": 54
    },
    {
      "epoch": 0.08283132530120482,
      "grad_norm": 0.1552734375,
      "learning_rate": 9.171686746987952e-06,
      "loss": 1.1774,
      "step": 55
    },
    {
      "epoch": 0.08433734939759036,
      "grad_norm": 0.1533203125,
      "learning_rate": 9.156626506024097e-06,
      "loss": 1.2013,
      "step": 56
    },
    {
      "epoch": 0.0858433734939759,
      "grad_norm": 0.154296875,
      "learning_rate": 9.141566265060242e-06,
      "loss": 1.1856,
      "step": 57
    },
    {
      "epoch": 0.08734939759036145,
      "grad_norm": 0.36328125,
      "learning_rate": 9.126506024096386e-06,
      "loss": 1.1949,
      "step": 58
    },
    {
      "epoch": 0.08885542168674698,
      "grad_norm": 0.15625,
      "learning_rate": 9.111445783132531e-06,
      "loss": 1.1413,
      "step": 59
    },
    {
      "epoch": 0.09036144578313253,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.096385542168675e-06,
      "loss": 1.1405,
      "step": 60
    },
    {
      "epoch": 0.09186746987951808,
      "grad_norm": 0.1474609375,
      "learning_rate": 9.08132530120482e-06,
      "loss": 1.1552,
      "step": 61
    },
    {
      "epoch": 0.09337349397590361,
      "grad_norm": 0.1640625,
      "learning_rate": 9.066265060240964e-06,
      "loss": 1.1527,
      "step": 62
    },
    {
      "epoch": 0.09487951807228916,
      "grad_norm": 0.150390625,
      "learning_rate": 9.051204819277109e-06,
      "loss": 1.1606,
      "step": 63
    },
    {
      "epoch": 0.0963855421686747,
      "grad_norm": 0.1484375,
      "learning_rate": 9.036144578313254e-06,
      "loss": 1.0633,
      "step": 64
    },
    {
      "epoch": 0.09789156626506024,
      "grad_norm": 0.146484375,
      "learning_rate": 9.021084337349398e-06,
      "loss": 1.1242,
      "step": 65
    },
    {
      "epoch": 0.09939759036144578,
      "grad_norm": 0.189453125,
      "learning_rate": 9.006024096385543e-06,
      "loss": 1.1872,
      "step": 66
    },
    {
      "epoch": 0.10090361445783133,
      "grad_norm": 0.2314453125,
      "learning_rate": 8.990963855421687e-06,
      "loss": 1.119,
      "step": 67
    },
    {
      "epoch": 0.10240963855421686,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.975903614457832e-06,
      "loss": 1.1324,
      "step": 68
    },
    {
      "epoch": 0.10391566265060241,
      "grad_norm": 0.150390625,
      "learning_rate": 8.960843373493977e-06,
      "loss": 1.1607,
      "step": 69
    },
    {
      "epoch": 0.10542168674698796,
      "grad_norm": 0.1484375,
      "learning_rate": 8.94578313253012e-06,
      "loss": 1.0983,
      "step": 70
    },
    {
      "epoch": 0.10692771084337349,
      "grad_norm": 0.1630859375,
      "learning_rate": 8.930722891566266e-06,
      "loss": 1.1718,
      "step": 71
    },
    {
      "epoch": 0.10843373493975904,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.91566265060241e-06,
      "loss": 1.1296,
      "step": 72
    },
    {
      "epoch": 0.10993975903614457,
      "grad_norm": 0.17578125,
      "learning_rate": 8.900602409638555e-06,
      "loss": 1.1293,
      "step": 73
    },
    {
      "epoch": 0.11144578313253012,
      "grad_norm": 0.181640625,
      "learning_rate": 8.885542168674699e-06,
      "loss": 1.1209,
      "step": 74
    },
    {
      "epoch": 0.11295180722891567,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.870481927710844e-06,
      "loss": 1.0991,
      "step": 75
    },
    {
      "epoch": 0.1144578313253012,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.855421686746989e-06,
      "loss": 1.1066,
      "step": 76
    },
    {
      "epoch": 0.11596385542168675,
      "grad_norm": 0.15234375,
      "learning_rate": 8.840361445783133e-06,
      "loss": 1.1067,
      "step": 77
    },
    {
      "epoch": 0.11746987951807229,
      "grad_norm": 0.146484375,
      "learning_rate": 8.825301204819278e-06,
      "loss": 1.1216,
      "step": 78
    },
    {
      "epoch": 0.11897590361445783,
      "grad_norm": 0.1376953125,
      "learning_rate": 8.810240963855422e-06,
      "loss": 1.1379,
      "step": 79
    },
    {
      "epoch": 0.12048192771084337,
      "grad_norm": 0.15234375,
      "learning_rate": 8.795180722891567e-06,
      "loss": 1.0659,
      "step": 80
    },
    {
      "epoch": 0.12198795180722892,
      "grad_norm": 0.158203125,
      "learning_rate": 8.780120481927712e-06,
      "loss": 1.1062,
      "step": 81
    },
    {
      "epoch": 0.12349397590361445,
      "grad_norm": 0.150390625,
      "learning_rate": 8.765060240963856e-06,
      "loss": 1.0756,
      "step": 82
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.16796875,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.1265,
      "step": 83
    },
    {
      "epoch": 0.12650602409638553,
      "grad_norm": 0.171875,
      "learning_rate": 8.734939759036145e-06,
      "loss": 1.0911,
      "step": 84
    },
    {
      "epoch": 0.1280120481927711,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.71987951807229e-06,
      "loss": 1.0481,
      "step": 85
    },
    {
      "epoch": 0.12951807228915663,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.704819277108435e-06,
      "loss": 1.075,
      "step": 86
    },
    {
      "epoch": 0.13102409638554216,
      "grad_norm": 0.16796875,
      "learning_rate": 8.689759036144579e-06,
      "loss": 1.0883,
      "step": 87
    },
    {
      "epoch": 0.13253012048192772,
      "grad_norm": 0.1630859375,
      "learning_rate": 8.674698795180724e-06,
      "loss": 1.0862,
      "step": 88
    },
    {
      "epoch": 0.13403614457831325,
      "grad_norm": 0.1669921875,
      "learning_rate": 8.659638554216867e-06,
      "loss": 1.0583,
      "step": 89
    },
    {
      "epoch": 0.1355421686746988,
      "grad_norm": 0.1630859375,
      "learning_rate": 8.644578313253013e-06,
      "loss": 1.0277,
      "step": 90
    },
    {
      "epoch": 0.13704819277108435,
      "grad_norm": 0.197265625,
      "learning_rate": 8.629518072289156e-06,
      "loss": 1.0803,
      "step": 91
    },
    {
      "epoch": 0.13855421686746988,
      "grad_norm": 0.2314453125,
      "learning_rate": 8.614457831325302e-06,
      "loss": 1.1316,
      "step": 92
    },
    {
      "epoch": 0.14006024096385541,
      "grad_norm": 0.169921875,
      "learning_rate": 8.599397590361447e-06,
      "loss": 1.0723,
      "step": 93
    },
    {
      "epoch": 0.14156626506024098,
      "grad_norm": 0.96875,
      "learning_rate": 8.58433734939759e-06,
      "loss": 1.1029,
      "step": 94
    },
    {
      "epoch": 0.1430722891566265,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.569277108433736e-06,
      "loss": 1.0479,
      "step": 95
    },
    {
      "epoch": 0.14457831325301204,
      "grad_norm": 0.166015625,
      "learning_rate": 8.55421686746988e-06,
      "loss": 1.1031,
      "step": 96
    },
    {
      "epoch": 0.1460843373493976,
      "grad_norm": 0.15625,
      "learning_rate": 8.539156626506025e-06,
      "loss": 1.0738,
      "step": 97
    },
    {
      "epoch": 0.14759036144578314,
      "grad_norm": 0.169921875,
      "learning_rate": 8.52409638554217e-06,
      "loss": 1.0639,
      "step": 98
    },
    {
      "epoch": 0.14909638554216867,
      "grad_norm": 0.26953125,
      "learning_rate": 8.509036144578313e-06,
      "loss": 1.0045,
      "step": 99
    },
    {
      "epoch": 0.15060240963855423,
      "grad_norm": 0.162109375,
      "learning_rate": 8.493975903614459e-06,
      "loss": 1.0362,
      "step": 100
    },
    {
      "epoch": 0.15210843373493976,
      "grad_norm": 0.166015625,
      "learning_rate": 8.478915662650602e-06,
      "loss": 1.0491,
      "step": 101
    },
    {
      "epoch": 0.1536144578313253,
      "grad_norm": 0.158203125,
      "learning_rate": 8.463855421686748e-06,
      "loss": 1.072,
      "step": 102
    },
    {
      "epoch": 0.15512048192771086,
      "grad_norm": 0.15625,
      "learning_rate": 8.448795180722891e-06,
      "loss": 1.0447,
      "step": 103
    },
    {
      "epoch": 0.1566265060240964,
      "grad_norm": 0.208984375,
      "learning_rate": 8.433734939759038e-06,
      "loss": 1.0931,
      "step": 104
    },
    {
      "epoch": 0.15813253012048192,
      "grad_norm": 0.17578125,
      "learning_rate": 8.418674698795182e-06,
      "loss": 1.0441,
      "step": 105
    },
    {
      "epoch": 0.15963855421686746,
      "grad_norm": 0.1826171875,
      "learning_rate": 8.403614457831325e-06,
      "loss": 1.059,
      "step": 106
    },
    {
      "epoch": 0.16114457831325302,
      "grad_norm": 0.15234375,
      "learning_rate": 8.38855421686747e-06,
      "loss": 1.0333,
      "step": 107
    },
    {
      "epoch": 0.16265060240963855,
      "grad_norm": 0.166015625,
      "learning_rate": 8.373493975903614e-06,
      "loss": 1.1011,
      "step": 108
    },
    {
      "epoch": 0.16415662650602408,
      "grad_norm": 0.203125,
      "learning_rate": 8.35843373493976e-06,
      "loss": 1.0635,
      "step": 109
    },
    {
      "epoch": 0.16566265060240964,
      "grad_norm": 0.2451171875,
      "learning_rate": 8.343373493975905e-06,
      "loss": 1.0353,
      "step": 110
    },
    {
      "epoch": 0.16716867469879518,
      "grad_norm": 0.181640625,
      "learning_rate": 8.32831325301205e-06,
      "loss": 1.0758,
      "step": 111
    },
    {
      "epoch": 0.1686746987951807,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.313253012048194e-06,
      "loss": 1.0596,
      "step": 112
    },
    {
      "epoch": 0.17018072289156627,
      "grad_norm": 0.1689453125,
      "learning_rate": 8.298192771084339e-06,
      "loss": 1.0519,
      "step": 113
    },
    {
      "epoch": 0.1716867469879518,
      "grad_norm": 0.154296875,
      "learning_rate": 8.283132530120482e-06,
      "loss": 1.0448,
      "step": 114
    },
    {
      "epoch": 0.17319277108433734,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.268072289156626e-06,
      "loss": 1.033,
      "step": 115
    },
    {
      "epoch": 0.1746987951807229,
      "grad_norm": 0.1630859375,
      "learning_rate": 8.253012048192773e-06,
      "loss": 1.0412,
      "step": 116
    },
    {
      "epoch": 0.17620481927710843,
      "grad_norm": 0.169921875,
      "learning_rate": 8.237951807228917e-06,
      "loss": 1.0562,
      "step": 117
    },
    {
      "epoch": 0.17771084337349397,
      "grad_norm": 0.189453125,
      "learning_rate": 8.222891566265062e-06,
      "loss": 1.0168,
      "step": 118
    },
    {
      "epoch": 0.17921686746987953,
      "grad_norm": 0.15234375,
      "learning_rate": 8.207831325301205e-06,
      "loss": 1.0295,
      "step": 119
    },
    {
      "epoch": 0.18072289156626506,
      "grad_norm": 0.2236328125,
      "learning_rate": 8.19277108433735e-06,
      "loss": 1.0971,
      "step": 120
    },
    {
      "epoch": 0.1822289156626506,
      "grad_norm": 0.1650390625,
      "learning_rate": 8.177710843373494e-06,
      "loss": 1.0501,
      "step": 121
    },
    {
      "epoch": 0.18373493975903615,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.16265060240964e-06,
      "loss": 1.0596,
      "step": 122
    },
    {
      "epoch": 0.1852409638554217,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.147590361445785e-06,
      "loss": 1.0345,
      "step": 123
    },
    {
      "epoch": 0.18674698795180722,
      "grad_norm": 0.150390625,
      "learning_rate": 8.132530120481928e-06,
      "loss": 0.9912,
      "step": 124
    },
    {
      "epoch": 0.18825301204819278,
      "grad_norm": 0.1982421875,
      "learning_rate": 8.117469879518074e-06,
      "loss": 1.0262,
      "step": 125
    },
    {
      "epoch": 0.1897590361445783,
      "grad_norm": 0.15625,
      "learning_rate": 8.102409638554217e-06,
      "loss": 1.0151,
      "step": 126
    },
    {
      "epoch": 0.19126506024096385,
      "grad_norm": 0.154296875,
      "learning_rate": 8.087349397590362e-06,
      "loss": 1.0169,
      "step": 127
    },
    {
      "epoch": 0.1927710843373494,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.072289156626508e-06,
      "loss": 1.0354,
      "step": 128
    },
    {
      "epoch": 0.19427710843373494,
      "grad_norm": 0.1640625,
      "learning_rate": 8.057228915662651e-06,
      "loss": 1.0328,
      "step": 129
    },
    {
      "epoch": 0.19578313253012047,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.042168674698797e-06,
      "loss": 1.0143,
      "step": 130
    },
    {
      "epoch": 0.19728915662650603,
      "grad_norm": 0.17578125,
      "learning_rate": 8.02710843373494e-06,
      "loss": 1.0673,
      "step": 131
    },
    {
      "epoch": 0.19879518072289157,
      "grad_norm": 0.162109375,
      "learning_rate": 8.012048192771085e-06,
      "loss": 1.0429,
      "step": 132
    },
    {
      "epoch": 0.2003012048192771,
      "grad_norm": 0.255859375,
      "learning_rate": 7.996987951807229e-06,
      "loss": 1.0198,
      "step": 133
    },
    {
      "epoch": 0.20180722891566266,
      "grad_norm": 0.265625,
      "learning_rate": 7.981927710843374e-06,
      "loss": 1.0198,
      "step": 134
    },
    {
      "epoch": 0.2033132530120482,
      "grad_norm": 0.181640625,
      "learning_rate": 7.96686746987952e-06,
      "loss": 1.0717,
      "step": 135
    },
    {
      "epoch": 0.20481927710843373,
      "grad_norm": 0.177734375,
      "learning_rate": 7.951807228915663e-06,
      "loss": 1.0131,
      "step": 136
    },
    {
      "epoch": 0.2063253012048193,
      "grad_norm": 0.162109375,
      "learning_rate": 7.936746987951808e-06,
      "loss": 1.0365,
      "step": 137
    },
    {
      "epoch": 0.20783132530120482,
      "grad_norm": 0.16796875,
      "learning_rate": 7.921686746987952e-06,
      "loss": 0.9985,
      "step": 138
    },
    {
      "epoch": 0.20933734939759036,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.906626506024097e-06,
      "loss": 1.037,
      "step": 139
    },
    {
      "epoch": 0.21084337349397592,
      "grad_norm": 0.1953125,
      "learning_rate": 7.891566265060243e-06,
      "loss": 0.9897,
      "step": 140
    },
    {
      "epoch": 0.21234939759036145,
      "grad_norm": 0.166015625,
      "learning_rate": 7.876506024096386e-06,
      "loss": 0.9719,
      "step": 141
    },
    {
      "epoch": 0.21385542168674698,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.861445783132531e-06,
      "loss": 1.0496,
      "step": 142
    },
    {
      "epoch": 0.21536144578313254,
      "grad_norm": 0.216796875,
      "learning_rate": 7.846385542168675e-06,
      "loss": 1.0327,
      "step": 143
    },
    {
      "epoch": 0.21686746987951808,
      "grad_norm": 0.2314453125,
      "learning_rate": 7.83132530120482e-06,
      "loss": 1.0296,
      "step": 144
    },
    {
      "epoch": 0.2183734939759036,
      "grad_norm": 0.1630859375,
      "learning_rate": 7.816265060240964e-06,
      "loss": 0.9811,
      "step": 145
    },
    {
      "epoch": 0.21987951807228914,
      "grad_norm": 0.19921875,
      "learning_rate": 7.801204819277109e-06,
      "loss": 1.0259,
      "step": 146
    },
    {
      "epoch": 0.2213855421686747,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.786144578313254e-06,
      "loss": 1.0157,
      "step": 147
    },
    {
      "epoch": 0.22289156626506024,
      "grad_norm": 0.1875,
      "learning_rate": 7.771084337349398e-06,
      "loss": 1.0016,
      "step": 148
    },
    {
      "epoch": 0.22439759036144577,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.756024096385543e-06,
      "loss": 1.0296,
      "step": 149
    },
    {
      "epoch": 0.22590361445783133,
      "grad_norm": 0.29296875,
      "learning_rate": 7.740963855421687e-06,
      "loss": 0.99,
      "step": 150
    },
    {
      "epoch": 0.22740963855421686,
      "grad_norm": 0.171875,
      "learning_rate": 7.725903614457832e-06,
      "loss": 0.9976,
      "step": 151
    },
    {
      "epoch": 0.2289156626506024,
      "grad_norm": 0.16015625,
      "learning_rate": 7.710843373493977e-06,
      "loss": 1.0379,
      "step": 152
    },
    {
      "epoch": 0.23042168674698796,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.695783132530121e-06,
      "loss": 0.9811,
      "step": 153
    },
    {
      "epoch": 0.2319277108433735,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.680722891566266e-06,
      "loss": 1.0035,
      "step": 154
    },
    {
      "epoch": 0.23343373493975902,
      "grad_norm": 0.169921875,
      "learning_rate": 7.66566265060241e-06,
      "loss": 1.0403,
      "step": 155
    },
    {
      "epoch": 0.23493975903614459,
      "grad_norm": 0.1845703125,
      "learning_rate": 7.650602409638555e-06,
      "loss": 0.9851,
      "step": 156
    },
    {
      "epoch": 0.23644578313253012,
      "grad_norm": 0.1728515625,
      "learning_rate": 7.635542168674699e-06,
      "loss": 0.9942,
      "step": 157
    },
    {
      "epoch": 0.23795180722891565,
      "grad_norm": 0.205078125,
      "learning_rate": 7.620481927710845e-06,
      "loss": 1.0072,
      "step": 158
    },
    {
      "epoch": 0.2394578313253012,
      "grad_norm": 0.2421875,
      "learning_rate": 7.605421686746988e-06,
      "loss": 0.9674,
      "step": 159
    },
    {
      "epoch": 0.24096385542168675,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.590361445783133e-06,
      "loss": 1.0468,
      "step": 160
    },
    {
      "epoch": 0.24246987951807228,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.575301204819277e-06,
      "loss": 0.9991,
      "step": 161
    },
    {
      "epoch": 0.24397590361445784,
      "grad_norm": 0.16796875,
      "learning_rate": 7.560240963855422e-06,
      "loss": 0.9725,
      "step": 162
    },
    {
      "epoch": 0.24548192771084337,
      "grad_norm": 0.166015625,
      "learning_rate": 7.545180722891566e-06,
      "loss": 1.0086,
      "step": 163
    },
    {
      "epoch": 0.2469879518072289,
      "grad_norm": 0.203125,
      "learning_rate": 7.530120481927712e-06,
      "loss": 1.0109,
      "step": 164
    },
    {
      "epoch": 0.24849397590361447,
      "grad_norm": 0.173828125,
      "learning_rate": 7.515060240963857e-06,
      "loss": 1.0107,
      "step": 165
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.0202,
      "step": 166
    },
    {
      "epoch": 0.25150602409638556,
      "grad_norm": 0.2021484375,
      "learning_rate": 7.484939759036145e-06,
      "loss": 1.0121,
      "step": 167
    },
    {
      "epoch": 0.25301204819277107,
      "grad_norm": 0.177734375,
      "learning_rate": 7.469879518072289e-06,
      "loss": 1.0056,
      "step": 168
    },
    {
      "epoch": 0.2545180722891566,
      "grad_norm": 0.1943359375,
      "learning_rate": 7.454819277108435e-06,
      "loss": 0.9952,
      "step": 169
    },
    {
      "epoch": 0.2560240963855422,
      "grad_norm": 0.2080078125,
      "learning_rate": 7.43975903614458e-06,
      "loss": 1.015,
      "step": 170
    },
    {
      "epoch": 0.2575301204819277,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.424698795180724e-06,
      "loss": 1.0114,
      "step": 171
    },
    {
      "epoch": 0.25903614457831325,
      "grad_norm": 0.1767578125,
      "learning_rate": 7.4096385542168684e-06,
      "loss": 0.9867,
      "step": 172
    },
    {
      "epoch": 0.2605421686746988,
      "grad_norm": 0.173828125,
      "learning_rate": 7.394578313253013e-06,
      "loss": 1.0081,
      "step": 173
    },
    {
      "epoch": 0.2620481927710843,
      "grad_norm": 0.21484375,
      "learning_rate": 7.3795180722891564e-06,
      "loss": 1.022,
      "step": 174
    },
    {
      "epoch": 0.2635542168674699,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.3644578313253026e-06,
      "loss": 1.0069,
      "step": 175
    },
    {
      "epoch": 0.26506024096385544,
      "grad_norm": 0.1640625,
      "learning_rate": 7.349397590361447e-06,
      "loss": 0.9935,
      "step": 176
    },
    {
      "epoch": 0.26656626506024095,
      "grad_norm": 0.494140625,
      "learning_rate": 7.334337349397591e-06,
      "loss": 0.9877,
      "step": 177
    },
    {
      "epoch": 0.2680722891566265,
      "grad_norm": 0.1630859375,
      "learning_rate": 7.319277108433736e-06,
      "loss": 0.9537,
      "step": 178
    },
    {
      "epoch": 0.26957831325301207,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.30421686746988e-06,
      "loss": 0.9627,
      "step": 179
    },
    {
      "epoch": 0.2710843373493976,
      "grad_norm": 0.1884765625,
      "learning_rate": 7.289156626506025e-06,
      "loss": 1.0223,
      "step": 180
    },
    {
      "epoch": 0.27259036144578314,
      "grad_norm": 0.23828125,
      "learning_rate": 7.27409638554217e-06,
      "loss": 0.9608,
      "step": 181
    },
    {
      "epoch": 0.2740963855421687,
      "grad_norm": 0.181640625,
      "learning_rate": 7.259036144578314e-06,
      "loss": 0.9585,
      "step": 182
    },
    {
      "epoch": 0.2756024096385542,
      "grad_norm": 0.1650390625,
      "learning_rate": 7.243975903614459e-06,
      "loss": 0.9503,
      "step": 183
    },
    {
      "epoch": 0.27710843373493976,
      "grad_norm": 0.166015625,
      "learning_rate": 7.228915662650603e-06,
      "loss": 0.9667,
      "step": 184
    },
    {
      "epoch": 0.2786144578313253,
      "grad_norm": 0.1787109375,
      "learning_rate": 7.213855421686748e-06,
      "loss": 0.9525,
      "step": 185
    },
    {
      "epoch": 0.28012048192771083,
      "grad_norm": 0.169921875,
      "learning_rate": 7.198795180722892e-06,
      "loss": 1.0228,
      "step": 186
    },
    {
      "epoch": 0.2816265060240964,
      "grad_norm": 0.19140625,
      "learning_rate": 7.183734939759037e-06,
      "loss": 1.0047,
      "step": 187
    },
    {
      "epoch": 0.28313253012048195,
      "grad_norm": 0.1669921875,
      "learning_rate": 7.168674698795182e-06,
      "loss": 0.9903,
      "step": 188
    },
    {
      "epoch": 0.28463855421686746,
      "grad_norm": 0.2099609375,
      "learning_rate": 7.153614457831326e-06,
      "loss": 0.984,
      "step": 189
    },
    {
      "epoch": 0.286144578313253,
      "grad_norm": 0.19921875,
      "learning_rate": 7.138554216867471e-06,
      "loss": 0.9856,
      "step": 190
    },
    {
      "epoch": 0.2876506024096386,
      "grad_norm": 0.1640625,
      "learning_rate": 7.123493975903615e-06,
      "loss": 0.9679,
      "step": 191
    },
    {
      "epoch": 0.2891566265060241,
      "grad_norm": 0.2255859375,
      "learning_rate": 7.1084337349397595e-06,
      "loss": 1.0065,
      "step": 192
    },
    {
      "epoch": 0.29066265060240964,
      "grad_norm": 0.1630859375,
      "learning_rate": 7.093373493975905e-06,
      "loss": 0.9611,
      "step": 193
    },
    {
      "epoch": 0.2921686746987952,
      "grad_norm": 0.18359375,
      "learning_rate": 7.078313253012049e-06,
      "loss": 1.0034,
      "step": 194
    },
    {
      "epoch": 0.2936746987951807,
      "grad_norm": 0.173828125,
      "learning_rate": 7.063253012048194e-06,
      "loss": 0.9546,
      "step": 195
    },
    {
      "epoch": 0.29518072289156627,
      "grad_norm": 0.162109375,
      "learning_rate": 7.048192771084338e-06,
      "loss": 0.9554,
      "step": 196
    },
    {
      "epoch": 0.29668674698795183,
      "grad_norm": 0.166015625,
      "learning_rate": 7.0331325301204825e-06,
      "loss": 0.9661,
      "step": 197
    },
    {
      "epoch": 0.29819277108433734,
      "grad_norm": 0.197265625,
      "learning_rate": 7.018072289156627e-06,
      "loss": 0.9337,
      "step": 198
    },
    {
      "epoch": 0.2996987951807229,
      "grad_norm": 0.1806640625,
      "learning_rate": 7.003012048192772e-06,
      "loss": 1.0222,
      "step": 199
    },
    {
      "epoch": 0.30120481927710846,
      "grad_norm": 0.1669921875,
      "learning_rate": 6.987951807228917e-06,
      "loss": 0.9915,
      "step": 200
    },
    {
      "epoch": 0.30271084337349397,
      "grad_norm": 0.25390625,
      "learning_rate": 6.972891566265061e-06,
      "loss": 0.9982,
      "step": 201
    },
    {
      "epoch": 0.3042168674698795,
      "grad_norm": 0.1787109375,
      "learning_rate": 6.9578313253012055e-06,
      "loss": 0.9811,
      "step": 202
    },
    {
      "epoch": 0.3057228915662651,
      "grad_norm": 0.1796875,
      "learning_rate": 6.94277108433735e-06,
      "loss": 0.9484,
      "step": 203
    },
    {
      "epoch": 0.3072289156626506,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.927710843373494e-06,
      "loss": 0.9968,
      "step": 204
    },
    {
      "epoch": 0.30873493975903615,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.91265060240964e-06,
      "loss": 0.9962,
      "step": 205
    },
    {
      "epoch": 0.3102409638554217,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.897590361445784e-06,
      "loss": 1.0162,
      "step": 206
    },
    {
      "epoch": 0.3117469879518072,
      "grad_norm": 0.21484375,
      "learning_rate": 6.8825301204819285e-06,
      "loss": 0.9987,
      "step": 207
    },
    {
      "epoch": 0.3132530120481928,
      "grad_norm": 0.1767578125,
      "learning_rate": 6.867469879518073e-06,
      "loss": 0.9803,
      "step": 208
    },
    {
      "epoch": 0.3147590361445783,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.852409638554217e-06,
      "loss": 0.9722,
      "step": 209
    },
    {
      "epoch": 0.31626506024096385,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.837349397590362e-06,
      "loss": 0.9755,
      "step": 210
    },
    {
      "epoch": 0.3177710843373494,
      "grad_norm": 0.173828125,
      "learning_rate": 6.822289156626507e-06,
      "loss": 0.9559,
      "step": 211
    },
    {
      "epoch": 0.3192771084337349,
      "grad_norm": 0.1728515625,
      "learning_rate": 6.8072289156626514e-06,
      "loss": 0.9503,
      "step": 212
    },
    {
      "epoch": 0.3207831325301205,
      "grad_norm": 0.181640625,
      "learning_rate": 6.792168674698796e-06,
      "loss": 0.991,
      "step": 213
    },
    {
      "epoch": 0.32228915662650603,
      "grad_norm": 0.171875,
      "learning_rate": 6.77710843373494e-06,
      "loss": 0.9722,
      "step": 214
    },
    {
      "epoch": 0.32379518072289154,
      "grad_norm": 0.220703125,
      "learning_rate": 6.762048192771085e-06,
      "loss": 1.0076,
      "step": 215
    },
    {
      "epoch": 0.3253012048192771,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.746987951807229e-06,
      "loss": 0.9693,
      "step": 216
    },
    {
      "epoch": 0.32680722891566266,
      "grad_norm": 0.193359375,
      "learning_rate": 6.731927710843374e-06,
      "loss": 0.9547,
      "step": 217
    },
    {
      "epoch": 0.32831325301204817,
      "grad_norm": 0.1669921875,
      "learning_rate": 6.716867469879519e-06,
      "loss": 0.9604,
      "step": 218
    },
    {
      "epoch": 0.32981927710843373,
      "grad_norm": 0.185546875,
      "learning_rate": 6.701807228915663e-06,
      "loss": 0.9602,
      "step": 219
    },
    {
      "epoch": 0.3313253012048193,
      "grad_norm": 0.166015625,
      "learning_rate": 6.686746987951808e-06,
      "loss": 0.9483,
      "step": 220
    },
    {
      "epoch": 0.3328313253012048,
      "grad_norm": 0.18359375,
      "learning_rate": 6.671686746987952e-06,
      "loss": 0.9506,
      "step": 221
    },
    {
      "epoch": 0.33433734939759036,
      "grad_norm": 0.181640625,
      "learning_rate": 6.6566265060240965e-06,
      "loss": 0.9832,
      "step": 222
    },
    {
      "epoch": 0.3358433734939759,
      "grad_norm": 0.1865234375,
      "learning_rate": 6.641566265060242e-06,
      "loss": 0.9915,
      "step": 223
    },
    {
      "epoch": 0.3373493975903614,
      "grad_norm": 0.1875,
      "learning_rate": 6.626506024096386e-06,
      "loss": 0.9779,
      "step": 224
    },
    {
      "epoch": 0.338855421686747,
      "grad_norm": 0.169921875,
      "learning_rate": 6.611445783132531e-06,
      "loss": 0.9648,
      "step": 225
    },
    {
      "epoch": 0.34036144578313254,
      "grad_norm": 0.1640625,
      "learning_rate": 6.596385542168675e-06,
      "loss": 0.9737,
      "step": 226
    },
    {
      "epoch": 0.34186746987951805,
      "grad_norm": 0.58203125,
      "learning_rate": 6.5813253012048195e-06,
      "loss": 0.9538,
      "step": 227
    },
    {
      "epoch": 0.3433734939759036,
      "grad_norm": 0.181640625,
      "learning_rate": 6.566265060240964e-06,
      "loss": 0.9875,
      "step": 228
    },
    {
      "epoch": 0.34487951807228917,
      "grad_norm": 0.255859375,
      "learning_rate": 6.551204819277109e-06,
      "loss": 0.9986,
      "step": 229
    },
    {
      "epoch": 0.3463855421686747,
      "grad_norm": 0.251953125,
      "learning_rate": 6.536144578313254e-06,
      "loss": 0.9283,
      "step": 230
    },
    {
      "epoch": 0.34789156626506024,
      "grad_norm": 0.203125,
      "learning_rate": 6.521084337349398e-06,
      "loss": 0.9372,
      "step": 231
    },
    {
      "epoch": 0.3493975903614458,
      "grad_norm": 0.173828125,
      "learning_rate": 6.5060240963855425e-06,
      "loss": 0.9665,
      "step": 232
    },
    {
      "epoch": 0.3509036144578313,
      "grad_norm": 0.169921875,
      "learning_rate": 6.490963855421687e-06,
      "loss": 0.9761,
      "step": 233
    },
    {
      "epoch": 0.35240963855421686,
      "grad_norm": 0.2470703125,
      "learning_rate": 6.475903614457831e-06,
      "loss": 0.9516,
      "step": 234
    },
    {
      "epoch": 0.3539156626506024,
      "grad_norm": 0.212890625,
      "learning_rate": 6.460843373493977e-06,
      "loss": 0.9346,
      "step": 235
    },
    {
      "epoch": 0.35542168674698793,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.445783132530121e-06,
      "loss": 0.9664,
      "step": 236
    },
    {
      "epoch": 0.3569277108433735,
      "grad_norm": 0.1904296875,
      "learning_rate": 6.4307228915662655e-06,
      "loss": 0.9432,
      "step": 237
    },
    {
      "epoch": 0.35843373493975905,
      "grad_norm": 0.2197265625,
      "learning_rate": 6.41566265060241e-06,
      "loss": 0.991,
      "step": 238
    },
    {
      "epoch": 0.35993975903614456,
      "grad_norm": 0.193359375,
      "learning_rate": 6.400602409638554e-06,
      "loss": 0.9323,
      "step": 239
    },
    {
      "epoch": 0.3614457831325301,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.385542168674699e-06,
      "loss": 0.909,
      "step": 240
    },
    {
      "epoch": 0.3629518072289157,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.370481927710844e-06,
      "loss": 0.9824,
      "step": 241
    },
    {
      "epoch": 0.3644578313253012,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.3554216867469885e-06,
      "loss": 0.971,
      "step": 242
    },
    {
      "epoch": 0.36596385542168675,
      "grad_norm": 0.2412109375,
      "learning_rate": 6.340361445783133e-06,
      "loss": 0.9554,
      "step": 243
    },
    {
      "epoch": 0.3674698795180723,
      "grad_norm": 0.1787109375,
      "learning_rate": 6.325301204819277e-06,
      "loss": 0.9393,
      "step": 244
    },
    {
      "epoch": 0.3689759036144578,
      "grad_norm": 0.201171875,
      "learning_rate": 6.310240963855422e-06,
      "loss": 0.9804,
      "step": 245
    },
    {
      "epoch": 0.3704819277108434,
      "grad_norm": 0.171875,
      "learning_rate": 6.295180722891566e-06,
      "loss": 0.965,
      "step": 246
    },
    {
      "epoch": 0.37198795180722893,
      "grad_norm": 0.2021484375,
      "learning_rate": 6.2801204819277114e-06,
      "loss": 0.9791,
      "step": 247
    },
    {
      "epoch": 0.37349397590361444,
      "grad_norm": 0.203125,
      "learning_rate": 6.265060240963856e-06,
      "loss": 0.9952,
      "step": 248
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.169921875,
      "learning_rate": 6.25e-06,
      "loss": 0.9708,
      "step": 249
    },
    {
      "epoch": 0.37650602409638556,
      "grad_norm": 0.181640625,
      "learning_rate": 6.234939759036145e-06,
      "loss": 0.9767,
      "step": 250
    },
    {
      "epoch": 0.37801204819277107,
      "grad_norm": 0.1806640625,
      "learning_rate": 6.219879518072289e-06,
      "loss": 0.9669,
      "step": 251
    },
    {
      "epoch": 0.3795180722891566,
      "grad_norm": 0.185546875,
      "learning_rate": 6.2048192771084344e-06,
      "loss": 0.9758,
      "step": 252
    },
    {
      "epoch": 0.3810240963855422,
      "grad_norm": 0.2177734375,
      "learning_rate": 6.189759036144579e-06,
      "loss": 0.9886,
      "step": 253
    },
    {
      "epoch": 0.3825301204819277,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.174698795180723e-06,
      "loss": 0.9807,
      "step": 254
    },
    {
      "epoch": 0.38403614457831325,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.159638554216868e-06,
      "loss": 0.9563,
      "step": 255
    },
    {
      "epoch": 0.3855421686746988,
      "grad_norm": 0.189453125,
      "learning_rate": 6.144578313253012e-06,
      "loss": 0.9666,
      "step": 256
    },
    {
      "epoch": 0.3870481927710843,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.1295180722891566e-06,
      "loss": 0.9659,
      "step": 257
    },
    {
      "epoch": 0.3885542168674699,
      "grad_norm": 0.1845703125,
      "learning_rate": 6.114457831325302e-06,
      "loss": 0.9476,
      "step": 258
    },
    {
      "epoch": 0.39006024096385544,
      "grad_norm": 0.193359375,
      "learning_rate": 6.099397590361446e-06,
      "loss": 0.9249,
      "step": 259
    },
    {
      "epoch": 0.39156626506024095,
      "grad_norm": 0.1787109375,
      "learning_rate": 6.084337349397591e-06,
      "loss": 0.9693,
      "step": 260
    },
    {
      "epoch": 0.3930722891566265,
      "grad_norm": 0.1884765625,
      "learning_rate": 6.069277108433735e-06,
      "loss": 0.931,
      "step": 261
    },
    {
      "epoch": 0.39457831325301207,
      "grad_norm": 0.171875,
      "learning_rate": 6.0542168674698795e-06,
      "loss": 0.9398,
      "step": 262
    },
    {
      "epoch": 0.3960843373493976,
      "grad_norm": 0.1708984375,
      "learning_rate": 6.039156626506024e-06,
      "loss": 0.9396,
      "step": 263
    },
    {
      "epoch": 0.39759036144578314,
      "grad_norm": 0.1875,
      "learning_rate": 6.02409638554217e-06,
      "loss": 0.9187,
      "step": 264
    },
    {
      "epoch": 0.3990963855421687,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.009036144578314e-06,
      "loss": 0.9473,
      "step": 265
    },
    {
      "epoch": 0.4006024096385542,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.993975903614458e-06,
      "loss": 0.9415,
      "step": 266
    },
    {
      "epoch": 0.40210843373493976,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.9789156626506025e-06,
      "loss": 0.9534,
      "step": 267
    },
    {
      "epoch": 0.4036144578313253,
      "grad_norm": 0.1845703125,
      "learning_rate": 5.963855421686747e-06,
      "loss": 0.9496,
      "step": 268
    },
    {
      "epoch": 0.40512048192771083,
      "grad_norm": 0.2890625,
      "learning_rate": 5.948795180722891e-06,
      "loss": 1.0058,
      "step": 269
    },
    {
      "epoch": 0.4066265060240964,
      "grad_norm": 0.228515625,
      "learning_rate": 5.9337349397590375e-06,
      "loss": 0.952,
      "step": 270
    },
    {
      "epoch": 0.40813253012048195,
      "grad_norm": 0.1708984375,
      "learning_rate": 5.918674698795182e-06,
      "loss": 0.9671,
      "step": 271
    },
    {
      "epoch": 0.40963855421686746,
      "grad_norm": 0.181640625,
      "learning_rate": 5.9036144578313255e-06,
      "loss": 0.9629,
      "step": 272
    },
    {
      "epoch": 0.411144578313253,
      "grad_norm": 0.177734375,
      "learning_rate": 5.88855421686747e-06,
      "loss": 0.9173,
      "step": 273
    },
    {
      "epoch": 0.4126506024096386,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.873493975903614e-06,
      "loss": 0.9529,
      "step": 274
    },
    {
      "epoch": 0.4141566265060241,
      "grad_norm": 0.1767578125,
      "learning_rate": 5.858433734939759e-06,
      "loss": 0.9627,
      "step": 275
    },
    {
      "epoch": 0.41566265060240964,
      "grad_norm": 0.173828125,
      "learning_rate": 5.843373493975905e-06,
      "loss": 0.9265,
      "step": 276
    },
    {
      "epoch": 0.4171686746987952,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.828313253012049e-06,
      "loss": 0.9488,
      "step": 277
    },
    {
      "epoch": 0.4186746987951807,
      "grad_norm": 0.205078125,
      "learning_rate": 5.813253012048194e-06,
      "loss": 0.951,
      "step": 278
    },
    {
      "epoch": 0.42018072289156627,
      "grad_norm": 0.17578125,
      "learning_rate": 5.798192771084338e-06,
      "loss": 0.9312,
      "step": 279
    },
    {
      "epoch": 0.42168674698795183,
      "grad_norm": 0.17578125,
      "learning_rate": 5.783132530120482e-06,
      "loss": 0.9475,
      "step": 280
    },
    {
      "epoch": 0.42319277108433734,
      "grad_norm": 0.2080078125,
      "learning_rate": 5.768072289156626e-06,
      "loss": 0.9655,
      "step": 281
    },
    {
      "epoch": 0.4246987951807229,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.753012048192772e-06,
      "loss": 0.9785,
      "step": 282
    },
    {
      "epoch": 0.42620481927710846,
      "grad_norm": 0.21484375,
      "learning_rate": 5.737951807228917e-06,
      "loss": 0.9595,
      "step": 283
    },
    {
      "epoch": 0.42771084337349397,
      "grad_norm": 0.1708984375,
      "learning_rate": 5.722891566265061e-06,
      "loss": 0.9427,
      "step": 284
    },
    {
      "epoch": 0.4292168674698795,
      "grad_norm": 0.212890625,
      "learning_rate": 5.707831325301206e-06,
      "loss": 0.9488,
      "step": 285
    },
    {
      "epoch": 0.4307228915662651,
      "grad_norm": 0.220703125,
      "learning_rate": 5.69277108433735e-06,
      "loss": 0.999,
      "step": 286
    },
    {
      "epoch": 0.4322289156626506,
      "grad_norm": 0.19140625,
      "learning_rate": 5.677710843373494e-06,
      "loss": 0.9489,
      "step": 287
    },
    {
      "epoch": 0.43373493975903615,
      "grad_norm": 0.1796875,
      "learning_rate": 5.66265060240964e-06,
      "loss": 0.9612,
      "step": 288
    },
    {
      "epoch": 0.4352409638554217,
      "grad_norm": 0.1787109375,
      "learning_rate": 5.647590361445784e-06,
      "loss": 0.92,
      "step": 289
    },
    {
      "epoch": 0.4367469879518072,
      "grad_norm": 0.1767578125,
      "learning_rate": 5.6325301204819286e-06,
      "loss": 0.9404,
      "step": 290
    },
    {
      "epoch": 0.4382530120481928,
      "grad_norm": 0.2236328125,
      "learning_rate": 5.617469879518073e-06,
      "loss": 0.9403,
      "step": 291
    },
    {
      "epoch": 0.4397590361445783,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.602409638554217e-06,
      "loss": 0.9067,
      "step": 292
    },
    {
      "epoch": 0.44126506024096385,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.587349397590362e-06,
      "loss": 0.9549,
      "step": 293
    },
    {
      "epoch": 0.4427710843373494,
      "grad_norm": 0.1875,
      "learning_rate": 5.572289156626507e-06,
      "loss": 0.9649,
      "step": 294
    },
    {
      "epoch": 0.4442771084337349,
      "grad_norm": 0.189453125,
      "learning_rate": 5.5572289156626515e-06,
      "loss": 0.9293,
      "step": 295
    },
    {
      "epoch": 0.4457831325301205,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.542168674698796e-06,
      "loss": 0.9349,
      "step": 296
    },
    {
      "epoch": 0.44728915662650603,
      "grad_norm": 0.181640625,
      "learning_rate": 5.52710843373494e-06,
      "loss": 1.0045,
      "step": 297
    },
    {
      "epoch": 0.44879518072289154,
      "grad_norm": 0.1904296875,
      "learning_rate": 5.512048192771085e-06,
      "loss": 0.9549,
      "step": 298
    },
    {
      "epoch": 0.4503012048192771,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.496987951807229e-06,
      "loss": 0.8953,
      "step": 299
    },
    {
      "epoch": 0.45180722891566266,
      "grad_norm": 0.1904296875,
      "learning_rate": 5.4819277108433745e-06,
      "loss": 0.9432,
      "step": 300
    },
    {
      "epoch": 0.45331325301204817,
      "grad_norm": 0.1796875,
      "learning_rate": 5.466867469879519e-06,
      "loss": 0.9425,
      "step": 301
    },
    {
      "epoch": 0.45481927710843373,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.451807228915663e-06,
      "loss": 0.945,
      "step": 302
    },
    {
      "epoch": 0.4563253012048193,
      "grad_norm": 0.173828125,
      "learning_rate": 5.436746987951808e-06,
      "loss": 0.9201,
      "step": 303
    },
    {
      "epoch": 0.4578313253012048,
      "grad_norm": 0.17578125,
      "learning_rate": 5.421686746987952e-06,
      "loss": 0.9662,
      "step": 304
    },
    {
      "epoch": 0.45933734939759036,
      "grad_norm": 0.2080078125,
      "learning_rate": 5.406626506024097e-06,
      "loss": 0.913,
      "step": 305
    },
    {
      "epoch": 0.4608433734939759,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.391566265060242e-06,
      "loss": 0.9598,
      "step": 306
    },
    {
      "epoch": 0.4623493975903614,
      "grad_norm": 0.2294921875,
      "learning_rate": 5.376506024096386e-06,
      "loss": 0.9247,
      "step": 307
    },
    {
      "epoch": 0.463855421686747,
      "grad_norm": 0.1796875,
      "learning_rate": 5.361445783132531e-06,
      "loss": 0.9741,
      "step": 308
    },
    {
      "epoch": 0.46536144578313254,
      "grad_norm": 0.287109375,
      "learning_rate": 5.346385542168675e-06,
      "loss": 0.9473,
      "step": 309
    },
    {
      "epoch": 0.46686746987951805,
      "grad_norm": 0.271484375,
      "learning_rate": 5.33132530120482e-06,
      "loss": 0.8972,
      "step": 310
    },
    {
      "epoch": 0.4683734939759036,
      "grad_norm": 0.17578125,
      "learning_rate": 5.316265060240964e-06,
      "loss": 0.9222,
      "step": 311
    },
    {
      "epoch": 0.46987951807228917,
      "grad_norm": 0.263671875,
      "learning_rate": 5.301204819277109e-06,
      "loss": 0.9161,
      "step": 312
    },
    {
      "epoch": 0.4713855421686747,
      "grad_norm": 0.1884765625,
      "learning_rate": 5.286144578313254e-06,
      "loss": 0.9208,
      "step": 313
    },
    {
      "epoch": 0.47289156626506024,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.271084337349398e-06,
      "loss": 0.9409,
      "step": 314
    },
    {
      "epoch": 0.4743975903614458,
      "grad_norm": 0.201171875,
      "learning_rate": 5.256024096385543e-06,
      "loss": 0.9424,
      "step": 315
    },
    {
      "epoch": 0.4759036144578313,
      "grad_norm": 0.1904296875,
      "learning_rate": 5.240963855421687e-06,
      "loss": 0.937,
      "step": 316
    },
    {
      "epoch": 0.47740963855421686,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.2259036144578315e-06,
      "loss": 0.929,
      "step": 317
    },
    {
      "epoch": 0.4789156626506024,
      "grad_norm": 0.18359375,
      "learning_rate": 5.210843373493977e-06,
      "loss": 0.9338,
      "step": 318
    },
    {
      "epoch": 0.48042168674698793,
      "grad_norm": 0.201171875,
      "learning_rate": 5.195783132530121e-06,
      "loss": 0.9515,
      "step": 319
    },
    {
      "epoch": 0.4819277108433735,
      "grad_norm": 0.1884765625,
      "learning_rate": 5.180722891566266e-06,
      "loss": 0.9422,
      "step": 320
    },
    {
      "epoch": 0.48343373493975905,
      "grad_norm": 0.177734375,
      "learning_rate": 5.16566265060241e-06,
      "loss": 0.9374,
      "step": 321
    },
    {
      "epoch": 0.48493975903614456,
      "grad_norm": 0.17578125,
      "learning_rate": 5.1506024096385544e-06,
      "loss": 0.9635,
      "step": 322
    },
    {
      "epoch": 0.4864457831325301,
      "grad_norm": 0.2451171875,
      "learning_rate": 5.135542168674699e-06,
      "loss": 0.9418,
      "step": 323
    },
    {
      "epoch": 0.4879518072289157,
      "grad_norm": 0.208984375,
      "learning_rate": 5.120481927710844e-06,
      "loss": 0.9157,
      "step": 324
    },
    {
      "epoch": 0.4894578313253012,
      "grad_norm": 0.18359375,
      "learning_rate": 5.1054216867469886e-06,
      "loss": 0.9607,
      "step": 325
    },
    {
      "epoch": 0.49096385542168675,
      "grad_norm": 0.1728515625,
      "learning_rate": 5.090361445783133e-06,
      "loss": 0.8981,
      "step": 326
    },
    {
      "epoch": 0.4924698795180723,
      "grad_norm": 0.2275390625,
      "learning_rate": 5.0753012048192774e-06,
      "loss": 0.9316,
      "step": 327
    },
    {
      "epoch": 0.4939759036144578,
      "grad_norm": 0.205078125,
      "learning_rate": 5.060240963855422e-06,
      "loss": 0.9304,
      "step": 328
    },
    {
      "epoch": 0.4954819277108434,
      "grad_norm": 0.2197265625,
      "learning_rate": 5.045180722891566e-06,
      "loss": 0.9158,
      "step": 329
    },
    {
      "epoch": 0.49698795180722893,
      "grad_norm": 0.1796875,
      "learning_rate": 5.0301204819277116e-06,
      "loss": 0.8974,
      "step": 330
    },
    {
      "epoch": 0.49849397590361444,
      "grad_norm": 0.21484375,
      "learning_rate": 5.015060240963856e-06,
      "loss": 0.9657,
      "step": 331
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.205078125,
      "learning_rate": 5e-06,
      "loss": 0.8869,
      "step": 332
    },
    {
      "epoch": 0.5015060240963856,
      "grad_norm": 0.177734375,
      "learning_rate": 4.984939759036145e-06,
      "loss": 0.9311,
      "step": 333
    },
    {
      "epoch": 0.5030120481927711,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.96987951807229e-06,
      "loss": 0.9436,
      "step": 334
    },
    {
      "epoch": 0.5045180722891566,
      "grad_norm": 0.2177734375,
      "learning_rate": 4.9548192771084345e-06,
      "loss": 0.9312,
      "step": 335
    },
    {
      "epoch": 0.5060240963855421,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.939759036144578e-06,
      "loss": 0.9825,
      "step": 336
    },
    {
      "epoch": 0.5075301204819277,
      "grad_norm": 0.25390625,
      "learning_rate": 4.924698795180723e-06,
      "loss": 0.9471,
      "step": 337
    },
    {
      "epoch": 0.5090361445783133,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.909638554216868e-06,
      "loss": 0.9287,
      "step": 338
    },
    {
      "epoch": 0.5105421686746988,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.894578313253012e-06,
      "loss": 0.9518,
      "step": 339
    },
    {
      "epoch": 0.5120481927710844,
      "grad_norm": 0.25390625,
      "learning_rate": 4.8795180722891575e-06,
      "loss": 0.9394,
      "step": 340
    },
    {
      "epoch": 0.5135542168674698,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.864457831325302e-06,
      "loss": 0.9214,
      "step": 341
    },
    {
      "epoch": 0.5150602409638554,
      "grad_norm": 0.2451171875,
      "learning_rate": 4.849397590361446e-06,
      "loss": 0.939,
      "step": 342
    },
    {
      "epoch": 0.516566265060241,
      "grad_norm": 0.240234375,
      "learning_rate": 4.834337349397591e-06,
      "loss": 0.9632,
      "step": 343
    },
    {
      "epoch": 0.5180722891566265,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.819277108433735e-06,
      "loss": 0.9567,
      "step": 344
    },
    {
      "epoch": 0.5195783132530121,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.80421686746988e-06,
      "loss": 0.8776,
      "step": 345
    },
    {
      "epoch": 0.5210843373493976,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.789156626506025e-06,
      "loss": 0.955,
      "step": 346
    },
    {
      "epoch": 0.5225903614457831,
      "grad_norm": 0.1875,
      "learning_rate": 4.774096385542169e-06,
      "loss": 0.8901,
      "step": 347
    },
    {
      "epoch": 0.5240963855421686,
      "grad_norm": 0.2109375,
      "learning_rate": 4.759036144578314e-06,
      "loss": 0.9396,
      "step": 348
    },
    {
      "epoch": 0.5256024096385542,
      "grad_norm": 0.1796875,
      "learning_rate": 4.743975903614458e-06,
      "loss": 0.9865,
      "step": 349
    },
    {
      "epoch": 0.5271084337349398,
      "grad_norm": 1.3984375,
      "learning_rate": 4.728915662650603e-06,
      "loss": 0.8973,
      "step": 350
    },
    {
      "epoch": 0.5286144578313253,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.713855421686747e-06,
      "loss": 0.9137,
      "step": 351
    },
    {
      "epoch": 0.5301204819277109,
      "grad_norm": 0.267578125,
      "learning_rate": 4.698795180722892e-06,
      "loss": 0.9022,
      "step": 352
    },
    {
      "epoch": 0.5316265060240963,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.683734939759037e-06,
      "loss": 0.9201,
      "step": 353
    },
    {
      "epoch": 0.5331325301204819,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.668674698795181e-06,
      "loss": 0.963,
      "step": 354
    },
    {
      "epoch": 0.5346385542168675,
      "grad_norm": 0.2490234375,
      "learning_rate": 4.653614457831326e-06,
      "loss": 0.9494,
      "step": 355
    },
    {
      "epoch": 0.536144578313253,
      "grad_norm": 0.189453125,
      "learning_rate": 4.63855421686747e-06,
      "loss": 0.9957,
      "step": 356
    },
    {
      "epoch": 0.5376506024096386,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.6234939759036145e-06,
      "loss": 0.9585,
      "step": 357
    },
    {
      "epoch": 0.5391566265060241,
      "grad_norm": 0.236328125,
      "learning_rate": 4.60843373493976e-06,
      "loss": 0.929,
      "step": 358
    },
    {
      "epoch": 0.5406626506024096,
      "grad_norm": 0.1875,
      "learning_rate": 4.593373493975904e-06,
      "loss": 0.906,
      "step": 359
    },
    {
      "epoch": 0.5421686746987951,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.578313253012049e-06,
      "loss": 0.9118,
      "step": 360
    },
    {
      "epoch": 0.5436746987951807,
      "grad_norm": 0.193359375,
      "learning_rate": 4.563253012048193e-06,
      "loss": 0.8907,
      "step": 361
    },
    {
      "epoch": 0.5451807228915663,
      "grad_norm": 0.189453125,
      "learning_rate": 4.5481927710843374e-06,
      "loss": 0.9147,
      "step": 362
    },
    {
      "epoch": 0.5466867469879518,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.533132530120482e-06,
      "loss": 0.9297,
      "step": 363
    },
    {
      "epoch": 0.5481927710843374,
      "grad_norm": 0.18359375,
      "learning_rate": 4.518072289156627e-06,
      "loss": 0.9228,
      "step": 364
    },
    {
      "epoch": 0.5496987951807228,
      "grad_norm": 0.2314453125,
      "learning_rate": 4.5030120481927716e-06,
      "loss": 0.9202,
      "step": 365
    },
    {
      "epoch": 0.5512048192771084,
      "grad_norm": 0.19140625,
      "learning_rate": 4.487951807228916e-06,
      "loss": 0.9678,
      "step": 366
    },
    {
      "epoch": 0.552710843373494,
      "grad_norm": 0.19140625,
      "learning_rate": 4.47289156626506e-06,
      "loss": 0.9576,
      "step": 367
    },
    {
      "epoch": 0.5542168674698795,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.457831325301205e-06,
      "loss": 0.965,
      "step": 368
    },
    {
      "epoch": 0.5557228915662651,
      "grad_norm": 0.189453125,
      "learning_rate": 4.442771084337349e-06,
      "loss": 0.9212,
      "step": 369
    },
    {
      "epoch": 0.5572289156626506,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.4277108433734945e-06,
      "loss": 0.9491,
      "step": 370
    },
    {
      "epoch": 0.5587349397590361,
      "grad_norm": 0.189453125,
      "learning_rate": 4.412650602409639e-06,
      "loss": 0.9109,
      "step": 371
    },
    {
      "epoch": 0.5602409638554217,
      "grad_norm": 0.20703125,
      "learning_rate": 4.397590361445783e-06,
      "loss": 0.9277,
      "step": 372
    },
    {
      "epoch": 0.5617469879518072,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.382530120481928e-06,
      "loss": 0.9133,
      "step": 373
    },
    {
      "epoch": 0.5632530120481928,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.367469879518072e-06,
      "loss": 0.9475,
      "step": 374
    },
    {
      "epoch": 0.5647590361445783,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.3524096385542175e-06,
      "loss": 0.9659,
      "step": 375
    },
    {
      "epoch": 0.5662650602409639,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.337349397590362e-06,
      "loss": 0.9767,
      "step": 376
    },
    {
      "epoch": 0.5677710843373494,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.322289156626506e-06,
      "loss": 0.945,
      "step": 377
    },
    {
      "epoch": 0.5692771084337349,
      "grad_norm": 0.25,
      "learning_rate": 4.307228915662651e-06,
      "loss": 0.9153,
      "step": 378
    },
    {
      "epoch": 0.5707831325301205,
      "grad_norm": 0.193359375,
      "learning_rate": 4.292168674698795e-06,
      "loss": 0.9344,
      "step": 379
    },
    {
      "epoch": 0.572289156626506,
      "grad_norm": 0.24609375,
      "learning_rate": 4.27710843373494e-06,
      "loss": 0.9527,
      "step": 380
    },
    {
      "epoch": 0.5737951807228916,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.262048192771085e-06,
      "loss": 0.9341,
      "step": 381
    },
    {
      "epoch": 0.5753012048192772,
      "grad_norm": 0.177734375,
      "learning_rate": 4.246987951807229e-06,
      "loss": 0.9088,
      "step": 382
    },
    {
      "epoch": 0.5768072289156626,
      "grad_norm": 0.25390625,
      "learning_rate": 4.231927710843374e-06,
      "loss": 0.9275,
      "step": 383
    },
    {
      "epoch": 0.5783132530120482,
      "grad_norm": 0.201171875,
      "learning_rate": 4.216867469879519e-06,
      "loss": 0.9012,
      "step": 384
    },
    {
      "epoch": 0.5798192771084337,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.201807228915663e-06,
      "loss": 0.9396,
      "step": 385
    },
    {
      "epoch": 0.5813253012048193,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.186746987951807e-06,
      "loss": 0.921,
      "step": 386
    },
    {
      "epoch": 0.5828313253012049,
      "grad_norm": 0.1796875,
      "learning_rate": 4.171686746987952e-06,
      "loss": 0.9099,
      "step": 387
    },
    {
      "epoch": 0.5843373493975904,
      "grad_norm": 0.1875,
      "learning_rate": 4.156626506024097e-06,
      "loss": 0.959,
      "step": 388
    },
    {
      "epoch": 0.5858433734939759,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.141566265060241e-06,
      "loss": 0.9326,
      "step": 389
    },
    {
      "epoch": 0.5873493975903614,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.1265060240963865e-06,
      "loss": 0.9209,
      "step": 390
    },
    {
      "epoch": 0.588855421686747,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.111445783132531e-06,
      "loss": 0.9525,
      "step": 391
    },
    {
      "epoch": 0.5903614457831325,
      "grad_norm": 0.181640625,
      "learning_rate": 4.096385542168675e-06,
      "loss": 0.9278,
      "step": 392
    },
    {
      "epoch": 0.5918674698795181,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.08132530120482e-06,
      "loss": 0.9258,
      "step": 393
    },
    {
      "epoch": 0.5933734939759037,
      "grad_norm": 0.314453125,
      "learning_rate": 4.066265060240964e-06,
      "loss": 0.9545,
      "step": 394
    },
    {
      "epoch": 0.5948795180722891,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.051204819277109e-06,
      "loss": 0.8999,
      "step": 395
    },
    {
      "epoch": 0.5963855421686747,
      "grad_norm": 0.203125,
      "learning_rate": 4.036144578313254e-06,
      "loss": 0.9248,
      "step": 396
    },
    {
      "epoch": 0.5978915662650602,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.021084337349398e-06,
      "loss": 0.8797,
      "step": 397
    },
    {
      "epoch": 0.5993975903614458,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.006024096385543e-06,
      "loss": 0.88,
      "step": 398
    },
    {
      "epoch": 0.6009036144578314,
      "grad_norm": 0.18359375,
      "learning_rate": 3.990963855421687e-06,
      "loss": 0.9659,
      "step": 399
    },
    {
      "epoch": 0.6024096385542169,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.975903614457832e-06,
      "loss": 0.9289,
      "step": 400
    },
    {
      "epoch": 0.6039156626506024,
      "grad_norm": 0.2080078125,
      "learning_rate": 3.960843373493976e-06,
      "loss": 0.9276,
      "step": 401
    },
    {
      "epoch": 0.6054216867469879,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.945783132530121e-06,
      "loss": 0.9747,
      "step": 402
    },
    {
      "epoch": 0.6069277108433735,
      "grad_norm": 0.2275390625,
      "learning_rate": 3.930722891566266e-06,
      "loss": 0.9463,
      "step": 403
    },
    {
      "epoch": 0.608433734939759,
      "grad_norm": 0.189453125,
      "learning_rate": 3.91566265060241e-06,
      "loss": 0.9286,
      "step": 404
    },
    {
      "epoch": 0.6099397590361446,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.9006024096385546e-06,
      "loss": 0.9218,
      "step": 405
    },
    {
      "epoch": 0.6114457831325302,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.885542168674699e-06,
      "loss": 0.9369,
      "step": 406
    },
    {
      "epoch": 0.6129518072289156,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.870481927710843e-06,
      "loss": 0.9215,
      "step": 407
    },
    {
      "epoch": 0.6144578313253012,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.855421686746989e-06,
      "loss": 0.9268,
      "step": 408
    },
    {
      "epoch": 0.6159638554216867,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.840361445783133e-06,
      "loss": 0.9379,
      "step": 409
    },
    {
      "epoch": 0.6174698795180723,
      "grad_norm": 0.197265625,
      "learning_rate": 3.8253012048192775e-06,
      "loss": 0.9246,
      "step": 410
    },
    {
      "epoch": 0.6189759036144579,
      "grad_norm": 0.220703125,
      "learning_rate": 3.8102409638554224e-06,
      "loss": 0.9164,
      "step": 411
    },
    {
      "epoch": 0.6204819277108434,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.7951807228915664e-06,
      "loss": 0.9299,
      "step": 412
    },
    {
      "epoch": 0.6219879518072289,
      "grad_norm": 0.193359375,
      "learning_rate": 3.780120481927711e-06,
      "loss": 0.9388,
      "step": 413
    },
    {
      "epoch": 0.6234939759036144,
      "grad_norm": 0.2412109375,
      "learning_rate": 3.765060240963856e-06,
      "loss": 0.9502,
      "step": 414
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.203125,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.9333,
      "step": 415
    },
    {
      "epoch": 0.6265060240963856,
      "grad_norm": 0.2265625,
      "learning_rate": 3.7349397590361445e-06,
      "loss": 0.9097,
      "step": 416
    },
    {
      "epoch": 0.6280120481927711,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.71987951807229e-06,
      "loss": 0.921,
      "step": 417
    },
    {
      "epoch": 0.6295180722891566,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.7048192771084342e-06,
      "loss": 0.9465,
      "step": 418
    },
    {
      "epoch": 0.6310240963855421,
      "grad_norm": 0.2333984375,
      "learning_rate": 3.6897590361445782e-06,
      "loss": 0.8774,
      "step": 419
    },
    {
      "epoch": 0.6325301204819277,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.6746987951807235e-06,
      "loss": 0.8849,
      "step": 420
    },
    {
      "epoch": 0.6340361445783133,
      "grad_norm": 0.205078125,
      "learning_rate": 3.659638554216868e-06,
      "loss": 0.9311,
      "step": 421
    },
    {
      "epoch": 0.6355421686746988,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.6445783132530124e-06,
      "loss": 0.933,
      "step": 422
    },
    {
      "epoch": 0.6370481927710844,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.629518072289157e-06,
      "loss": 0.9062,
      "step": 423
    },
    {
      "epoch": 0.6385542168674698,
      "grad_norm": 0.423828125,
      "learning_rate": 3.6144578313253016e-06,
      "loss": 0.9291,
      "step": 424
    },
    {
      "epoch": 0.6400602409638554,
      "grad_norm": 0.189453125,
      "learning_rate": 3.599397590361446e-06,
      "loss": 0.9463,
      "step": 425
    },
    {
      "epoch": 0.641566265060241,
      "grad_norm": 0.30078125,
      "learning_rate": 3.584337349397591e-06,
      "loss": 0.9736,
      "step": 426
    },
    {
      "epoch": 0.6430722891566265,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.5692771084337353e-06,
      "loss": 0.921,
      "step": 427
    },
    {
      "epoch": 0.6445783132530121,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.5542168674698798e-06,
      "loss": 0.9263,
      "step": 428
    },
    {
      "epoch": 0.6460843373493976,
      "grad_norm": 0.19140625,
      "learning_rate": 3.5391566265060246e-06,
      "loss": 0.9052,
      "step": 429
    },
    {
      "epoch": 0.6475903614457831,
      "grad_norm": 0.5,
      "learning_rate": 3.524096385542169e-06,
      "loss": 1.0386,
      "step": 430
    },
    {
      "epoch": 0.6490963855421686,
      "grad_norm": 0.185546875,
      "learning_rate": 3.5090361445783135e-06,
      "loss": 0.9168,
      "step": 431
    },
    {
      "epoch": 0.6506024096385542,
      "grad_norm": 0.1826171875,
      "learning_rate": 3.4939759036144583e-06,
      "loss": 0.9217,
      "step": 432
    },
    {
      "epoch": 0.6521084337349398,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.4789156626506027e-06,
      "loss": 0.93,
      "step": 433
    },
    {
      "epoch": 0.6536144578313253,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.463855421686747e-06,
      "loss": 0.9099,
      "step": 434
    },
    {
      "epoch": 0.6551204819277109,
      "grad_norm": 0.333984375,
      "learning_rate": 3.448795180722892e-06,
      "loss": 0.9585,
      "step": 435
    },
    {
      "epoch": 0.6566265060240963,
      "grad_norm": 0.197265625,
      "learning_rate": 3.4337349397590364e-06,
      "loss": 0.9295,
      "step": 436
    },
    {
      "epoch": 0.6581325301204819,
      "grad_norm": 0.255859375,
      "learning_rate": 3.418674698795181e-06,
      "loss": 0.8764,
      "step": 437
    },
    {
      "epoch": 0.6596385542168675,
      "grad_norm": 0.2265625,
      "learning_rate": 3.4036144578313257e-06,
      "loss": 0.9239,
      "step": 438
    },
    {
      "epoch": 0.661144578313253,
      "grad_norm": 0.189453125,
      "learning_rate": 3.38855421686747e-06,
      "loss": 0.9504,
      "step": 439
    },
    {
      "epoch": 0.6626506024096386,
      "grad_norm": 0.201171875,
      "learning_rate": 3.3734939759036146e-06,
      "loss": 0.9006,
      "step": 440
    },
    {
      "epoch": 0.6641566265060241,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.3584337349397594e-06,
      "loss": 0.9302,
      "step": 441
    },
    {
      "epoch": 0.6656626506024096,
      "grad_norm": 0.23046875,
      "learning_rate": 3.343373493975904e-06,
      "loss": 0.9068,
      "step": 442
    },
    {
      "epoch": 0.6671686746987951,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.3283132530120483e-06,
      "loss": 0.8961,
      "step": 443
    },
    {
      "epoch": 0.6686746987951807,
      "grad_norm": 0.19921875,
      "learning_rate": 3.313253012048193e-06,
      "loss": 0.9165,
      "step": 444
    },
    {
      "epoch": 0.6701807228915663,
      "grad_norm": 0.203125,
      "learning_rate": 3.2981927710843376e-06,
      "loss": 0.9184,
      "step": 445
    },
    {
      "epoch": 0.6716867469879518,
      "grad_norm": 0.216796875,
      "learning_rate": 3.283132530120482e-06,
      "loss": 0.9437,
      "step": 446
    },
    {
      "epoch": 0.6731927710843374,
      "grad_norm": 0.51171875,
      "learning_rate": 3.268072289156627e-06,
      "loss": 0.9562,
      "step": 447
    },
    {
      "epoch": 0.6746987951807228,
      "grad_norm": 0.18359375,
      "learning_rate": 3.2530120481927713e-06,
      "loss": 0.9525,
      "step": 448
    },
    {
      "epoch": 0.6762048192771084,
      "grad_norm": 0.19921875,
      "learning_rate": 3.2379518072289157e-06,
      "loss": 0.9659,
      "step": 449
    },
    {
      "epoch": 0.677710843373494,
      "grad_norm": 0.271484375,
      "learning_rate": 3.2228915662650605e-06,
      "loss": 1.0181,
      "step": 450
    },
    {
      "epoch": 0.6792168674698795,
      "grad_norm": 0.193359375,
      "learning_rate": 3.207831325301205e-06,
      "loss": 0.9153,
      "step": 451
    },
    {
      "epoch": 0.6807228915662651,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.1927710843373494e-06,
      "loss": 0.9514,
      "step": 452
    },
    {
      "epoch": 0.6822289156626506,
      "grad_norm": 0.19921875,
      "learning_rate": 3.1777108433734942e-06,
      "loss": 0.8957,
      "step": 453
    },
    {
      "epoch": 0.6837349397590361,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.1626506024096387e-06,
      "loss": 0.9267,
      "step": 454
    },
    {
      "epoch": 0.6852409638554217,
      "grad_norm": 0.2275390625,
      "learning_rate": 3.147590361445783e-06,
      "loss": 0.9551,
      "step": 455
    },
    {
      "epoch": 0.6867469879518072,
      "grad_norm": 0.28125,
      "learning_rate": 3.132530120481928e-06,
      "loss": 0.9308,
      "step": 456
    },
    {
      "epoch": 0.6882530120481928,
      "grad_norm": 0.18359375,
      "learning_rate": 3.1174698795180724e-06,
      "loss": 0.8992,
      "step": 457
    },
    {
      "epoch": 0.6897590361445783,
      "grad_norm": 0.29296875,
      "learning_rate": 3.1024096385542172e-06,
      "loss": 0.9173,
      "step": 458
    },
    {
      "epoch": 0.6912650602409639,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.0873493975903616e-06,
      "loss": 0.8848,
      "step": 459
    },
    {
      "epoch": 0.6927710843373494,
      "grad_norm": 0.1953125,
      "learning_rate": 3.072289156626506e-06,
      "loss": 0.9465,
      "step": 460
    },
    {
      "epoch": 0.6942771084337349,
      "grad_norm": 0.255859375,
      "learning_rate": 3.057228915662651e-06,
      "loss": 0.9308,
      "step": 461
    },
    {
      "epoch": 0.6957831325301205,
      "grad_norm": 0.1875,
      "learning_rate": 3.0421686746987953e-06,
      "loss": 0.9274,
      "step": 462
    },
    {
      "epoch": 0.697289156626506,
      "grad_norm": 0.2431640625,
      "learning_rate": 3.0271084337349398e-06,
      "loss": 0.9405,
      "step": 463
    },
    {
      "epoch": 0.6987951807228916,
      "grad_norm": 0.2392578125,
      "learning_rate": 3.012048192771085e-06,
      "loss": 0.9163,
      "step": 464
    },
    {
      "epoch": 0.7003012048192772,
      "grad_norm": 0.189453125,
      "learning_rate": 2.996987951807229e-06,
      "loss": 0.941,
      "step": 465
    },
    {
      "epoch": 0.7018072289156626,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.9819277108433735e-06,
      "loss": 0.9054,
      "step": 466
    },
    {
      "epoch": 0.7033132530120482,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.9668674698795187e-06,
      "loss": 0.9472,
      "step": 467
    },
    {
      "epoch": 0.7048192771084337,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.9518072289156627e-06,
      "loss": 0.9451,
      "step": 468
    },
    {
      "epoch": 0.7063253012048193,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.936746987951807e-06,
      "loss": 0.9696,
      "step": 469
    },
    {
      "epoch": 0.7078313253012049,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.9216867469879524e-06,
      "loss": 0.9095,
      "step": 470
    },
    {
      "epoch": 0.7093373493975904,
      "grad_norm": 0.189453125,
      "learning_rate": 2.906626506024097e-06,
      "loss": 0.9395,
      "step": 471
    },
    {
      "epoch": 0.7108433734939759,
      "grad_norm": 0.28515625,
      "learning_rate": 2.891566265060241e-06,
      "loss": 0.9205,
      "step": 472
    },
    {
      "epoch": 0.7123493975903614,
      "grad_norm": 0.19140625,
      "learning_rate": 2.876506024096386e-06,
      "loss": 0.9489,
      "step": 473
    },
    {
      "epoch": 0.713855421686747,
      "grad_norm": 0.1875,
      "learning_rate": 2.8614457831325306e-06,
      "loss": 0.925,
      "step": 474
    },
    {
      "epoch": 0.7153614457831325,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.846385542168675e-06,
      "loss": 0.9041,
      "step": 475
    },
    {
      "epoch": 0.7168674698795181,
      "grad_norm": 0.19140625,
      "learning_rate": 2.83132530120482e-06,
      "loss": 0.9385,
      "step": 476
    },
    {
      "epoch": 0.7183734939759037,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.8162650602409643e-06,
      "loss": 0.9326,
      "step": 477
    },
    {
      "epoch": 0.7198795180722891,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.8012048192771087e-06,
      "loss": 0.88,
      "step": 478
    },
    {
      "epoch": 0.7213855421686747,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.7861445783132536e-06,
      "loss": 0.9096,
      "step": 479
    },
    {
      "epoch": 0.7228915662650602,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.771084337349398e-06,
      "loss": 0.891,
      "step": 480
    },
    {
      "epoch": 0.7243975903614458,
      "grad_norm": 0.193359375,
      "learning_rate": 2.7560240963855424e-06,
      "loss": 0.9029,
      "step": 481
    },
    {
      "epoch": 0.7259036144578314,
      "grad_norm": 0.185546875,
      "learning_rate": 2.7409638554216873e-06,
      "loss": 0.9201,
      "step": 482
    },
    {
      "epoch": 0.7274096385542169,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.7259036144578317e-06,
      "loss": 0.9015,
      "step": 483
    },
    {
      "epoch": 0.7289156626506024,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.710843373493976e-06,
      "loss": 0.9402,
      "step": 484
    },
    {
      "epoch": 0.7304216867469879,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.695783132530121e-06,
      "loss": 0.931,
      "step": 485
    },
    {
      "epoch": 0.7319277108433735,
      "grad_norm": 0.181640625,
      "learning_rate": 2.6807228915662654e-06,
      "loss": 0.9036,
      "step": 486
    },
    {
      "epoch": 0.733433734939759,
      "grad_norm": 0.294921875,
      "learning_rate": 2.66566265060241e-06,
      "loss": 0.9242,
      "step": 487
    },
    {
      "epoch": 0.7349397590361446,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.6506024096385547e-06,
      "loss": 0.8932,
      "step": 488
    },
    {
      "epoch": 0.7364457831325302,
      "grad_norm": 0.197265625,
      "learning_rate": 2.635542168674699e-06,
      "loss": 0.9367,
      "step": 489
    },
    {
      "epoch": 0.7379518072289156,
      "grad_norm": 0.1875,
      "learning_rate": 2.6204819277108435e-06,
      "loss": 0.9091,
      "step": 490
    },
    {
      "epoch": 0.7394578313253012,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.6054216867469884e-06,
      "loss": 0.9249,
      "step": 491
    },
    {
      "epoch": 0.7409638554216867,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.590361445783133e-06,
      "loss": 0.893,
      "step": 492
    },
    {
      "epoch": 0.7424698795180723,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.5753012048192772e-06,
      "loss": 0.9492,
      "step": 493
    },
    {
      "epoch": 0.7439759036144579,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.560240963855422e-06,
      "loss": 0.9127,
      "step": 494
    },
    {
      "epoch": 0.7454819277108434,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.5451807228915665e-06,
      "loss": 0.9011,
      "step": 495
    },
    {
      "epoch": 0.7469879518072289,
      "grad_norm": 0.1953125,
      "learning_rate": 2.530120481927711e-06,
      "loss": 0.9478,
      "step": 496
    },
    {
      "epoch": 0.7484939759036144,
      "grad_norm": 0.29296875,
      "learning_rate": 2.5150602409638558e-06,
      "loss": 0.9589,
      "step": 497
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.5e-06,
      "loss": 0.9117,
      "step": 498
    },
    {
      "epoch": 0.7515060240963856,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.484939759036145e-06,
      "loss": 0.9263,
      "step": 499
    },
    {
      "epoch": 0.7530120481927711,
      "grad_norm": 0.20703125,
      "learning_rate": 2.469879518072289e-06,
      "loss": 0.998,
      "step": 500
    },
    {
      "epoch": 0.7545180722891566,
      "grad_norm": 0.19921875,
      "learning_rate": 2.454819277108434e-06,
      "loss": 0.8925,
      "step": 501
    },
    {
      "epoch": 0.7560240963855421,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.4397590361445788e-06,
      "loss": 0.9535,
      "step": 502
    },
    {
      "epoch": 0.7575301204819277,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.424698795180723e-06,
      "loss": 0.9093,
      "step": 503
    },
    {
      "epoch": 0.7590361445783133,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.4096385542168676e-06,
      "loss": 0.9159,
      "step": 504
    },
    {
      "epoch": 0.7605421686746988,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3945783132530125e-06,
      "loss": 0.9613,
      "step": 505
    },
    {
      "epoch": 0.7620481927710844,
      "grad_norm": 0.2353515625,
      "learning_rate": 2.379518072289157e-06,
      "loss": 0.9133,
      "step": 506
    },
    {
      "epoch": 0.7635542168674698,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3644578313253013e-06,
      "loss": 0.958,
      "step": 507
    },
    {
      "epoch": 0.7650602409638554,
      "grad_norm": 0.19140625,
      "learning_rate": 2.349397590361446e-06,
      "loss": 0.9253,
      "step": 508
    },
    {
      "epoch": 0.766566265060241,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3343373493975906e-06,
      "loss": 0.9329,
      "step": 509
    },
    {
      "epoch": 0.7680722891566265,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.319277108433735e-06,
      "loss": 0.9412,
      "step": 510
    },
    {
      "epoch": 0.7695783132530121,
      "grad_norm": 0.2109375,
      "learning_rate": 2.30421686746988e-06,
      "loss": 0.9099,
      "step": 511
    },
    {
      "epoch": 0.7710843373493976,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2891566265060243e-06,
      "loss": 0.9155,
      "step": 512
    },
    {
      "epoch": 0.7725903614457831,
      "grad_norm": 0.208984375,
      "learning_rate": 2.2740963855421687e-06,
      "loss": 0.9182,
      "step": 513
    },
    {
      "epoch": 0.7740963855421686,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2590361445783136e-06,
      "loss": 0.9211,
      "step": 514
    },
    {
      "epoch": 0.7756024096385542,
      "grad_norm": 0.1953125,
      "learning_rate": 2.243975903614458e-06,
      "loss": 0.9205,
      "step": 515
    },
    {
      "epoch": 0.7771084337349398,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.2289156626506024e-06,
      "loss": 0.9182,
      "step": 516
    },
    {
      "epoch": 0.7786144578313253,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2138554216867473e-06,
      "loss": 0.9043,
      "step": 517
    },
    {
      "epoch": 0.7801204819277109,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.1987951807228917e-06,
      "loss": 0.9299,
      "step": 518
    },
    {
      "epoch": 0.7816265060240963,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.183734939759036e-06,
      "loss": 0.9268,
      "step": 519
    },
    {
      "epoch": 0.7831325301204819,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.168674698795181e-06,
      "loss": 0.9246,
      "step": 520
    },
    {
      "epoch": 0.7846385542168675,
      "grad_norm": 0.181640625,
      "learning_rate": 2.1536144578313254e-06,
      "loss": 0.9514,
      "step": 521
    },
    {
      "epoch": 0.786144578313253,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.13855421686747e-06,
      "loss": 0.8994,
      "step": 522
    },
    {
      "epoch": 0.7876506024096386,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.1234939759036147e-06,
      "loss": 0.9106,
      "step": 523
    },
    {
      "epoch": 0.7891566265060241,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.1084337349397595e-06,
      "loss": 0.9043,
      "step": 524
    },
    {
      "epoch": 0.7906626506024096,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.0933734939759035e-06,
      "loss": 0.895,
      "step": 525
    },
    {
      "epoch": 0.7921686746987951,
      "grad_norm": 0.39453125,
      "learning_rate": 2.0783132530120484e-06,
      "loss": 0.9457,
      "step": 526
    },
    {
      "epoch": 0.7936746987951807,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.0632530120481932e-06,
      "loss": 0.9343,
      "step": 527
    },
    {
      "epoch": 0.7951807228915663,
      "grad_norm": 0.201171875,
      "learning_rate": 2.0481927710843377e-06,
      "loss": 0.9402,
      "step": 528
    },
    {
      "epoch": 0.7966867469879518,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.033132530120482e-06,
      "loss": 0.9171,
      "step": 529
    },
    {
      "epoch": 0.7981927710843374,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.018072289156627e-06,
      "loss": 0.8961,
      "step": 530
    },
    {
      "epoch": 0.7996987951807228,
      "grad_norm": 0.19921875,
      "learning_rate": 2.0030120481927714e-06,
      "loss": 0.9184,
      "step": 531
    },
    {
      "epoch": 0.8012048192771084,
      "grad_norm": 0.193359375,
      "learning_rate": 1.987951807228916e-06,
      "loss": 0.9145,
      "step": 532
    },
    {
      "epoch": 0.802710843373494,
      "grad_norm": 0.271484375,
      "learning_rate": 1.9728915662650606e-06,
      "loss": 0.9461,
      "step": 533
    },
    {
      "epoch": 0.8042168674698795,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.957831325301205e-06,
      "loss": 0.8839,
      "step": 534
    },
    {
      "epoch": 0.8057228915662651,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.9427710843373495e-06,
      "loss": 0.9423,
      "step": 535
    },
    {
      "epoch": 0.8072289156626506,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.9277108433734943e-06,
      "loss": 0.893,
      "step": 536
    },
    {
      "epoch": 0.8087349397590361,
      "grad_norm": 0.248046875,
      "learning_rate": 1.9126506024096388e-06,
      "loss": 0.8957,
      "step": 537
    },
    {
      "epoch": 0.8102409638554217,
      "grad_norm": 0.197265625,
      "learning_rate": 1.8975903614457832e-06,
      "loss": 0.9693,
      "step": 538
    },
    {
      "epoch": 0.8117469879518072,
      "grad_norm": 0.193359375,
      "learning_rate": 1.882530120481928e-06,
      "loss": 0.9408,
      "step": 539
    },
    {
      "epoch": 0.8132530120481928,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.8674698795180723e-06,
      "loss": 0.99,
      "step": 540
    },
    {
      "epoch": 0.8147590361445783,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.8524096385542171e-06,
      "loss": 0.9168,
      "step": 541
    },
    {
      "epoch": 0.8162650602409639,
      "grad_norm": 0.2373046875,
      "learning_rate": 1.8373493975903617e-06,
      "loss": 0.9519,
      "step": 542
    },
    {
      "epoch": 0.8177710843373494,
      "grad_norm": 0.1875,
      "learning_rate": 1.8222891566265062e-06,
      "loss": 0.932,
      "step": 543
    },
    {
      "epoch": 0.8192771084337349,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.8072289156626508e-06,
      "loss": 0.9158,
      "step": 544
    },
    {
      "epoch": 0.8207831325301205,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.7921686746987955e-06,
      "loss": 0.9167,
      "step": 545
    },
    {
      "epoch": 0.822289156626506,
      "grad_norm": 0.185546875,
      "learning_rate": 1.7771084337349399e-06,
      "loss": 0.8845,
      "step": 546
    },
    {
      "epoch": 0.8237951807228916,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.7620481927710845e-06,
      "loss": 0.9324,
      "step": 547
    },
    {
      "epoch": 0.8253012048192772,
      "grad_norm": 0.1845703125,
      "learning_rate": 1.7469879518072292e-06,
      "loss": 0.9297,
      "step": 548
    },
    {
      "epoch": 0.8268072289156626,
      "grad_norm": 0.1796875,
      "learning_rate": 1.7319277108433736e-06,
      "loss": 0.9223,
      "step": 549
    },
    {
      "epoch": 0.8283132530120482,
      "grad_norm": 0.208984375,
      "learning_rate": 1.7168674698795182e-06,
      "loss": 0.9424,
      "step": 550
    },
    {
      "epoch": 0.8298192771084337,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.7018072289156629e-06,
      "loss": 0.9062,
      "step": 551
    },
    {
      "epoch": 0.8313253012048193,
      "grad_norm": 0.265625,
      "learning_rate": 1.6867469879518073e-06,
      "loss": 0.9142,
      "step": 552
    },
    {
      "epoch": 0.8328313253012049,
      "grad_norm": 0.21875,
      "learning_rate": 1.671686746987952e-06,
      "loss": 0.9349,
      "step": 553
    },
    {
      "epoch": 0.8343373493975904,
      "grad_norm": 0.20703125,
      "learning_rate": 1.6566265060240966e-06,
      "loss": 0.9309,
      "step": 554
    },
    {
      "epoch": 0.8358433734939759,
      "grad_norm": 0.197265625,
      "learning_rate": 1.641566265060241e-06,
      "loss": 0.9448,
      "step": 555
    },
    {
      "epoch": 0.8373493975903614,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.6265060240963856e-06,
      "loss": 0.93,
      "step": 556
    },
    {
      "epoch": 0.838855421686747,
      "grad_norm": 0.181640625,
      "learning_rate": 1.6114457831325303e-06,
      "loss": 0.9096,
      "step": 557
    },
    {
      "epoch": 0.8403614457831325,
      "grad_norm": 0.1884765625,
      "learning_rate": 1.5963855421686747e-06,
      "loss": 0.8934,
      "step": 558
    },
    {
      "epoch": 0.8418674698795181,
      "grad_norm": 0.25390625,
      "learning_rate": 1.5813253012048193e-06,
      "loss": 0.8913,
      "step": 559
    },
    {
      "epoch": 0.8433734939759037,
      "grad_norm": 0.1845703125,
      "learning_rate": 1.566265060240964e-06,
      "loss": 0.9307,
      "step": 560
    },
    {
      "epoch": 0.8448795180722891,
      "grad_norm": 0.208984375,
      "learning_rate": 1.5512048192771086e-06,
      "loss": 0.9236,
      "step": 561
    },
    {
      "epoch": 0.8463855421686747,
      "grad_norm": 0.201171875,
      "learning_rate": 1.536144578313253e-06,
      "loss": 0.9036,
      "step": 562
    },
    {
      "epoch": 0.8478915662650602,
      "grad_norm": 0.189453125,
      "learning_rate": 1.5210843373493977e-06,
      "loss": 0.8784,
      "step": 563
    },
    {
      "epoch": 0.8493975903614458,
      "grad_norm": 0.1875,
      "learning_rate": 1.5060240963855425e-06,
      "loss": 0.9217,
      "step": 564
    },
    {
      "epoch": 0.8509036144578314,
      "grad_norm": 0.18359375,
      "learning_rate": 1.4909638554216867e-06,
      "loss": 0.9193,
      "step": 565
    },
    {
      "epoch": 0.8524096385542169,
      "grad_norm": 0.56640625,
      "learning_rate": 1.4759036144578314e-06,
      "loss": 0.9356,
      "step": 566
    },
    {
      "epoch": 0.8539156626506024,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.4608433734939762e-06,
      "loss": 0.9236,
      "step": 567
    },
    {
      "epoch": 0.8554216867469879,
      "grad_norm": 0.1845703125,
      "learning_rate": 1.4457831325301204e-06,
      "loss": 0.9239,
      "step": 568
    },
    {
      "epoch": 0.8569277108433735,
      "grad_norm": 0.201171875,
      "learning_rate": 1.4307228915662653e-06,
      "loss": 0.9066,
      "step": 569
    },
    {
      "epoch": 0.858433734939759,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.41566265060241e-06,
      "loss": 0.9156,
      "step": 570
    },
    {
      "epoch": 0.8599397590361446,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.4006024096385544e-06,
      "loss": 0.9249,
      "step": 571
    },
    {
      "epoch": 0.8614457831325302,
      "grad_norm": 0.279296875,
      "learning_rate": 1.385542168674699e-06,
      "loss": 0.9129,
      "step": 572
    },
    {
      "epoch": 0.8629518072289156,
      "grad_norm": 0.2119140625,
      "learning_rate": 1.3704819277108436e-06,
      "loss": 0.9019,
      "step": 573
    },
    {
      "epoch": 0.8644578313253012,
      "grad_norm": 0.193359375,
      "learning_rate": 1.355421686746988e-06,
      "loss": 0.9654,
      "step": 574
    },
    {
      "epoch": 0.8659638554216867,
      "grad_norm": 0.1845703125,
      "learning_rate": 1.3403614457831327e-06,
      "loss": 0.8949,
      "step": 575
    },
    {
      "epoch": 0.8674698795180723,
      "grad_norm": 0.1953125,
      "learning_rate": 1.3253012048192773e-06,
      "loss": 0.8927,
      "step": 576
    },
    {
      "epoch": 0.8689759036144579,
      "grad_norm": 0.189453125,
      "learning_rate": 1.3102409638554218e-06,
      "loss": 0.9113,
      "step": 577
    },
    {
      "epoch": 0.8704819277108434,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.2951807228915664e-06,
      "loss": 0.9136,
      "step": 578
    },
    {
      "epoch": 0.8719879518072289,
      "grad_norm": 0.3046875,
      "learning_rate": 1.280120481927711e-06,
      "loss": 0.884,
      "step": 579
    },
    {
      "epoch": 0.8734939759036144,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.2650602409638555e-06,
      "loss": 0.9002,
      "step": 580
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.189453125,
      "learning_rate": 1.25e-06,
      "loss": 0.9362,
      "step": 581
    },
    {
      "epoch": 0.8765060240963856,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.2349397590361445e-06,
      "loss": 0.9401,
      "step": 582
    },
    {
      "epoch": 0.8780120481927711,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.2198795180722894e-06,
      "loss": 0.9322,
      "step": 583
    },
    {
      "epoch": 0.8795180722891566,
      "grad_norm": 0.330078125,
      "learning_rate": 1.2048192771084338e-06,
      "loss": 0.9905,
      "step": 584
    },
    {
      "epoch": 0.8810240963855421,
      "grad_norm": 0.19921875,
      "learning_rate": 1.1897590361445784e-06,
      "loss": 0.9377,
      "step": 585
    },
    {
      "epoch": 0.8825301204819277,
      "grad_norm": 0.23046875,
      "learning_rate": 1.174698795180723e-06,
      "loss": 0.9394,
      "step": 586
    },
    {
      "epoch": 0.8840361445783133,
      "grad_norm": 0.189453125,
      "learning_rate": 1.1596385542168675e-06,
      "loss": 0.8716,
      "step": 587
    },
    {
      "epoch": 0.8855421686746988,
      "grad_norm": 0.1884765625,
      "learning_rate": 1.1445783132530121e-06,
      "loss": 0.9204,
      "step": 588
    },
    {
      "epoch": 0.8870481927710844,
      "grad_norm": 0.302734375,
      "learning_rate": 1.1295180722891568e-06,
      "loss": 0.9481,
      "step": 589
    },
    {
      "epoch": 0.8885542168674698,
      "grad_norm": 0.1884765625,
      "learning_rate": 1.1144578313253012e-06,
      "loss": 0.8887,
      "step": 590
    },
    {
      "epoch": 0.8900602409638554,
      "grad_norm": 0.197265625,
      "learning_rate": 1.0993975903614459e-06,
      "loss": 0.9949,
      "step": 591
    },
    {
      "epoch": 0.891566265060241,
      "grad_norm": 0.193359375,
      "learning_rate": 1.0843373493975905e-06,
      "loss": 0.9556,
      "step": 592
    },
    {
      "epoch": 0.8930722891566265,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.069277108433735e-06,
      "loss": 0.921,
      "step": 593
    },
    {
      "epoch": 0.8945783132530121,
      "grad_norm": 0.1845703125,
      "learning_rate": 1.0542168674698798e-06,
      "loss": 0.8998,
      "step": 594
    },
    {
      "epoch": 0.8960843373493976,
      "grad_norm": 0.1884765625,
      "learning_rate": 1.0391566265060242e-06,
      "loss": 0.8747,
      "step": 595
    },
    {
      "epoch": 0.8975903614457831,
      "grad_norm": 0.1953125,
      "learning_rate": 1.0240963855421688e-06,
      "loss": 0.9286,
      "step": 596
    },
    {
      "epoch": 0.8990963855421686,
      "grad_norm": 0.21875,
      "learning_rate": 1.0090361445783135e-06,
      "loss": 0.9195,
      "step": 597
    },
    {
      "epoch": 0.9006024096385542,
      "grad_norm": 0.2353515625,
      "learning_rate": 9.93975903614458e-07,
      "loss": 0.9209,
      "step": 598
    },
    {
      "epoch": 0.9021084337349398,
      "grad_norm": 0.1943359375,
      "learning_rate": 9.789156626506025e-07,
      "loss": 0.9205,
      "step": 599
    },
    {
      "epoch": 0.9036144578313253,
      "grad_norm": 0.189453125,
      "learning_rate": 9.638554216867472e-07,
      "loss": 0.9264,
      "step": 600
    },
    {
      "epoch": 0.9051204819277109,
      "grad_norm": 0.1875,
      "learning_rate": 9.487951807228916e-07,
      "loss": 0.917,
      "step": 601
    },
    {
      "epoch": 0.9066265060240963,
      "grad_norm": 0.91796875,
      "learning_rate": 9.337349397590361e-07,
      "loss": 0.8826,
      "step": 602
    },
    {
      "epoch": 0.9081325301204819,
      "grad_norm": 0.1982421875,
      "learning_rate": 9.186746987951809e-07,
      "loss": 0.9075,
      "step": 603
    },
    {
      "epoch": 0.9096385542168675,
      "grad_norm": 0.1953125,
      "learning_rate": 9.036144578313254e-07,
      "loss": 0.909,
      "step": 604
    },
    {
      "epoch": 0.911144578313253,
      "grad_norm": 0.2392578125,
      "learning_rate": 8.885542168674699e-07,
      "loss": 0.9121,
      "step": 605
    },
    {
      "epoch": 0.9126506024096386,
      "grad_norm": 0.2216796875,
      "learning_rate": 8.734939759036146e-07,
      "loss": 0.9607,
      "step": 606
    },
    {
      "epoch": 0.9141566265060241,
      "grad_norm": 0.185546875,
      "learning_rate": 8.584337349397591e-07,
      "loss": 0.9157,
      "step": 607
    },
    {
      "epoch": 0.9156626506024096,
      "grad_norm": 0.1962890625,
      "learning_rate": 8.433734939759036e-07,
      "loss": 0.9551,
      "step": 608
    },
    {
      "epoch": 0.9171686746987951,
      "grad_norm": 0.2001953125,
      "learning_rate": 8.283132530120483e-07,
      "loss": 0.8956,
      "step": 609
    },
    {
      "epoch": 0.9186746987951807,
      "grad_norm": 0.185546875,
      "learning_rate": 8.132530120481928e-07,
      "loss": 0.8996,
      "step": 610
    },
    {
      "epoch": 0.9201807228915663,
      "grad_norm": 0.1982421875,
      "learning_rate": 7.981927710843373e-07,
      "loss": 0.9452,
      "step": 611
    },
    {
      "epoch": 0.9216867469879518,
      "grad_norm": 0.1953125,
      "learning_rate": 7.83132530120482e-07,
      "loss": 0.9051,
      "step": 612
    },
    {
      "epoch": 0.9231927710843374,
      "grad_norm": 0.19140625,
      "learning_rate": 7.680722891566265e-07,
      "loss": 0.9442,
      "step": 613
    },
    {
      "epoch": 0.9246987951807228,
      "grad_norm": 0.2041015625,
      "learning_rate": 7.530120481927713e-07,
      "loss": 0.9165,
      "step": 614
    },
    {
      "epoch": 0.9262048192771084,
      "grad_norm": 0.19140625,
      "learning_rate": 7.379518072289157e-07,
      "loss": 0.9527,
      "step": 615
    },
    {
      "epoch": 0.927710843373494,
      "grad_norm": 0.1884765625,
      "learning_rate": 7.228915662650602e-07,
      "loss": 0.9116,
      "step": 616
    },
    {
      "epoch": 0.9292168674698795,
      "grad_norm": 0.244140625,
      "learning_rate": 7.07831325301205e-07,
      "loss": 0.958,
      "step": 617
    },
    {
      "epoch": 0.9307228915662651,
      "grad_norm": 0.1845703125,
      "learning_rate": 6.927710843373495e-07,
      "loss": 0.8975,
      "step": 618
    },
    {
      "epoch": 0.9322289156626506,
      "grad_norm": 0.1845703125,
      "learning_rate": 6.77710843373494e-07,
      "loss": 0.9357,
      "step": 619
    },
    {
      "epoch": 0.9337349397590361,
      "grad_norm": 0.189453125,
      "learning_rate": 6.626506024096387e-07,
      "loss": 0.9423,
      "step": 620
    },
    {
      "epoch": 0.9352409638554217,
      "grad_norm": 0.2041015625,
      "learning_rate": 6.475903614457832e-07,
      "loss": 0.9072,
      "step": 621
    },
    {
      "epoch": 0.9367469879518072,
      "grad_norm": 0.1962890625,
      "learning_rate": 6.325301204819277e-07,
      "loss": 0.9324,
      "step": 622
    },
    {
      "epoch": 0.9382530120481928,
      "grad_norm": 0.306640625,
      "learning_rate": 6.174698795180723e-07,
      "loss": 0.9504,
      "step": 623
    },
    {
      "epoch": 0.9397590361445783,
      "grad_norm": 0.185546875,
      "learning_rate": 6.024096385542169e-07,
      "loss": 0.9149,
      "step": 624
    },
    {
      "epoch": 0.9412650602409639,
      "grad_norm": 0.2353515625,
      "learning_rate": 5.873493975903615e-07,
      "loss": 0.9331,
      "step": 625
    },
    {
      "epoch": 0.9427710843373494,
      "grad_norm": 0.201171875,
      "learning_rate": 5.722891566265061e-07,
      "loss": 0.8771,
      "step": 626
    },
    {
      "epoch": 0.9442771084337349,
      "grad_norm": 0.2314453125,
      "learning_rate": 5.572289156626506e-07,
      "loss": 0.9011,
      "step": 627
    },
    {
      "epoch": 0.9457831325301205,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.421686746987952e-07,
      "loss": 0.9246,
      "step": 628
    },
    {
      "epoch": 0.947289156626506,
      "grad_norm": 0.205078125,
      "learning_rate": 5.271084337349399e-07,
      "loss": 0.8982,
      "step": 629
    },
    {
      "epoch": 0.9487951807228916,
      "grad_norm": 0.1865234375,
      "learning_rate": 5.120481927710844e-07,
      "loss": 0.9147,
      "step": 630
    },
    {
      "epoch": 0.9503012048192772,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.96987951807229e-07,
      "loss": 0.9326,
      "step": 631
    },
    {
      "epoch": 0.9518072289156626,
      "grad_norm": 0.2236328125,
      "learning_rate": 4.819277108433736e-07,
      "loss": 0.9635,
      "step": 632
    },
    {
      "epoch": 0.9533132530120482,
      "grad_norm": 0.322265625,
      "learning_rate": 4.6686746987951807e-07,
      "loss": 0.922,
      "step": 633
    },
    {
      "epoch": 0.9548192771084337,
      "grad_norm": 0.2099609375,
      "learning_rate": 4.518072289156627e-07,
      "loss": 0.9175,
      "step": 634
    },
    {
      "epoch": 0.9563253012048193,
      "grad_norm": 0.1953125,
      "learning_rate": 4.367469879518073e-07,
      "loss": 0.9235,
      "step": 635
    },
    {
      "epoch": 0.9578313253012049,
      "grad_norm": 0.2333984375,
      "learning_rate": 4.216867469879518e-07,
      "loss": 0.9155,
      "step": 636
    },
    {
      "epoch": 0.9593373493975904,
      "grad_norm": 0.203125,
      "learning_rate": 4.066265060240964e-07,
      "loss": 0.9512,
      "step": 637
    },
    {
      "epoch": 0.9608433734939759,
      "grad_norm": 0.19140625,
      "learning_rate": 3.91566265060241e-07,
      "loss": 0.9534,
      "step": 638
    },
    {
      "epoch": 0.9623493975903614,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.7650602409638563e-07,
      "loss": 0.9117,
      "step": 639
    },
    {
      "epoch": 0.963855421686747,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.614457831325301e-07,
      "loss": 0.9556,
      "step": 640
    },
    {
      "epoch": 0.9653614457831325,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.4638554216867475e-07,
      "loss": 0.8901,
      "step": 641
    },
    {
      "epoch": 0.9668674698795181,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.3132530120481933e-07,
      "loss": 0.907,
      "step": 642
    },
    {
      "epoch": 0.9683734939759037,
      "grad_norm": 0.193359375,
      "learning_rate": 3.1626506024096387e-07,
      "loss": 0.8773,
      "step": 643
    },
    {
      "epoch": 0.9698795180722891,
      "grad_norm": 0.1796875,
      "learning_rate": 3.0120481927710845e-07,
      "loss": 0.896,
      "step": 644
    },
    {
      "epoch": 0.9713855421686747,
      "grad_norm": 0.208984375,
      "learning_rate": 2.8614457831325304e-07,
      "loss": 0.8964,
      "step": 645
    },
    {
      "epoch": 0.9728915662650602,
      "grad_norm": 0.1875,
      "learning_rate": 2.710843373493976e-07,
      "loss": 0.9323,
      "step": 646
    },
    {
      "epoch": 0.9743975903614458,
      "grad_norm": 1.734375,
      "learning_rate": 2.560240963855422e-07,
      "loss": 0.9402,
      "step": 647
    },
    {
      "epoch": 0.9759036144578314,
      "grad_norm": 0.19140625,
      "learning_rate": 2.409638554216868e-07,
      "loss": 0.895,
      "step": 648
    },
    {
      "epoch": 0.9774096385542169,
      "grad_norm": 0.185546875,
      "learning_rate": 2.2590361445783135e-07,
      "loss": 0.9357,
      "step": 649
    },
    {
      "epoch": 0.9789156626506024,
      "grad_norm": 0.189453125,
      "learning_rate": 2.108433734939759e-07,
      "loss": 0.9333,
      "step": 650
    },
    {
      "epoch": 0.9804216867469879,
      "grad_norm": 0.220703125,
      "learning_rate": 1.957831325301205e-07,
      "loss": 0.8999,
      "step": 651
    },
    {
      "epoch": 0.9819277108433735,
      "grad_norm": 0.27734375,
      "learning_rate": 1.8072289156626505e-07,
      "loss": 0.8891,
      "step": 652
    },
    {
      "epoch": 0.983433734939759,
      "grad_norm": 0.201171875,
      "learning_rate": 1.6566265060240967e-07,
      "loss": 0.9329,
      "step": 653
    },
    {
      "epoch": 0.9849397590361446,
      "grad_norm": 0.212890625,
      "learning_rate": 1.5060240963855423e-07,
      "loss": 0.989,
      "step": 654
    },
    {
      "epoch": 0.9864457831325302,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.355421686746988e-07,
      "loss": 0.9469,
      "step": 655
    },
    {
      "epoch": 0.9879518072289156,
      "grad_norm": 0.275390625,
      "learning_rate": 1.204819277108434e-07,
      "loss": 0.928,
      "step": 656
    },
    {
      "epoch": 0.9894578313253012,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.0542168674698796e-07,
      "loss": 0.8903,
      "step": 657
    },
    {
      "epoch": 0.9909638554216867,
      "grad_norm": 0.189453125,
      "learning_rate": 9.036144578313253e-08,
      "loss": 0.8711,
      "step": 658
    },
    {
      "epoch": 0.9924698795180723,
      "grad_norm": 0.232421875,
      "learning_rate": 7.530120481927711e-08,
      "loss": 0.8955,
      "step": 659
    },
    {
      "epoch": 0.9939759036144579,
      "grad_norm": 0.1884765625,
      "learning_rate": 6.02409638554217e-08,
      "loss": 0.9138,
      "step": 660
    },
    {
      "epoch": 0.9954819277108434,
      "grad_norm": 0.19140625,
      "learning_rate": 4.5180722891566264e-08,
      "loss": 0.9462,
      "step": 661
    },
    {
      "epoch": 0.9969879518072289,
      "grad_norm": 0.1953125,
      "learning_rate": 3.012048192771085e-08,
      "loss": 0.9727,
      "step": 662
    },
    {
      "epoch": 0.9984939759036144,
      "grad_norm": 0.19140625,
      "learning_rate": 1.5060240963855425e-08,
      "loss": 0.9021,
      "step": 663
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0,
      "loss": 0.9495,
      "step": 664
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 664,
  "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": 2.129060370482987e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}