{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9996048367975974,
  "eval_steps": 500,
  "global_step": 1581,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.2656612225200705,
      "learning_rate": 1.0416666666666667e-07,
      "loss": 0.0917,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 37.71678549668734,
      "learning_rate": 2.0833333333333333e-07,
      "loss": 1.137,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.895397158947514,
      "learning_rate": 3.125e-07,
      "loss": 0.7479,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.183793101679505,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.4486,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.891246608292718,
      "learning_rate": 5.208333333333334e-07,
      "loss": 0.761,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.994012792789633,
      "learning_rate": 6.25e-07,
      "loss": 0.4221,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.526634963381646,
      "learning_rate": 7.291666666666667e-07,
      "loss": 0.6891,
      "step": 7
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.301318633524378,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.2271,
      "step": 8
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.30560285482862,
      "learning_rate": 9.375000000000001e-07,
      "loss": 0.8797,
      "step": 9
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.6060244744932,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 1.0926,
      "step": 10
    },
    {
      "epoch": 0.01,
      "grad_norm": 22.285069634472343,
      "learning_rate": 1.1458333333333333e-06,
      "loss": 0.6605,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.910361966967372,
      "learning_rate": 1.25e-06,
      "loss": 0.4612,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.847411644670956,
      "learning_rate": 1.3541666666666667e-06,
      "loss": 0.827,
      "step": 13
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.670988107560447,
      "learning_rate": 1.4583333333333335e-06,
      "loss": 0.4954,
      "step": 14
    },
    {
      "epoch": 0.01,
      "grad_norm": 47.28509413074193,
      "learning_rate": 1.5625e-06,
      "loss": 1.0251,
      "step": 15
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.371926282375087,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.2646,
      "step": 16
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.58598956011841,
      "learning_rate": 1.7708333333333337e-06,
      "loss": 0.4352,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.707206567726946,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.6815,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.703607804492943,
      "learning_rate": 1.9791666666666666e-06,
      "loss": 0.5448,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.897272953634102,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.4308,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.118015556739405,
      "learning_rate": 2.1875000000000002e-06,
      "loss": 1.1068,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.298309356391181,
      "learning_rate": 2.2916666666666666e-06,
      "loss": 0.2084,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.37628919494901,
      "learning_rate": 2.395833333333334e-06,
      "loss": 1.1782,
      "step": 23
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.479434977474586,
      "learning_rate": 2.5e-06,
      "loss": 0.4243,
      "step": 24
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.259739501649532,
      "learning_rate": 2.604166666666667e-06,
      "loss": 0.4217,
      "step": 25
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.960559893553361,
      "learning_rate": 2.7083333333333334e-06,
      "loss": 0.4158,
      "step": 26
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.429086598913813,
      "learning_rate": 2.8125e-06,
      "loss": 0.6773,
      "step": 27
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.209207852868984,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.3294,
      "step": 28
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.940346431292438,
      "learning_rate": 3.0208333333333334e-06,
      "loss": 0.2523,
      "step": 29
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.523395752029726,
      "learning_rate": 3.125e-06,
      "loss": 0.3717,
      "step": 30
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.435033084102253,
      "learning_rate": 3.229166666666667e-06,
      "loss": 0.5776,
      "step": 31
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.350376484804269,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.5697,
      "step": 32
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.46023112043865,
      "learning_rate": 3.4375e-06,
      "loss": 0.3445,
      "step": 33
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.91670954958463,
      "learning_rate": 3.5416666666666673e-06,
      "loss": 0.2192,
      "step": 34
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.227868702792659,
      "learning_rate": 3.6458333333333333e-06,
      "loss": 0.4932,
      "step": 35
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.284980136108494,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.7055,
      "step": 36
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.405195303041342,
      "learning_rate": 3.854166666666667e-06,
      "loss": 0.7916,
      "step": 37
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.7766714829237134,
      "learning_rate": 3.958333333333333e-06,
      "loss": 0.0268,
      "step": 38
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.013506801089294,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.2811,
      "step": 39
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.745139379595656,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.2401,
      "step": 40
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.713988395103588,
      "learning_rate": 4.270833333333333e-06,
      "loss": 0.5153,
      "step": 41
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.301871898454228,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.2485,
      "step": 42
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.848396848766385,
      "learning_rate": 4.479166666666667e-06,
      "loss": 0.2515,
      "step": 43
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.7180885166743,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.1705,
      "step": 44
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.707560595021471,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.2393,
      "step": 45
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.348481263119,
      "learning_rate": 4.791666666666668e-06,
      "loss": 0.401,
      "step": 46
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.277477398342935,
      "learning_rate": 4.895833333333333e-06,
      "loss": 0.1771,
      "step": 47
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.225147278092627,
      "learning_rate": 5e-06,
      "loss": 0.1883,
      "step": 48
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.714093120128615,
      "learning_rate": 4.999994750411077e-06,
      "loss": 0.3014,
      "step": 49
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.29428007285362,
      "learning_rate": 4.999979001666353e-06,
      "loss": 0.6187,
      "step": 50
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.868248877610313,
      "learning_rate": 4.999952753831968e-06,
      "loss": 0.4989,
      "step": 51
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.321802600524457,
      "learning_rate": 4.999916007018154e-06,
      "loss": 0.2043,
      "step": 52
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.505912071091073,
      "learning_rate": 4.999868761379236e-06,
      "loss": 0.3569,
      "step": 53
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.737221159631332,
      "learning_rate": 4.99981101711363e-06,
      "loss": 0.1102,
      "step": 54
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.9970130112468105,
      "learning_rate": 4.999742774463843e-06,
      "loss": 0.2492,
      "step": 55
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.322204039774054,
      "learning_rate": 4.999664033716471e-06,
      "loss": 0.48,
      "step": 56
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.5526073475709055,
      "learning_rate": 4.9995747952022e-06,
      "loss": 0.2816,
      "step": 57
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6641124730067647,
      "learning_rate": 4.999475059295803e-06,
      "loss": 0.0647,
      "step": 58
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.362074836839993,
      "learning_rate": 4.999364826416136e-06,
      "loss": 0.9832,
      "step": 59
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.495240926785474,
      "learning_rate": 4.999244097026143e-06,
      "loss": 0.6477,
      "step": 60
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.202760008252028,
      "learning_rate": 4.999112871632847e-06,
      "loss": 0.3342,
      "step": 61
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.140751521029765,
      "learning_rate": 4.998971150787351e-06,
      "loss": 0.2422,
      "step": 62
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.531221088546207,
      "learning_rate": 4.998818935084836e-06,
      "loss": 0.2217,
      "step": 63
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.304101101586628,
      "learning_rate": 4.998656225164557e-06,
      "loss": 0.3276,
      "step": 64
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.351477084601795,
      "learning_rate": 4.998483021709846e-06,
      "loss": 0.2839,
      "step": 65
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.319785689748254,
      "learning_rate": 4.998299325448096e-06,
      "loss": 0.2311,
      "step": 66
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.552400967967682,
      "learning_rate": 4.9981051371507735e-06,
      "loss": 0.1399,
      "step": 67
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.4528212357398,
      "learning_rate": 4.997900457633405e-06,
      "loss": 0.33,
      "step": 68
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.549916931064769,
      "learning_rate": 4.9976852877555755e-06,
      "loss": 0.249,
      "step": 69
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.289074465413135,
      "learning_rate": 4.99745962842093e-06,
      "loss": 0.2407,
      "step": 70
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.255157682372891,
      "learning_rate": 4.997223480577162e-06,
      "loss": 0.3071,
      "step": 71
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.21536451342183,
      "learning_rate": 4.996976845216016e-06,
      "loss": 0.3498,
      "step": 72
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.694160378243387,
      "learning_rate": 4.9967197233732786e-06,
      "loss": 0.5605,
      "step": 73
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.000976165200353,
      "learning_rate": 4.996452116128778e-06,
      "loss": 0.0838,
      "step": 74
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.964975743028065,
      "learning_rate": 4.996174024606375e-06,
      "loss": 0.6742,
      "step": 75
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.168516070752894,
      "learning_rate": 4.9958854499739625e-06,
      "loss": 0.1488,
      "step": 76
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6416967558036006,
      "learning_rate": 4.995586393443461e-06,
      "loss": 0.0552,
      "step": 77
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.733989414358419,
      "learning_rate": 4.9952768562708085e-06,
      "loss": 0.6977,
      "step": 78
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.299548344755562,
      "learning_rate": 4.994956839755959e-06,
      "loss": 0.5162,
      "step": 79
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.600930630119668,
      "learning_rate": 4.994626345242878e-06,
      "loss": 0.5782,
      "step": 80
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.323689437212801,
      "learning_rate": 4.994285374119531e-06,
      "loss": 0.4152,
      "step": 81
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.109158140013431,
      "learning_rate": 4.993933927817888e-06,
      "loss": 0.414,
      "step": 82
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.800126058194779,
      "learning_rate": 4.993572007813905e-06,
      "loss": 0.1281,
      "step": 83
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.125007512015758,
      "learning_rate": 4.9931996156275285e-06,
      "loss": 0.3613,
      "step": 84
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.019597005987227,
      "learning_rate": 4.992816752822682e-06,
      "loss": 0.1949,
      "step": 85
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.425014271709907,
      "learning_rate": 4.992423421007265e-06,
      "loss": 0.5576,
      "step": 86
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.815926514021896,
      "learning_rate": 4.992019621833142e-06,
      "loss": 0.7708,
      "step": 87
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.011332290033770304,
      "learning_rate": 4.991605356996135e-06,
      "loss": 0.0,
      "step": 88
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4658112852872014,
      "learning_rate": 4.99118062823602e-06,
      "loss": 0.0134,
      "step": 89
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.64181278769754,
      "learning_rate": 4.99074543733652e-06,
      "loss": 0.485,
      "step": 90
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.460222683362177,
      "learning_rate": 4.9902997861252924e-06,
      "loss": 0.4733,
      "step": 91
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.417865384844604,
      "learning_rate": 4.989843676473926e-06,
      "loss": 0.3127,
      "step": 92
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.679053058190546,
      "learning_rate": 4.9893771102979305e-06,
      "loss": 0.043,
      "step": 93
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.30453362455745,
      "learning_rate": 4.9889000895567315e-06,
      "loss": 0.5676,
      "step": 94
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.12439385133809,
      "learning_rate": 4.98841261625366e-06,
      "loss": 0.3245,
      "step": 95
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.59403708586379,
      "learning_rate": 4.987914692435942e-06,
      "loss": 0.2386,
      "step": 96
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.993527037187059,
      "learning_rate": 4.987406320194695e-06,
      "loss": 0.1594,
      "step": 97
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.598896307863153,
      "learning_rate": 4.986887501664912e-06,
      "loss": 0.4927,
      "step": 98
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.572072341140025,
      "learning_rate": 4.986358239025465e-06,
      "loss": 0.1074,
      "step": 99
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.158141169039942,
      "learning_rate": 4.98581853449908e-06,
      "loss": 0.2743,
      "step": 100
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.177560611734636,
      "learning_rate": 4.985268390352339e-06,
      "loss": 0.4104,
      "step": 101
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.624361318630898,
      "learning_rate": 4.984707808895668e-06,
      "loss": 0.4447,
      "step": 102
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.489156859299493,
      "learning_rate": 4.984136792483322e-06,
      "loss": 0.6547,
      "step": 103
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.5095593832872147,
      "learning_rate": 4.9835553435133845e-06,
      "loss": 0.0884,
      "step": 104
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.265172937696828,
      "learning_rate": 4.982963464427749e-06,
      "loss": 0.1859,
      "step": 105
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.604090979244088,
      "learning_rate": 4.982361157712114e-06,
      "loss": 0.2245,
      "step": 106
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.861771386959831,
      "learning_rate": 4.981748425895968e-06,
      "loss": 0.2213,
      "step": 107
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.006355876944076,
      "learning_rate": 4.9811252715525835e-06,
      "loss": 0.8064,
      "step": 108
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.294557387145929,
      "learning_rate": 4.980491697299005e-06,
      "loss": 0.249,
      "step": 109
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.018720984463257,
      "learning_rate": 4.979847705796035e-06,
      "loss": 0.1311,
      "step": 110
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.046383542180305,
      "learning_rate": 4.979193299748225e-06,
      "loss": 0.7158,
      "step": 111
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.139919871450443,
      "learning_rate": 4.978528481903868e-06,
      "loss": 0.2277,
      "step": 112
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.719032459088086,
      "learning_rate": 4.977853255054978e-06,
      "loss": 0.1229,
      "step": 113
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.035410233181668,
      "learning_rate": 4.977167622037287e-06,
      "loss": 0.2215,
      "step": 114
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.55615821929029,
      "learning_rate": 4.976471585730227e-06,
      "loss": 0.3229,
      "step": 115
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.252747198966202,
      "learning_rate": 4.9757651490569235e-06,
      "loss": 0.1689,
      "step": 116
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.26860256356231,
      "learning_rate": 4.975048314984176e-06,
      "loss": 0.3926,
      "step": 117
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.439627879036101,
      "learning_rate": 4.974321086522453e-06,
      "loss": 0.5022,
      "step": 118
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.200371477671347,
      "learning_rate": 4.973583466725875e-06,
      "loss": 0.453,
      "step": 119
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.137229388267069,
      "learning_rate": 4.972835458692202e-06,
      "loss": 0.1897,
      "step": 120
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.7150017289995985,
      "learning_rate": 4.9720770655628216e-06,
      "loss": 0.2803,
      "step": 121
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.141197932858308,
      "learning_rate": 4.971308290522737e-06,
      "loss": 0.1684,
      "step": 122
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.373554535908345,
      "learning_rate": 4.9705291368005485e-06,
      "loss": 0.2033,
      "step": 123
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.776634419185687,
      "learning_rate": 4.9697396076684465e-06,
      "loss": 0.6946,
      "step": 124
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.34632163911218,
      "learning_rate": 4.968939706442195e-06,
      "loss": 0.4237,
      "step": 125
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.199015141722331,
      "learning_rate": 4.968129436481114e-06,
      "loss": 0.5535,
      "step": 126
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.115336542030487,
      "learning_rate": 4.967308801188072e-06,
      "loss": 0.1513,
      "step": 127
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.83142325605739,
      "learning_rate": 4.966477804009467e-06,
      "loss": 0.6237,
      "step": 128
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.880253167962748,
      "learning_rate": 4.965636448435214e-06,
      "loss": 0.1413,
      "step": 129
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.647769719523428,
      "learning_rate": 4.964784737998728e-06,
      "loss": 0.6148,
      "step": 130
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.417627793803241,
      "learning_rate": 4.963922676276916e-06,
      "loss": 0.1604,
      "step": 131
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.489083928200404,
      "learning_rate": 4.963050266890152e-06,
      "loss": 0.5219,
      "step": 132
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.857201755335243,
      "learning_rate": 4.962167513502268e-06,
      "loss": 0.256,
      "step": 133
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.730581582797649,
      "learning_rate": 4.96127441982054e-06,
      "loss": 0.1392,
      "step": 134
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8117295304767658,
      "learning_rate": 4.960370989595665e-06,
      "loss": 0.0117,
      "step": 135
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.822915610587546,
      "learning_rate": 4.9594572266217545e-06,
      "loss": 0.3114,
      "step": 136
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.887064750003579,
      "learning_rate": 4.9585331347363115e-06,
      "loss": 0.1163,
      "step": 137
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.813427922823778,
      "learning_rate": 4.95759871782022e-06,
      "loss": 0.1726,
      "step": 138
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.198693498974976,
      "learning_rate": 4.956653979797722e-06,
      "loss": 0.3887,
      "step": 139
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.186200895249503,
      "learning_rate": 4.955698924636406e-06,
      "loss": 0.2827,
      "step": 140
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.502346836809142,
      "learning_rate": 4.954733556347192e-06,
      "loss": 0.2718,
      "step": 141
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.759273760439989,
      "learning_rate": 4.953757878984306e-06,
      "loss": 0.0605,
      "step": 142
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.083775563940655,
      "learning_rate": 4.9527718966452756e-06,
      "loss": 0.328,
      "step": 143
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.973868778063047,
      "learning_rate": 4.9517756134709005e-06,
      "loss": 0.3812,
      "step": 144
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.8341193153548887,
      "learning_rate": 4.9507690336452425e-06,
      "loss": 0.0354,
      "step": 145
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.416234317055171,
      "learning_rate": 4.949752161395606e-06,
      "loss": 0.0572,
      "step": 146
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.324931736130027,
      "learning_rate": 4.948725000992519e-06,
      "loss": 0.258,
      "step": 147
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.833407872820073,
      "learning_rate": 4.947687556749719e-06,
      "loss": 0.0653,
      "step": 148
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.186974397163941,
      "learning_rate": 4.9466398330241305e-06,
      "loss": 0.2876,
      "step": 149
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.006865624117305,
      "learning_rate": 4.945581834215848e-06,
      "loss": 0.8994,
      "step": 150
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.082199999798531,
      "learning_rate": 4.944513564768119e-06,
      "loss": 0.7066,
      "step": 151
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.091613985696003,
      "learning_rate": 4.943435029167322e-06,
      "loss": 0.3657,
      "step": 152
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.091645616692901,
      "learning_rate": 4.942346231942955e-06,
      "loss": 0.4636,
      "step": 153
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.60548066413454,
      "learning_rate": 4.941247177667606e-06,
      "loss": 0.4686,
      "step": 154
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.308989498160375,
      "learning_rate": 4.940137870956942e-06,
      "loss": 0.1011,
      "step": 155
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.307909212086813,
      "learning_rate": 4.939018316469687e-06,
      "loss": 0.1337,
      "step": 156
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.32445644400113,
      "learning_rate": 4.937888518907601e-06,
      "loss": 0.6179,
      "step": 157
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.119911315893658,
      "learning_rate": 4.936748483015462e-06,
      "loss": 0.7428,
      "step": 158
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.807390547452164,
      "learning_rate": 4.935598213581047e-06,
      "loss": 0.1299,
      "step": 159
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.948990271526885,
      "learning_rate": 4.934437715435108e-06,
      "loss": 0.8448,
      "step": 160
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.513856425965154,
      "learning_rate": 4.933266993451357e-06,
      "loss": 0.2622,
      "step": 161
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.631467484008082,
      "learning_rate": 4.93208605254644e-06,
      "loss": 0.3311,
      "step": 162
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.988981661744699,
      "learning_rate": 4.930894897679921e-06,
      "loss": 0.3608,
      "step": 163
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.093000921556436,
      "learning_rate": 4.929693533854259e-06,
      "loss": 0.7365,
      "step": 164
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.056080018556442,
      "learning_rate": 4.928481966114785e-06,
      "loss": 0.4492,
      "step": 165
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.683090324772103,
      "learning_rate": 4.927260199549689e-06,
      "loss": 0.3204,
      "step": 166
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.425554154039748,
      "learning_rate": 4.926028239289985e-06,
      "loss": 0.2467,
      "step": 167
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.33885327098014,
      "learning_rate": 4.924786090509504e-06,
      "loss": 0.1732,
      "step": 168
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.58341017408635,
      "learning_rate": 4.923533758424858e-06,
      "loss": 0.1798,
      "step": 169
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.90838859493241,
      "learning_rate": 4.922271248295436e-06,
      "loss": 0.4632,
      "step": 170
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.761841962324697,
      "learning_rate": 4.92099856542336e-06,
      "loss": 0.2391,
      "step": 171
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.079810792898726,
      "learning_rate": 4.919715715153482e-06,
      "loss": 0.1513,
      "step": 172
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.926120307168409,
      "learning_rate": 4.9184227028733526e-06,
      "loss": 0.5402,
      "step": 173
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.185515669973498,
      "learning_rate": 4.917119534013194e-06,
      "loss": 0.6567,
      "step": 174
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.49548192252073,
      "learning_rate": 4.915806214045891e-06,
      "loss": 0.443,
      "step": 175
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.866687314590607,
      "learning_rate": 4.914482748486953e-06,
      "loss": 0.4633,
      "step": 176
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.839924897344879,
      "learning_rate": 4.913149142894501e-06,
      "loss": 0.2298,
      "step": 177
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.581544137825711,
      "learning_rate": 4.91180540286924e-06,
      "loss": 0.4352,
      "step": 178
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.974583659171536,
      "learning_rate": 4.910451534054436e-06,
      "loss": 0.3001,
      "step": 179
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.895168804752402,
      "learning_rate": 4.909087542135893e-06,
      "loss": 0.4718,
      "step": 180
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.10700239632405,
      "learning_rate": 4.907713432841929e-06,
      "loss": 0.2801,
      "step": 181
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.166896838308176,
      "learning_rate": 4.906329211943349e-06,
      "loss": 0.4782,
      "step": 182
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.69390248788744,
      "learning_rate": 4.904934885253428e-06,
      "loss": 0.1194,
      "step": 183
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.701907718254184,
      "learning_rate": 4.903530458627878e-06,
      "loss": 0.3517,
      "step": 184
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.913263760748024,
      "learning_rate": 4.90211593796483e-06,
      "loss": 0.2148,
      "step": 185
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.457947770775371,
      "learning_rate": 4.9006913292048055e-06,
      "loss": 0.3311,
      "step": 186
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.808477075172256,
      "learning_rate": 4.899256638330693e-06,
      "loss": 0.6076,
      "step": 187
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.952034609239373,
      "learning_rate": 4.89781187136772e-06,
      "loss": 0.6339,
      "step": 188
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.38088281610856,
      "learning_rate": 4.896357034383436e-06,
      "loss": 0.2371,
      "step": 189
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.31560991666473,
      "learning_rate": 4.8948921334876755e-06,
      "loss": 0.526,
      "step": 190
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.023614289262058,
      "learning_rate": 4.893417174832542e-06,
      "loss": 0.1653,
      "step": 191
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6420920545956146,
      "learning_rate": 4.891932164612376e-06,
      "loss": 0.0117,
      "step": 192
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.266238213667865,
      "learning_rate": 4.890437109063733e-06,
      "loss": 0.1514,
      "step": 193
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.652913786511836,
      "learning_rate": 4.8889320144653525e-06,
      "loss": 0.1414,
      "step": 194
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.875838904783274,
      "learning_rate": 4.887416887138139e-06,
      "loss": 0.2616,
      "step": 195
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.250095519289617,
      "learning_rate": 4.885891733445127e-06,
      "loss": 0.1971,
      "step": 196
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.129772035859265,
      "learning_rate": 4.884356559791463e-06,
      "loss": 0.5796,
      "step": 197
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.590763211081543,
      "learning_rate": 4.882811372624369e-06,
      "loss": 0.2874,
      "step": 198
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0113034310365054,
      "learning_rate": 4.8812561784331255e-06,
      "loss": 0.0168,
      "step": 199
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.595816123163086,
      "learning_rate": 4.879690983749035e-06,
      "loss": 0.433,
      "step": 200
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.134903299468987,
      "learning_rate": 4.8781157951454e-06,
      "loss": 0.3736,
      "step": 201
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.270510474054294,
      "learning_rate": 4.876530619237495e-06,
      "loss": 0.3223,
      "step": 202
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.494284831575592,
      "learning_rate": 4.874935462682539e-06,
      "loss": 0.2917,
      "step": 203
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.339106223756011,
      "learning_rate": 4.873330332179663e-06,
      "loss": 0.4017,
      "step": 204
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.793226714871994,
      "learning_rate": 4.8717152344698884e-06,
      "loss": 0.5368,
      "step": 205
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.007605496887672106,
      "learning_rate": 4.870090176336094e-06,
      "loss": 0.0,
      "step": 206
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.934248171845968,
      "learning_rate": 4.86845516460299e-06,
      "loss": 0.1657,
      "step": 207
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.719543202185832,
      "learning_rate": 4.866810206137086e-06,
      "loss": 0.2729,
      "step": 208
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.018804860094635,
      "learning_rate": 4.86515530784667e-06,
      "loss": 0.3355,
      "step": 209
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.521138255140858,
      "learning_rate": 4.863490476681768e-06,
      "loss": 0.042,
      "step": 210
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.854944651393597,
      "learning_rate": 4.861815719634124e-06,
      "loss": 0.3011,
      "step": 211
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.5541484717514167,
      "learning_rate": 4.860131043737167e-06,
      "loss": 0.209,
      "step": 212
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.6168549606899765,
      "learning_rate": 4.858436456065982e-06,
      "loss": 0.3621,
      "step": 213
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.850190206951416,
      "learning_rate": 4.856731963737279e-06,
      "loss": 0.3514,
      "step": 214
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.616166023256547,
      "learning_rate": 4.855017573909367e-06,
      "loss": 0.0938,
      "step": 215
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.490503885540791,
      "learning_rate": 4.853293293782118e-06,
      "loss": 0.3176,
      "step": 216
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.815167945813124,
      "learning_rate": 4.851559130596942e-06,
      "loss": 0.2464,
      "step": 217
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.773913593911098,
      "learning_rate": 4.849815091636754e-06,
      "loss": 0.3162,
      "step": 218
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.619651615154147,
      "learning_rate": 4.8480611842259435e-06,
      "loss": 0.6477,
      "step": 219
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.136921658980324,
      "learning_rate": 4.846297415730346e-06,
      "loss": 0.4822,
      "step": 220
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.580845798620424,
      "learning_rate": 4.844523793557207e-06,
      "loss": 0.4707,
      "step": 221
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.854711454069392,
      "learning_rate": 4.842740325155159e-06,
      "loss": 0.66,
      "step": 222
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.887155990232335,
      "learning_rate": 4.8409470180141825e-06,
      "loss": 0.3607,
      "step": 223
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.030220145246533,
      "learning_rate": 4.839143879665577e-06,
      "loss": 0.4508,
      "step": 224
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.049187003686174,
      "learning_rate": 4.83733091768193e-06,
      "loss": 0.1506,
      "step": 225
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.041797169102463,
      "learning_rate": 4.835508139677086e-06,
      "loss": 0.2972,
      "step": 226
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.465887562386847,
      "learning_rate": 4.833675553306112e-06,
      "loss": 0.378,
      "step": 227
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.1022007958787015,
      "learning_rate": 4.831833166265271e-06,
      "loss": 0.3064,
      "step": 228
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.811905211809423,
      "learning_rate": 4.82998098629198e-06,
      "loss": 0.3165,
      "step": 229
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.243049098521701,
      "learning_rate": 4.8281190211647876e-06,
      "loss": 0.0759,
      "step": 230
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.741882565245166,
      "learning_rate": 4.826247278703333e-06,
      "loss": 0.329,
      "step": 231
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.282475138507657,
      "learning_rate": 4.824365766768322e-06,
      "loss": 0.2201,
      "step": 232
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.906386302632517,
      "learning_rate": 4.822474493261483e-06,
      "loss": 0.3049,
      "step": 233
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5989551440609946,
      "learning_rate": 4.820573466125544e-06,
      "loss": 0.0179,
      "step": 234
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.25735767805942,
      "learning_rate": 4.818662693344195e-06,
      "loss": 0.6544,
      "step": 235
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.251767788609671,
      "learning_rate": 4.8167421829420505e-06,
      "loss": 0.324,
      "step": 236
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.816084732299098,
      "learning_rate": 4.814811942984625e-06,
      "loss": 0.3753,
      "step": 237
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.732009703913429,
      "learning_rate": 4.812871981578291e-06,
      "loss": 0.2821,
      "step": 238
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.68447713609987,
      "learning_rate": 4.810922306870247e-06,
      "loss": 0.3349,
      "step": 239
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.716199981642604,
      "learning_rate": 4.8089629270484875e-06,
      "loss": 0.0707,
      "step": 240
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.484711425070972,
      "learning_rate": 4.806993850341763e-06,
      "loss": 0.1182,
      "step": 241
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.396622851142999,
      "learning_rate": 4.805015085019547e-06,
      "loss": 0.4418,
      "step": 242
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.017667817058326,
      "learning_rate": 4.8030266393920045e-06,
      "loss": 0.1671,
      "step": 243
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.91983323603094,
      "learning_rate": 4.801028521809951e-06,
      "loss": 0.7713,
      "step": 244
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.789966853365322,
      "learning_rate": 4.799020740664827e-06,
      "loss": 0.0282,
      "step": 245
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.612661630915982,
      "learning_rate": 4.79700330438865e-06,
      "loss": 0.4338,
      "step": 246
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.569635467076427,
      "learning_rate": 4.79497622145399e-06,
      "loss": 0.4263,
      "step": 247
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.023934283650843,
      "learning_rate": 4.792939500373928e-06,
      "loss": 0.0597,
      "step": 248
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.647239352966388,
      "learning_rate": 4.790893149702023e-06,
      "loss": 0.2972,
      "step": 249
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.792282969046761,
      "learning_rate": 4.788837178032275e-06,
      "loss": 0.5494,
      "step": 250
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.746679107152121,
      "learning_rate": 4.78677159399909e-06,
      "loss": 0.1171,
      "step": 251
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.5558840379031444,
      "learning_rate": 4.78469640627724e-06,
      "loss": 0.0948,
      "step": 252
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.412106427399646,
      "learning_rate": 4.782611623581831e-06,
      "loss": 0.3801,
      "step": 253
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.157126797197075,
      "learning_rate": 4.780517254668265e-06,
      "loss": 0.0801,
      "step": 254
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.015971106062372,
      "learning_rate": 4.778413308332204e-06,
      "loss": 0.3776,
      "step": 255
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.289155558496217,
      "learning_rate": 4.776299793409529e-06,
      "loss": 0.7922,
      "step": 256
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.287668871417681,
      "learning_rate": 4.774176718776309e-06,
      "loss": 0.467,
      "step": 257
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.8340233598408,
      "learning_rate": 4.772044093348757e-06,
      "loss": 0.802,
      "step": 258
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.366158636305297,
      "learning_rate": 4.769901926083202e-06,
      "loss": 0.6071,
      "step": 259
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.4178176671571,
      "learning_rate": 4.767750225976039e-06,
      "loss": 0.2315,
      "step": 260
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.3592580273717,
      "learning_rate": 4.765589002063702e-06,
      "loss": 0.1362,
      "step": 261
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.813344495634473,
      "learning_rate": 4.76341826342262e-06,
      "loss": 0.2995,
      "step": 262
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.8247550836571635,
      "learning_rate": 4.761238019169183e-06,
      "loss": 0.2216,
      "step": 263
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.396924292106436,
      "learning_rate": 4.759048278459698e-06,
      "loss": 0.3645,
      "step": 264
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.310670066266062,
      "learning_rate": 4.756849050490357e-06,
      "loss": 0.1536,
      "step": 265
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.613794316344615,
      "learning_rate": 4.754640344497195e-06,
      "loss": 0.4279,
      "step": 266
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.738743253823156,
      "learning_rate": 4.752422169756048e-06,
      "loss": 0.2885,
      "step": 267
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.8228132943823105,
      "learning_rate": 4.750194535582523e-06,
      "loss": 0.3673,
      "step": 268
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.195239581278578,
      "learning_rate": 4.7479574513319505e-06,
      "loss": 0.9428,
      "step": 269
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.547670694579577,
      "learning_rate": 4.745710926399347e-06,
      "loss": 0.1266,
      "step": 270
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.473438274334972,
      "learning_rate": 4.743454970219382e-06,
      "loss": 0.4765,
      "step": 271
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.042287073136041,
      "learning_rate": 4.741189592266326e-06,
      "loss": 0.3018,
      "step": 272
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.701635483230024,
      "learning_rate": 4.738914802054022e-06,
      "loss": 0.3544,
      "step": 273
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.6243596042552975,
      "learning_rate": 4.736630609135843e-06,
      "loss": 0.3046,
      "step": 274
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.318410965955888,
      "learning_rate": 4.734337023104645e-06,
      "loss": 0.2677,
      "step": 275
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.16942579840848,
      "learning_rate": 4.732034053592738e-06,
      "loss": 0.2028,
      "step": 276
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.019452450691545,
      "learning_rate": 4.729721710271834e-06,
      "loss": 0.2126,
      "step": 277
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.840398685906719,
      "learning_rate": 4.727400002853017e-06,
      "loss": 0.3519,
      "step": 278
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.6210832385563085,
      "learning_rate": 4.725068941086693e-06,
      "loss": 0.2438,
      "step": 279
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.334328732728344,
      "learning_rate": 4.722728534762554e-06,
      "loss": 0.0618,
      "step": 280
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.570676657248557,
      "learning_rate": 4.720378793709539e-06,
      "loss": 0.5638,
      "step": 281
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.47696006300138,
      "learning_rate": 4.718019727795787e-06,
      "loss": 0.2949,
      "step": 282
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.3630084592841567,
      "learning_rate": 4.715651346928598e-06,
      "loss": 0.0478,
      "step": 283
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.2526694128489595,
      "learning_rate": 4.713273661054394e-06,
      "loss": 0.2678,
      "step": 284
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.9107314539428915,
      "learning_rate": 4.710886680158674e-06,
      "loss": 0.3437,
      "step": 285
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.190842467430662,
      "learning_rate": 4.708490414265972e-06,
      "loss": 0.2772,
      "step": 286
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.424816302245013,
      "learning_rate": 4.706084873439815e-06,
      "loss": 0.7814,
      "step": 287
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.92682724473768,
      "learning_rate": 4.7036700677826876e-06,
      "loss": 0.3288,
      "step": 288
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.884500216515442,
      "learning_rate": 4.701246007435975e-06,
      "loss": 0.0483,
      "step": 289
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.8542197992694085,
      "learning_rate": 4.6988127025799376e-06,
      "loss": 0.1828,
      "step": 290
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.886115276920187,
      "learning_rate": 4.696370163433652e-06,
      "loss": 0.6249,
      "step": 291
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.612995427687164,
      "learning_rate": 4.6939184002549805e-06,
      "loss": 0.2071,
      "step": 292
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.302681354698541,
      "learning_rate": 4.691457423340524e-06,
      "loss": 0.4143,
      "step": 293
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.13586769540285,
      "learning_rate": 4.688987243025574e-06,
      "loss": 0.2545,
      "step": 294
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.086471525239894,
      "learning_rate": 4.686507869684076e-06,
      "loss": 0.3258,
      "step": 295
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.333696545760031,
      "learning_rate": 4.684019313728583e-06,
      "loss": 0.0972,
      "step": 296
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.896131188297605,
      "learning_rate": 4.681521585610212e-06,
      "loss": 0.4128,
      "step": 297
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.408882161856064,
      "learning_rate": 4.679014695818598e-06,
      "loss": 0.252,
      "step": 298
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.728542594442993,
      "learning_rate": 4.676498654881855e-06,
      "loss": 0.175,
      "step": 299
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.906921080986179,
      "learning_rate": 4.6739734733665275e-06,
      "loss": 0.0217,
      "step": 300
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.540570584710826,
      "learning_rate": 4.671439161877548e-06,
      "loss": 0.0623,
      "step": 301
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.92013933765738,
      "learning_rate": 4.6688957310581905e-06,
      "loss": 0.5718,
      "step": 302
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.779258330072647,
      "learning_rate": 4.666343191590027e-06,
      "loss": 0.2939,
      "step": 303
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.220533518311544,
      "learning_rate": 4.663781554192886e-06,
      "loss": 0.6022,
      "step": 304
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.011876826816918,
      "learning_rate": 4.661210829624802e-06,
      "loss": 0.5797,
      "step": 305
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.733390076167311,
      "learning_rate": 4.65863102868197e-06,
      "loss": 0.1799,
      "step": 306
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.6604052420012145,
      "learning_rate": 4.656042162198708e-06,
      "loss": 0.3506,
      "step": 307
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4477626238031727,
      "learning_rate": 4.653444241047403e-06,
      "loss": 0.0399,
      "step": 308
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.488359648065368,
      "learning_rate": 4.650837276138471e-06,
      "loss": 0.4692,
      "step": 309
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9915621031726336,
      "learning_rate": 4.6482212784203055e-06,
      "loss": 0.041,
      "step": 310
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.602561259333772,
      "learning_rate": 4.645596258879237e-06,
      "loss": 0.5356,
      "step": 311
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.138638923373422,
      "learning_rate": 4.642962228539485e-06,
      "loss": 0.0997,
      "step": 312
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.098187320358129,
      "learning_rate": 4.640319198463109e-06,
      "loss": 0.1054,
      "step": 313
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.054088302003616,
      "learning_rate": 4.637667179749968e-06,
      "loss": 0.3498,
      "step": 314
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.099179129744174,
      "learning_rate": 4.635006183537668e-06,
      "loss": 0.6047,
      "step": 315
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.776208928766074,
      "learning_rate": 4.6323362210015176e-06,
      "loss": 0.3915,
      "step": 316
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.500286940857086,
      "learning_rate": 4.629657303354482e-06,
      "loss": 0.1304,
      "step": 317
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.528960182141249,
      "learning_rate": 4.626969441847133e-06,
      "loss": 0.1004,
      "step": 318
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.16164129393346,
      "learning_rate": 4.624272647767607e-06,
      "loss": 0.5955,
      "step": 319
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.581160032224291,
      "learning_rate": 4.6215669324415505e-06,
      "loss": 0.1906,
      "step": 320
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.681609694889568,
      "learning_rate": 4.618852307232078e-06,
      "loss": 0.6354,
      "step": 321
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.765863479933287,
      "learning_rate": 4.616128783539725e-06,
      "loss": 0.5332,
      "step": 322
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.042645535150689,
      "learning_rate": 4.613396372802392e-06,
      "loss": 0.2459,
      "step": 323
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.304337125634598,
      "learning_rate": 4.610655086495308e-06,
      "loss": 1.121,
      "step": 324
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.138452198881182,
      "learning_rate": 4.607904936130974e-06,
      "loss": 0.1494,
      "step": 325
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.324245025193733,
      "learning_rate": 4.605145933259115e-06,
      "loss": 0.6364,
      "step": 326
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.174242307681336,
      "learning_rate": 4.602378089466637e-06,
      "loss": 0.4634,
      "step": 327
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.03943801640145,
      "learning_rate": 4.599601416377575e-06,
      "loss": 0.4802,
      "step": 328
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.0185706532664165,
      "learning_rate": 4.59681592565304e-06,
      "loss": 0.4142,
      "step": 329
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.175639231787874,
      "learning_rate": 4.59402162899118e-06,
      "loss": 0.0195,
      "step": 330
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.313862394962,
      "learning_rate": 4.5912185381271205e-06,
      "loss": 0.294,
      "step": 331
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.336083479714063,
      "learning_rate": 4.588406664832921e-06,
      "loss": 0.374,
      "step": 332
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.981125375904826,
      "learning_rate": 4.585586020917524e-06,
      "loss": 0.4972,
      "step": 333
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.240178158889226,
      "learning_rate": 4.582756618226709e-06,
      "loss": 0.0465,
      "step": 334
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.700701227100131,
      "learning_rate": 4.579918468643035e-06,
      "loss": 0.5969,
      "step": 335
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.094599838135833,
      "learning_rate": 4.577071584085797e-06,
      "loss": 0.4755,
      "step": 336
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.0528229235818785,
      "learning_rate": 4.574215976510973e-06,
      "loss": 0.249,
      "step": 337
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.9219379799941887,
      "learning_rate": 4.571351657911178e-06,
      "loss": 0.0711,
      "step": 338
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.685544307350515,
      "learning_rate": 4.568478640315606e-06,
      "loss": 0.3457,
      "step": 339
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.159724288471223,
      "learning_rate": 4.565596935789987e-06,
      "loss": 0.1901,
      "step": 340
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.037164325768962,
      "learning_rate": 4.562706556436534e-06,
      "loss": 0.0451,
      "step": 341
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.629381918758996,
      "learning_rate": 4.5598075143938855e-06,
      "loss": 0.5185,
      "step": 342
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.513771637546047,
      "learning_rate": 4.556899821837069e-06,
      "loss": 0.1977,
      "step": 343
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.112397287828078,
      "learning_rate": 4.553983490977435e-06,
      "loss": 0.2676,
      "step": 344
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.473957879645747,
      "learning_rate": 4.551058534062614e-06,
      "loss": 0.2595,
      "step": 345
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.076851365992341,
      "learning_rate": 4.5481249633764635e-06,
      "loss": 0.1709,
      "step": 346
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.886193555972856,
      "learning_rate": 4.545182791239015e-06,
      "loss": 0.1557,
      "step": 347
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.42694233431067,
      "learning_rate": 4.5422320300064246e-06,
      "loss": 0.4042,
      "step": 348
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.74570424341322,
      "learning_rate": 4.53927269207092e-06,
      "loss": 0.0405,
      "step": 349
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.293017670661733,
      "learning_rate": 4.536304789860746e-06,
      "loss": 0.584,
      "step": 350
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.763428651850486,
      "learning_rate": 4.5333283358401155e-06,
      "loss": 0.2972,
      "step": 351
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.708866766208018,
      "learning_rate": 4.530343342509158e-06,
      "loss": 0.6355,
      "step": 352
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.057864787308724,
      "learning_rate": 4.527349822403862e-06,
      "loss": 0.1121,
      "step": 353
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.983738106783992,
      "learning_rate": 4.524347788096029e-06,
      "loss": 0.3805,
      "step": 354
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.688560264204164,
      "learning_rate": 4.521337252193215e-06,
      "loss": 0.2027,
      "step": 355
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.511093942636213,
      "learning_rate": 4.518318227338682e-06,
      "loss": 0.3418,
      "step": 356
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.280807859127777,
      "learning_rate": 4.51529072621134e-06,
      "loss": 0.1645,
      "step": 357
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.3348787681936995,
      "learning_rate": 4.5122547615257e-06,
      "loss": 0.104,
      "step": 358
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.53142930159218,
      "learning_rate": 4.509210346031812e-06,
      "loss": 0.5503,
      "step": 359
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.714571628049478,
      "learning_rate": 4.506157492515223e-06,
      "loss": 0.6116,
      "step": 360
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.243656378294672,
      "learning_rate": 4.503096213796912e-06,
      "loss": 0.2708,
      "step": 361
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.7711328770370227,
      "learning_rate": 4.5000265227332455e-06,
      "loss": 0.2498,
      "step": 362
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.1023789784995754,
      "learning_rate": 4.4969484322159125e-06,
      "loss": 0.1407,
      "step": 363
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.345248609310735,
      "learning_rate": 4.493861955171885e-06,
      "loss": 0.5385,
      "step": 364
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3255143649706245,
      "learning_rate": 4.490767104563348e-06,
      "loss": 0.0422,
      "step": 365
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.319241880209145,
      "learning_rate": 4.487663893387658e-06,
      "loss": 0.4725,
      "step": 366
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.979939462164577,
      "learning_rate": 4.484552334677281e-06,
      "loss": 0.3704,
      "step": 367
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.078710371082373,
      "learning_rate": 4.481432441499741e-06,
      "loss": 0.8272,
      "step": 368
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.6832591198738704,
      "learning_rate": 4.478304226957563e-06,
      "loss": 0.3487,
      "step": 369
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.208336328289693,
      "learning_rate": 4.475167704188219e-06,
      "loss": 0.3827,
      "step": 370
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.437385879983391,
      "learning_rate": 4.472022886364073e-06,
      "loss": 0.2293,
      "step": 371
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.102030383585948,
      "learning_rate": 4.468869786692327e-06,
      "loss": 0.1278,
      "step": 372
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.879508054226713,
      "learning_rate": 4.46570841841496e-06,
      "loss": 0.3139,
      "step": 373
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.2286608123286085,
      "learning_rate": 4.462538794808683e-06,
      "loss": 0.2995,
      "step": 374
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.431765263681682,
      "learning_rate": 4.459360929184869e-06,
      "loss": 0.0296,
      "step": 375
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.00714482387539,
      "learning_rate": 4.4561748348895106e-06,
      "loss": 0.211,
      "step": 376
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.65933170528292,
      "learning_rate": 4.452980525303156e-06,
      "loss": 0.2855,
      "step": 377
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.519982650887477,
      "learning_rate": 4.449778013840854e-06,
      "loss": 0.0479,
      "step": 378
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.751698782046443,
      "learning_rate": 4.4465673139521e-06,
      "loss": 0.2207,
      "step": 379
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.255366573963133,
      "learning_rate": 4.443348439120778e-06,
      "loss": 0.4336,
      "step": 380
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.216785906484164,
      "learning_rate": 4.440121402865104e-06,
      "loss": 0.1264,
      "step": 381
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.038760908061135,
      "learning_rate": 4.436886218737568e-06,
      "loss": 0.3318,
      "step": 382
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.532180455927987,
      "learning_rate": 4.433642900324881e-06,
      "loss": 0.4184,
      "step": 383
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.688890378107093,
      "learning_rate": 4.430391461247911e-06,
      "loss": 0.2135,
      "step": 384
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.346098779390515,
      "learning_rate": 4.427131915161635e-06,
      "loss": 0.0838,
      "step": 385
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.3411319444873175,
      "learning_rate": 4.423864275755075e-06,
      "loss": 0.1006,
      "step": 386
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.296116543501431,
      "learning_rate": 4.42058855675124e-06,
      "loss": 0.1309,
      "step": 387
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.566110874838561,
      "learning_rate": 4.4173047719070745e-06,
      "loss": 0.3441,
      "step": 388
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.973664216756589,
      "learning_rate": 4.414012935013395e-06,
      "loss": 0.4388,
      "step": 389
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.992032378507416,
      "learning_rate": 4.410713059894831e-06,
      "loss": 0.1431,
      "step": 390
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.035479552028947,
      "learning_rate": 4.4074051604097755e-06,
      "loss": 0.2138,
      "step": 391
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.297789703647399,
      "learning_rate": 4.404089250450317e-06,
      "loss": 0.3008,
      "step": 392
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.871278224221298,
      "learning_rate": 4.400765343942189e-06,
      "loss": 0.5762,
      "step": 393
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.901936661106918,
      "learning_rate": 4.397433454844704e-06,
      "loss": 0.4035,
      "step": 394
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.066499413051595,
      "learning_rate": 4.3940935971507e-06,
      "loss": 0.2631,
      "step": 395
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.632133773878889,
      "learning_rate": 4.390745784886483e-06,
      "loss": 0.4767,
      "step": 396
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.890438710849303,
      "learning_rate": 4.387390032111762e-06,
      "loss": 0.4167,
      "step": 397
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.070736624872489,
      "learning_rate": 4.384026352919595e-06,
      "loss": 0.2567,
      "step": 398
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.073674937086958,
      "learning_rate": 4.380654761436329e-06,
      "loss": 0.4412,
      "step": 399
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.441670539391129,
      "learning_rate": 4.377275271821539e-06,
      "loss": 0.1941,
      "step": 400
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.839444292880805,
      "learning_rate": 4.37388789826797e-06,
      "loss": 0.1258,
      "step": 401
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.215905233733528,
      "learning_rate": 4.370492655001478e-06,
      "loss": 0.1758,
      "step": 402
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.980578141859587,
      "learning_rate": 4.3670895562809675e-06,
      "loss": 0.2898,
      "step": 403
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.282653919911857,
      "learning_rate": 4.363678616398334e-06,
      "loss": 0.3226,
      "step": 404
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.063399415301817,
      "learning_rate": 4.360259849678402e-06,
      "loss": 0.2041,
      "step": 405
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.3143282178086615,
      "learning_rate": 4.356833270478869e-06,
      "loss": 0.3638,
      "step": 406
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.231448979903139,
      "learning_rate": 4.353398893190241e-06,
      "loss": 0.4857,
      "step": 407
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.831446021230515,
      "learning_rate": 4.349956732235772e-06,
      "loss": 0.4654,
      "step": 408
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.278094457020618,
      "learning_rate": 4.346506802071406e-06,
      "loss": 0.4976,
      "step": 409
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.479196026767647,
      "learning_rate": 4.343049117185717e-06,
      "loss": 0.0527,
      "step": 410
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.678965211892643,
      "learning_rate": 4.339583692099843e-06,
      "loss": 0.1032,
      "step": 411
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.200423662359617,
      "learning_rate": 4.336110541367428e-06,
      "loss": 0.364,
      "step": 412
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.721084535125891,
      "learning_rate": 4.332629679574566e-06,
      "loss": 0.2819,
      "step": 413
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3321076682876747,
      "learning_rate": 4.329141121339731e-06,
      "loss": 0.0294,
      "step": 414
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.2734787239629024,
      "learning_rate": 4.325644881313718e-06,
      "loss": 0.0577,
      "step": 415
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.010488878826746,
      "learning_rate": 4.322140974179589e-06,
      "loss": 0.1327,
      "step": 416
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.561795114909279,
      "learning_rate": 4.3186294146526e-06,
      "loss": 0.1491,
      "step": 417
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.32636865127843,
      "learning_rate": 4.315110217480145e-06,
      "loss": 0.4162,
      "step": 418
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.775930033804638,
      "learning_rate": 4.3115833974416965e-06,
      "loss": 0.375,
      "step": 419
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.33849063232613,
      "learning_rate": 4.308048969348738e-06,
      "loss": 0.6308,
      "step": 420
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.957389569572968,
      "learning_rate": 4.304506948044704e-06,
      "loss": 0.4595,
      "step": 421
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.645789701665759,
      "learning_rate": 4.300957348404923e-06,
      "loss": 0.4273,
      "step": 422
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.466706447268282,
      "learning_rate": 4.297400185336541e-06,
      "loss": 0.2687,
      "step": 423
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.85523642991325,
      "learning_rate": 4.293835473778477e-06,
      "loss": 0.1872,
      "step": 424
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.908384405538566,
      "learning_rate": 4.290263228701346e-06,
      "loss": 0.2586,
      "step": 425
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.656316764696141,
      "learning_rate": 4.286683465107403e-06,
      "loss": 0.4652,
      "step": 426
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2745372590901196,
      "learning_rate": 4.283096198030477e-06,
      "loss": 0.0445,
      "step": 427
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.993416839295216,
      "learning_rate": 4.2795014425359095e-06,
      "loss": 0.132,
      "step": 428
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.572081918402075,
      "learning_rate": 4.275899213720493e-06,
      "loss": 0.1834,
      "step": 429
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.758615999468492,
      "learning_rate": 4.272289526712403e-06,
      "loss": 0.3461,
      "step": 430
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.646954570773951,
      "learning_rate": 4.268672396671139e-06,
      "loss": 0.1242,
      "step": 431
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.325819773592492,
      "learning_rate": 4.265047838787455e-06,
      "loss": 0.4949,
      "step": 432
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.032490642480162,
      "learning_rate": 4.261415868283304e-06,
      "loss": 0.1983,
      "step": 433
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6534399988330506,
      "learning_rate": 4.257776500411768e-06,
      "loss": 0.0399,
      "step": 434
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.7451464313728895,
      "learning_rate": 4.254129750456995e-06,
      "loss": 0.5699,
      "step": 435
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.833969279778525,
      "learning_rate": 4.250475633734135e-06,
      "loss": 0.1851,
      "step": 436
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.076512664709991,
      "learning_rate": 4.246814165589277e-06,
      "loss": 0.1268,
      "step": 437
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.261292560668046,
      "learning_rate": 4.243145361399383e-06,
      "loss": 0.2889,
      "step": 438
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.157137579278704,
      "learning_rate": 4.239469236572222e-06,
      "loss": 0.3119,
      "step": 439
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.319839335137427,
      "learning_rate": 4.235785806546313e-06,
      "loss": 0.4011,
      "step": 440
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.92563902387483,
      "learning_rate": 4.232095086790849e-06,
      "loss": 0.347,
      "step": 441
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.313336093701249,
      "learning_rate": 4.228397092805639e-06,
      "loss": 0.4254,
      "step": 442
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.785311864888827,
      "learning_rate": 4.224691840121042e-06,
      "loss": 0.182,
      "step": 443
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.099194095093149,
      "learning_rate": 4.220979344297901e-06,
      "loss": 0.2243,
      "step": 444
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.742114297468367,
      "learning_rate": 4.217259620927476e-06,
      "loss": 0.0591,
      "step": 445
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.065907225872378,
      "learning_rate": 4.213532685631384e-06,
      "loss": 0.0399,
      "step": 446
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.837702766601384,
      "learning_rate": 4.209798554061527e-06,
      "loss": 0.0264,
      "step": 447
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.624665627010382,
      "learning_rate": 4.206057241900029e-06,
      "loss": 0.4901,
      "step": 448
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.567189512612613,
      "learning_rate": 4.202308764859171e-06,
      "loss": 0.7284,
      "step": 449
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.410828152936395,
      "learning_rate": 4.198553138681324e-06,
      "loss": 0.0097,
      "step": 450
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.488429296268352,
      "learning_rate": 4.194790379138882e-06,
      "loss": 0.0699,
      "step": 451
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.055795272114857,
      "learning_rate": 4.191020502034198e-06,
      "loss": 0.6005,
      "step": 452
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.385892110136627,
      "learning_rate": 4.187243523199518e-06,
      "loss": 0.3304,
      "step": 453
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.411432914363283,
      "learning_rate": 4.1834594584969084e-06,
      "loss": 0.297,
      "step": 454
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.0139030263357505,
      "learning_rate": 4.179668323818198e-06,
      "loss": 0.39,
      "step": 455
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.27903800552657,
      "learning_rate": 4.175870135084905e-06,
      "loss": 0.0633,
      "step": 456
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.097437412252207,
      "learning_rate": 4.1720649082481735e-06,
      "loss": 0.7544,
      "step": 457
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.4988279019467114,
      "learning_rate": 4.168252659288704e-06,
      "loss": 0.2817,
      "step": 458
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.6865387002995305,
      "learning_rate": 4.164433404216689e-06,
      "loss": 0.1691,
      "step": 459
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.785666426884982,
      "learning_rate": 4.160607159071744e-06,
      "loss": 0.1772,
      "step": 460
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.605003777677964,
      "learning_rate": 4.156773939922841e-06,
      "loss": 0.6325,
      "step": 461
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.118944638083486,
      "learning_rate": 4.1529337628682375e-06,
      "loss": 0.0584,
      "step": 462
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.633787429121124,
      "learning_rate": 4.149086644035417e-06,
      "loss": 0.6426,
      "step": 463
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.196798926948138,
      "learning_rate": 4.145232599581011e-06,
      "loss": 0.4067,
      "step": 464
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.0593519368329,
      "learning_rate": 4.14137164569074e-06,
      "loss": 0.106,
      "step": 465
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.188050960706681,
      "learning_rate": 4.137503798579341e-06,
      "loss": 0.5153,
      "step": 466
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.90584308494099,
      "learning_rate": 4.1336290744905e-06,
      "loss": 0.2388,
      "step": 467
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.919577602658904,
      "learning_rate": 4.129747489696781e-06,
      "loss": 0.1712,
      "step": 468
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7047643412375464,
      "learning_rate": 4.125859060499569e-06,
      "loss": 0.0032,
      "step": 469
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.85964324921822,
      "learning_rate": 4.121963803228982e-06,
      "loss": 0.5518,
      "step": 470
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.757176675700209,
      "learning_rate": 4.118061734243824e-06,
      "loss": 0.3864,
      "step": 471
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.3281177122165255,
      "learning_rate": 4.1141528699314995e-06,
      "loss": 0.2804,
      "step": 472
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.039482913873313,
      "learning_rate": 4.110237226707953e-06,
      "loss": 0.31,
      "step": 473
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.70171702311622,
      "learning_rate": 4.1063148210176e-06,
      "loss": 0.4733,
      "step": 474
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.856575828778487,
      "learning_rate": 4.102385669333252e-06,
      "loss": 0.3028,
      "step": 475
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.473986924735284,
      "learning_rate": 4.098449788156056e-06,
      "loss": 0.2117,
      "step": 476
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.875672322214464,
      "learning_rate": 4.094507194015417e-06,
      "loss": 0.2796,
      "step": 477
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.169668352658258,
      "learning_rate": 4.090557903468935e-06,
      "loss": 0.4738,
      "step": 478
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.306268363658799,
      "learning_rate": 4.086601933102331e-06,
      "loss": 0.0848,
      "step": 479
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.081158592125722,
      "learning_rate": 4.08263929952938e-06,
      "loss": 0.205,
      "step": 480
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.955295735718712,
      "learning_rate": 4.07867001939184e-06,
      "loss": 0.0559,
      "step": 481
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.033866374746554,
      "learning_rate": 4.0746941093593815e-06,
      "loss": 0.5994,
      "step": 482
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.338877041882072,
      "learning_rate": 4.070711586129519e-06,
      "loss": 0.1351,
      "step": 483
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.948280351271775,
      "learning_rate": 4.066722466427541e-06,
      "loss": 0.123,
      "step": 484
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.938709860318418,
      "learning_rate": 4.062726767006439e-06,
      "loss": 0.0669,
      "step": 485
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.28614179863707,
      "learning_rate": 4.058724504646834e-06,
      "loss": 0.4046,
      "step": 486
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.566973312673025,
      "learning_rate": 4.054715696156914e-06,
      "loss": 0.1975,
      "step": 487
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.301531941486158,
      "learning_rate": 4.050700358372357e-06,
      "loss": 0.2406,
      "step": 488
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.607999415137768,
      "learning_rate": 4.046678508156259e-06,
      "loss": 0.3169,
      "step": 489
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.548200487083982,
      "learning_rate": 4.042650162399069e-06,
      "loss": 0.3722,
      "step": 490
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.466660903980532,
      "learning_rate": 4.038615338018515e-06,
      "loss": 0.2581,
      "step": 491
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.7044290427508364,
      "learning_rate": 4.034574051959532e-06,
      "loss": 0.0706,
      "step": 492
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.211011816575825,
      "learning_rate": 4.030526321194194e-06,
      "loss": 0.1761,
      "step": 493
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.092352291127704,
      "learning_rate": 4.026472162721636e-06,
      "loss": 0.3531,
      "step": 494
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.452091851505345,
      "learning_rate": 4.022411593567992e-06,
      "loss": 0.1989,
      "step": 495
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.931724550258274,
      "learning_rate": 4.018344630786318e-06,
      "loss": 0.1829,
      "step": 496
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.408700014643758,
      "learning_rate": 4.0142712914565186e-06,
      "loss": 0.4181,
      "step": 497
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.460201198409054,
      "learning_rate": 4.0101915926852795e-06,
      "loss": 0.2097,
      "step": 498
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.572793789141728,
      "learning_rate": 4.006105551605995e-06,
      "loss": 0.1717,
      "step": 499
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.778938028021749,
      "learning_rate": 4.002013185378694e-06,
      "loss": 0.1805,
      "step": 500
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.484465596129168,
      "learning_rate": 3.997914511189968e-06,
      "loss": 0.5012,
      "step": 501
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.424697782087838,
      "learning_rate": 3.993809546252901e-06,
      "loss": 0.408,
      "step": 502
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.074791343992947,
      "learning_rate": 3.989698307806995e-06,
      "loss": 0.498,
      "step": 503
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.072288538307756,
      "learning_rate": 3.9855808131181e-06,
      "loss": 0.4073,
      "step": 504
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.087831955935733,
      "learning_rate": 3.981457079478341e-06,
      "loss": 0.1355,
      "step": 505
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.2255848156211115,
      "learning_rate": 3.9773271242060405e-06,
      "loss": 0.3494,
      "step": 506
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.962668481844557,
      "learning_rate": 3.973190964645655e-06,
      "loss": 0.1183,
      "step": 507
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.452887706122831,
      "learning_rate": 3.969048618167693e-06,
      "loss": 0.0991,
      "step": 508
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.159313342804909,
      "learning_rate": 3.964900102168647e-06,
      "loss": 0.4468,
      "step": 509
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.423591477189401,
      "learning_rate": 3.9607454340709215e-06,
      "loss": 0.3173,
      "step": 510
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.243627145868191,
      "learning_rate": 3.956584631322755e-06,
      "loss": 0.2281,
      "step": 511
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.854213014888493,
      "learning_rate": 3.952417711398151e-06,
      "loss": 0.3971,
      "step": 512
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.380350035946153,
      "learning_rate": 3.948244691796803e-06,
      "loss": 0.5623,
      "step": 513
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.3256107563167125,
      "learning_rate": 3.944065590044021e-06,
      "loss": 0.2549,
      "step": 514
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.693745296740134,
      "learning_rate": 3.939880423690657e-06,
      "loss": 0.1227,
      "step": 515
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.590973328184677,
      "learning_rate": 3.935689210313036e-06,
      "loss": 0.3157,
      "step": 516
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.52645929549164,
      "learning_rate": 3.931491967512872e-06,
      "loss": 0.5785,
      "step": 517
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.141805538695314,
      "learning_rate": 3.927288712917209e-06,
      "loss": 0.185,
      "step": 518
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.716130126233855,
      "learning_rate": 3.92307946417833e-06,
      "loss": 0.3371,
      "step": 519
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.426152942224686,
      "learning_rate": 3.918864238973697e-06,
      "loss": 0.5731,
      "step": 520
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.926517127155523,
      "learning_rate": 3.91464305500587e-06,
      "loss": 0.2902,
      "step": 521
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.19851060692581,
      "learning_rate": 3.910415930002433e-06,
      "loss": 0.4182,
      "step": 522
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.025790646729712,
      "learning_rate": 3.906182881715922e-06,
      "loss": 0.1609,
      "step": 523
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.634598912990908,
      "learning_rate": 3.901943927923745e-06,
      "loss": 0.2222,
      "step": 524
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.7465537448124984,
      "learning_rate": 3.897699086428117e-06,
      "loss": 0.2473,
      "step": 525
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.65937294926426,
      "learning_rate": 3.893448375055973e-06,
      "loss": 0.1741,
      "step": 526
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.670711371192274,
      "learning_rate": 3.889191811658907e-06,
      "loss": 0.1265,
      "step": 527
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.077946496758454,
      "learning_rate": 3.884929414113082e-06,
      "loss": 0.3901,
      "step": 528
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.7458163766332184,
      "learning_rate": 3.880661200319168e-06,
      "loss": 0.197,
      "step": 529
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.553998108394142,
      "learning_rate": 3.876387188202258e-06,
      "loss": 0.3539,
      "step": 530
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.285249920646573,
      "learning_rate": 3.872107395711799e-06,
      "loss": 0.5698,
      "step": 531
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.359354002441556,
      "learning_rate": 3.867821840821509e-06,
      "loss": 0.221,
      "step": 532
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.315099782889805,
      "learning_rate": 3.863530541529313e-06,
      "loss": 0.2987,
      "step": 533
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.926422263998903,
      "learning_rate": 3.859233515857253e-06,
      "loss": 0.7797,
      "step": 534
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.8648743772632983,
      "learning_rate": 3.854930781851426e-06,
      "loss": 0.0488,
      "step": 535
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.375334134048924,
      "learning_rate": 3.8506223575819e-06,
      "loss": 0.4402,
      "step": 536
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.648127081288748,
      "learning_rate": 3.846308261142639e-06,
      "loss": 0.4045,
      "step": 537
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.550294327039207,
      "learning_rate": 3.84198851065143e-06,
      "loss": 0.2589,
      "step": 538
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.057065409798057,
      "learning_rate": 3.837663124249803e-06,
      "loss": 0.0294,
      "step": 539
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.565796233190705,
      "learning_rate": 3.833332120102961e-06,
      "loss": 0.1514,
      "step": 540
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.971535339837574,
      "learning_rate": 3.828995516399695e-06,
      "loss": 0.1534,
      "step": 541
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.567438704082053,
      "learning_rate": 3.824653331352316e-06,
      "loss": 0.2992,
      "step": 542
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.496221480640524,
      "learning_rate": 3.820305583196571e-06,
      "loss": 0.1934,
      "step": 543
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.530072411782659,
      "learning_rate": 3.815952290191575e-06,
      "loss": 0.2904,
      "step": 544
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.382579450480106,
      "learning_rate": 3.8115934706197248e-06,
      "loss": 0.2244,
      "step": 545
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.891447702931509,
      "learning_rate": 3.807229142786631e-06,
      "loss": 0.3195,
      "step": 546
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.021732382181241,
      "learning_rate": 3.8028593250210337e-06,
      "loss": 0.2305,
      "step": 547
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.649333461072385,
      "learning_rate": 3.798484035674732e-06,
      "loss": 0.3756,
      "step": 548
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.815812286121932,
      "learning_rate": 3.7941032931225007e-06,
      "loss": 0.3032,
      "step": 549
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.376647970732145,
      "learning_rate": 3.789717115762019e-06,
      "loss": 0.0415,
      "step": 550
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.282498886993564,
      "learning_rate": 3.7853255220137885e-06,
      "loss": 0.1783,
      "step": 551
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.842565982766671,
      "learning_rate": 3.78092853032106e-06,
      "loss": 0.4399,
      "step": 552
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.064656925923377,
      "learning_rate": 3.7765261591497502e-06,
      "loss": 0.3379,
      "step": 553
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.837069722441042,
      "learning_rate": 3.7721184269883735e-06,
      "loss": 0.3289,
      "step": 554
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.809046998862238,
      "learning_rate": 3.7677053523479534e-06,
      "loss": 0.1118,
      "step": 555
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.81892192111834,
      "learning_rate": 3.763286953761952e-06,
      "loss": 0.4513,
      "step": 556
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.667934027715425,
      "learning_rate": 3.758863249786191e-06,
      "loss": 0.7158,
      "step": 557
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.750197579753873,
      "learning_rate": 3.754434258998772e-06,
      "loss": 0.5956,
      "step": 558
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.725275627094032,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.7041,
      "step": 559
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.155266207947589,
      "learning_rate": 3.745560491412305e-06,
      "loss": 0.1983,
      "step": 560
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.481014106894056,
      "learning_rate": 3.7411157518801622e-06,
      "loss": 0.2349,
      "step": 561
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.42416083220858,
      "learning_rate": 3.7366658000700164e-06,
      "loss": 0.1869,
      "step": 562
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.608690310373774,
      "learning_rate": 3.732210654670201e-06,
      "loss": 0.1505,
      "step": 563
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.146990606032226,
      "learning_rate": 3.7277503343908627e-06,
      "loss": 0.1055,
      "step": 564
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.1348605843047,
      "learning_rate": 3.72328485796388e-06,
      "loss": 0.4122,
      "step": 565
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.165249616737636,
      "learning_rate": 3.718814244142784e-06,
      "loss": 0.4394,
      "step": 566
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.344305355370828,
      "learning_rate": 3.714338511702683e-06,
      "loss": 0.0903,
      "step": 567
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.450327094149253,
      "learning_rate": 3.709857679440182e-06,
      "loss": 0.3096,
      "step": 568
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.598028498160248,
      "learning_rate": 3.705371766173303e-06,
      "loss": 0.9059,
      "step": 569
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.144789465998734,
      "learning_rate": 3.700880790741405e-06,
      "loss": 0.5633,
      "step": 570
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.592664427333508,
      "learning_rate": 3.69638477200511e-06,
      "loss": 0.6696,
      "step": 571
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.551634999264025,
      "learning_rate": 3.691883728846216e-06,
      "loss": 0.3798,
      "step": 572
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.114513904125447,
      "learning_rate": 3.6873776801676265e-06,
      "loss": 0.5475,
      "step": 573
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.077178015949125,
      "learning_rate": 3.6828666448932615e-06,
      "loss": 0.4892,
      "step": 574
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.7550479335042,
      "learning_rate": 3.6783506419679878e-06,
      "loss": 0.0305,
      "step": 575
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.619133865912793,
      "learning_rate": 3.6738296903575303e-06,
      "loss": 0.5245,
      "step": 576
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.816562856266707,
      "learning_rate": 3.669303809048401e-06,
      "loss": 0.5877,
      "step": 577
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.058456488209786,
      "learning_rate": 3.6647730170478124e-06,
      "loss": 0.3524,
      "step": 578
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.488515837473997,
      "learning_rate": 3.6602373333836004e-06,
      "loss": 0.5515,
      "step": 579
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.008600428866071,
      "learning_rate": 3.655696777104146e-06,
      "loss": 0.3129,
      "step": 580
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.694389915349598,
      "learning_rate": 3.651151367278291e-06,
      "loss": 0.4325,
      "step": 581
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3934275212234715,
      "learning_rate": 3.646601122995263e-06,
      "loss": 0.0488,
      "step": 582
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.194649585369792,
      "learning_rate": 3.6420460633645904e-06,
      "loss": 0.131,
      "step": 583
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.970086451801307,
      "learning_rate": 3.637486207516027e-06,
      "loss": 0.1761,
      "step": 584
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.756777151540958,
      "learning_rate": 3.6329215745994656e-06,
      "loss": 0.2108,
      "step": 585
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.328596914440564,
      "learning_rate": 3.6283521837848655e-06,
      "loss": 0.0486,
      "step": 586
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.35619647848426,
      "learning_rate": 3.623778054262165e-06,
      "loss": 0.2675,
      "step": 587
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.5291643855885395,
      "learning_rate": 3.619199205241204e-06,
      "loss": 0.3329,
      "step": 588
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.11179299587398,
      "learning_rate": 3.614615655951641e-06,
      "loss": 0.1971,
      "step": 589
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.822432696361728,
      "learning_rate": 3.6100274256428773e-06,
      "loss": 0.2352,
      "step": 590
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.646554973507976,
      "learning_rate": 3.605434533583971e-06,
      "loss": 0.254,
      "step": 591
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.059289770525368,
      "learning_rate": 3.6008369990635583e-06,
      "loss": 0.2312,
      "step": 592
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.808202665546631,
      "learning_rate": 3.596234841389771e-06,
      "loss": 0.2267,
      "step": 593
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.383365408355296,
      "learning_rate": 3.5916280798901604e-06,
      "loss": 0.4627,
      "step": 594
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.287491311888257,
      "learning_rate": 3.5870167339116075e-06,
      "loss": 0.3406,
      "step": 595
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.794404112465292,
      "learning_rate": 3.58240082282025e-06,
      "loss": 0.2973,
      "step": 596
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.69207379810026,
      "learning_rate": 3.577780366001396e-06,
      "loss": 0.463,
      "step": 597
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.941540550418985,
      "learning_rate": 3.5731553828594446e-06,
      "loss": 0.3988,
      "step": 598
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.707521495510718,
      "learning_rate": 3.5685258928178054e-06,
      "loss": 0.4192,
      "step": 599
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.236522839387826,
      "learning_rate": 3.5638919153188125e-06,
      "loss": 0.5069,
      "step": 600
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.299558426797743,
      "learning_rate": 3.5592534698236477e-06,
      "loss": 0.1596,
      "step": 601
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0906154281421703,
      "learning_rate": 3.554610575812257e-06,
      "loss": 0.0426,
      "step": 602
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.0012822369296657781,
      "learning_rate": 3.5499632527832683e-06,
      "loss": 0.0,
      "step": 603
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.57191123713371,
      "learning_rate": 3.54531152025391e-06,
      "loss": 0.2251,
      "step": 604
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.31522755412989,
      "learning_rate": 3.54065539775993e-06,
      "loss": 0.3553,
      "step": 605
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.726450211388549,
      "learning_rate": 3.535994904855509e-06,
      "loss": 0.3344,
      "step": 606
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.912298528059726,
      "learning_rate": 3.5313300611131874e-06,
      "loss": 0.4693,
      "step": 607
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.197976877169555,
      "learning_rate": 3.526660886123773e-06,
      "loss": 0.3403,
      "step": 608
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.43994159419505,
      "learning_rate": 3.521987399496266e-06,
      "loss": 0.0843,
      "step": 609
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.22676489269629,
      "learning_rate": 3.517309620857773e-06,
      "loss": 0.0739,
      "step": 610
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.4999483546626164,
      "learning_rate": 3.5126275698534255e-06,
      "loss": 0.413,
      "step": 611
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.044424306425061,
      "learning_rate": 3.507941266146299e-06,
      "loss": 0.2406,
      "step": 612
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.330248413703988,
      "learning_rate": 3.5032507294173275e-06,
      "loss": 0.5586,
      "step": 613
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.962067530930888,
      "learning_rate": 3.4985559793652223e-06,
      "loss": 0.152,
      "step": 614
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.772229224355687,
      "learning_rate": 3.4938570357063906e-06,
      "loss": 0.1259,
      "step": 615
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8995872335347888,
      "learning_rate": 3.489153918174849e-06,
      "loss": 0.0429,
      "step": 616
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.154732299238771,
      "learning_rate": 3.484446646522146e-06,
      "loss": 0.2469,
      "step": 617
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.221432299199805,
      "learning_rate": 3.4797352405172735e-06,
      "loss": 0.1814,
      "step": 618
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.765988374681921,
      "learning_rate": 3.475019719946588e-06,
      "loss": 0.5377,
      "step": 619
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.284518969277898,
      "learning_rate": 3.4703001046137253e-06,
      "loss": 0.4524,
      "step": 620
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.491568826283332,
      "learning_rate": 3.4655764143395163e-06,
      "loss": 0.125,
      "step": 621
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.229204050050088,
      "learning_rate": 3.4608486689619087e-06,
      "loss": 0.6578,
      "step": 622
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.420714459405207,
      "learning_rate": 3.4561168883358766e-06,
      "loss": 0.4392,
      "step": 623
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.791198748158608,
      "learning_rate": 3.4513810923333436e-06,
      "loss": 0.2087,
      "step": 624
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.0539738315514935,
      "learning_rate": 3.4466413008430943e-06,
      "loss": 0.204,
      "step": 625
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.419946566523178,
      "learning_rate": 3.4418975337706957e-06,
      "loss": 0.431,
      "step": 626
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.696142933306922,
      "learning_rate": 3.4371498110384083e-06,
      "loss": 0.6483,
      "step": 627
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.394725180088116,
      "learning_rate": 3.4323981525851075e-06,
      "loss": 0.2136,
      "step": 628
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.8987016794037,
      "learning_rate": 3.4276425783661945e-06,
      "loss": 0.3772,
      "step": 629
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.807960952942797,
      "learning_rate": 3.4228831083535185e-06,
      "loss": 0.267,
      "step": 630
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.267958900346725,
      "learning_rate": 3.4181197625352874e-06,
      "loss": 0.131,
      "step": 631
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.30303447021465,
      "learning_rate": 3.4133525609159883e-06,
      "loss": 0.1255,
      "step": 632
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.253820530760357,
      "learning_rate": 3.408581523516299e-06,
      "loss": 0.3077,
      "step": 633
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6422984979801796,
      "learning_rate": 3.403806670373008e-06,
      "loss": 0.0678,
      "step": 634
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.2746809112730055,
      "learning_rate": 3.399028021538929e-06,
      "loss": 0.3641,
      "step": 635
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.7660628038378805,
      "learning_rate": 3.3942455970828146e-06,
      "loss": 0.3673,
      "step": 636
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.259821689491524,
      "learning_rate": 3.3894594170892747e-06,
      "loss": 0.216,
      "step": 637
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.251441711033382,
      "learning_rate": 3.3846695016586915e-06,
      "loss": 0.1541,
      "step": 638
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.74081112796774,
      "learning_rate": 3.3798758709071354e-06,
      "loss": 0.5547,
      "step": 639
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.785835410986747,
      "learning_rate": 3.375078544966278e-06,
      "loss": 0.1474,
      "step": 640
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.74835028918615,
      "learning_rate": 3.3702775439833126e-06,
      "loss": 0.0708,
      "step": 641
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.46243647631831,
      "learning_rate": 3.3654728881208607e-06,
      "loss": 0.458,
      "step": 642
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.001441840071807,
      "learning_rate": 3.360664597556901e-06,
      "loss": 0.4055,
      "step": 643
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.33990609928784,
      "learning_rate": 3.3558526924846695e-06,
      "loss": 0.4857,
      "step": 644
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.910268419120786,
      "learning_rate": 3.3510371931125875e-06,
      "loss": 0.4183,
      "step": 645
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.582500627241547,
      "learning_rate": 3.346218119664166e-06,
      "loss": 0.4943,
      "step": 646
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.771240807572005,
      "learning_rate": 3.3413954923779313e-06,
      "loss": 0.1844,
      "step": 647
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.9605775327217363,
      "learning_rate": 3.33656933150733e-06,
      "loss": 0.2458,
      "step": 648
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.461514250287404,
      "learning_rate": 3.3317396573206528e-06,
      "loss": 0.1039,
      "step": 649
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.6908110644522765,
      "learning_rate": 3.326906490100941e-06,
      "loss": 0.5968,
      "step": 650
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.353282102445988,
      "learning_rate": 3.3220698501459082e-06,
      "loss": 0.0937,
      "step": 651
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.8408826025195895,
      "learning_rate": 3.3172297577678515e-06,
      "loss": 0.2298,
      "step": 652
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.534472852360349,
      "learning_rate": 3.3123862332935674e-06,
      "loss": 0.2475,
      "step": 653
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.577137498820138,
      "learning_rate": 3.3075392970642655e-06,
      "loss": 0.0879,
      "step": 654
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.33067748382168,
      "learning_rate": 3.3026889694354845e-06,
      "loss": 0.1781,
      "step": 655
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.428135552177555,
      "learning_rate": 3.297835270777005e-06,
      "loss": 0.6271,
      "step": 656
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.577155000695369,
      "learning_rate": 3.2929782214727657e-06,
      "loss": 0.0708,
      "step": 657
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.735077192280396,
      "learning_rate": 3.2881178419207754e-06,
      "loss": 0.7341,
      "step": 658
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.902137263626237,
      "learning_rate": 3.2832541525330307e-06,
      "loss": 0.3174,
      "step": 659
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.704656332682939,
      "learning_rate": 3.2783871737354272e-06,
      "loss": 0.384,
      "step": 660
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.20714266724514,
      "learning_rate": 3.2735169259676754e-06,
      "loss": 0.4284,
      "step": 661
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.117987324182064,
      "learning_rate": 3.268643429683214e-06,
      "loss": 0.1292,
      "step": 662
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.8741767383788135,
      "learning_rate": 3.263766705349125e-06,
      "loss": 0.2135,
      "step": 663
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.504234644362891,
      "learning_rate": 3.2588867734460467e-06,
      "loss": 0.3344,
      "step": 664
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.94017168989941,
      "learning_rate": 3.254003654468088e-06,
      "loss": 0.2153,
      "step": 665
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.415277901569533,
      "learning_rate": 3.249117368922744e-06,
      "loss": 0.5916,
      "step": 666
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.828794210678742,
      "learning_rate": 3.244227937330805e-06,
      "loss": 0.284,
      "step": 667
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.107075212019553,
      "learning_rate": 3.2393353802262777e-06,
      "loss": 0.2186,
      "step": 668
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.80001101370792,
      "learning_rate": 3.234439718156292e-06,
      "loss": 0.5572,
      "step": 669
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.40616740927132,
      "learning_rate": 3.229540971681019e-06,
      "loss": 0.3547,
      "step": 670
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.9635428038572424,
      "learning_rate": 3.224639161373582e-06,
      "loss": 0.1532,
      "step": 671
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.384910519522065,
      "learning_rate": 3.2197343078199735e-06,
      "loss": 0.2449,
      "step": 672
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.827283214844652,
      "learning_rate": 3.2148264316189638e-06,
      "loss": 0.5185,
      "step": 673
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.861391449667313,
      "learning_rate": 3.2099155533820207e-06,
      "loss": 0.2988,
      "step": 674
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.31015680758051,
      "learning_rate": 3.2050016937332166e-06,
      "loss": 0.1083,
      "step": 675
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.312843258302744,
      "learning_rate": 3.2000848733091473e-06,
      "loss": 0.3997,
      "step": 676
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.8860401441663655,
      "learning_rate": 3.1951651127588403e-06,
      "loss": 0.108,
      "step": 677
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.732600215496707,
      "learning_rate": 3.190242432743673e-06,
      "loss": 0.2456,
      "step": 678
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.165008116436,
      "learning_rate": 3.185316853937281e-06,
      "loss": 0.3988,
      "step": 679
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.344600806765619,
      "learning_rate": 3.1803883970254773e-06,
      "loss": 0.2684,
      "step": 680
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.47365976724781,
      "learning_rate": 3.17545708270616e-06,
      "loss": 0.5079,
      "step": 681
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.518229330834231,
      "learning_rate": 3.1705229316892263e-06,
      "loss": 0.1876,
      "step": 682
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.8916402401892185,
      "learning_rate": 3.1655859646964882e-06,
      "loss": 0.1962,
      "step": 683
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.847496422957769,
      "learning_rate": 3.1606462024615838e-06,
      "loss": 0.1164,
      "step": 684
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.399327552754383,
      "learning_rate": 3.1557036657298906e-06,
      "loss": 0.1209,
      "step": 685
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.45055607274162,
      "learning_rate": 3.150758375258436e-06,
      "loss": 0.5516,
      "step": 686
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.256604176803525,
      "learning_rate": 3.145810351815815e-06,
      "loss": 0.2048,
      "step": 687
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4313041843335819,
      "learning_rate": 3.140859616182098e-06,
      "loss": 0.0273,
      "step": 688
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.83878147958253,
      "learning_rate": 3.1359061891487473e-06,
      "loss": 0.0585,
      "step": 689
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.938524357249688,
      "learning_rate": 3.1309500915185265e-06,
      "loss": 0.0961,
      "step": 690
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.987872468147123,
      "learning_rate": 3.125991344105417e-06,
      "loss": 0.41,
      "step": 691
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.726756165094635,
      "learning_rate": 3.1210299677345256e-06,
      "loss": 0.2588,
      "step": 692
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.351286182622627,
      "learning_rate": 3.1160659832420033e-06,
      "loss": 0.3137,
      "step": 693
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.102458971034878,
      "learning_rate": 3.111099411474951e-06,
      "loss": 0.1375,
      "step": 694
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9289523398289603,
      "learning_rate": 3.106130273291338e-06,
      "loss": 0.2306,
      "step": 695
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.286081124614515,
      "learning_rate": 3.1011585895599083e-06,
      "loss": 0.2769,
      "step": 696
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7685540670661409,
      "learning_rate": 3.0961843811601016e-06,
      "loss": 0.006,
      "step": 697
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.358175160084128,
      "learning_rate": 3.091207668981955e-06,
      "loss": 0.1439,
      "step": 698
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.539215633007578,
      "learning_rate": 3.0862284739260247e-06,
      "loss": 0.3104,
      "step": 699
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.370554511640458,
      "learning_rate": 3.081246816903292e-06,
      "loss": 0.726,
      "step": 700
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.460448905063149,
      "learning_rate": 3.076262718835077e-06,
      "loss": 0.0923,
      "step": 701
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6729677562799097,
      "learning_rate": 3.0712762006529533e-06,
      "loss": 0.1067,
      "step": 702
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.322218678118633,
      "learning_rate": 3.0662872832986574e-06,
      "loss": 0.109,
      "step": 703
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.078091269771491,
      "learning_rate": 3.0612959877240024e-06,
      "loss": 0.0223,
      "step": 704
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.485777168486294,
      "learning_rate": 3.056302334890786e-06,
      "loss": 0.427,
      "step": 705
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.579777815757636,
      "learning_rate": 3.051306345770711e-06,
      "loss": 0.2521,
      "step": 706
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.687302345848805,
      "learning_rate": 3.0463080413452856e-06,
      "loss": 0.4441,
      "step": 707
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.01544559505207,
      "learning_rate": 3.0413074426057472e-06,
      "loss": 0.4639,
      "step": 708
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.927990817661355,
      "learning_rate": 3.0363045705529638e-06,
      "loss": 0.2244,
      "step": 709
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.664134707205871,
      "learning_rate": 3.0312994461973544e-06,
      "loss": 0.6734,
      "step": 710
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.311725542839963,
      "learning_rate": 3.0262920905587945e-06,
      "loss": 0.2679,
      "step": 711
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4334897203665629,
      "learning_rate": 3.021282524666531e-06,
      "loss": 0.0143,
      "step": 712
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.786103845052255,
      "learning_rate": 3.016270769559094e-06,
      "loss": 0.3318,
      "step": 713
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.136857532049963,
      "learning_rate": 3.011256846284206e-06,
      "loss": 0.0677,
      "step": 714
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.6911212412125805,
      "learning_rate": 3.006240775898696e-06,
      "loss": 0.3064,
      "step": 715
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.80582893891379,
      "learning_rate": 3.001222579468411e-06,
      "loss": 0.3059,
      "step": 716
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5434229572582017,
      "learning_rate": 2.9962022780681237e-06,
      "loss": 0.0193,
      "step": 717
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.66356231649076,
      "learning_rate": 2.991179892781451e-06,
      "loss": 0.4298,
      "step": 718
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.296739365210787,
      "learning_rate": 2.986155444700758e-06,
      "loss": 0.6059,
      "step": 719
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.80721612064858,
      "learning_rate": 2.981128954927075e-06,
      "loss": 0.6935,
      "step": 720
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.20388174424918,
      "learning_rate": 2.9761004445700063e-06,
      "loss": 0.1102,
      "step": 721
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.274031807768026,
      "learning_rate": 2.9710699347476415e-06,
      "loss": 0.4478,
      "step": 722
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.294650620703173,
      "learning_rate": 2.966037446586467e-06,
      "loss": 0.2077,
      "step": 723
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.823860984468512,
      "learning_rate": 2.9610030012212783e-06,
      "loss": 0.1845,
      "step": 724
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.146084055123757,
      "learning_rate": 2.955966619795091e-06,
      "loss": 0.3648,
      "step": 725
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.51193924679981,
      "learning_rate": 2.95092832345905e-06,
      "loss": 0.4524,
      "step": 726
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.0467729600301205,
      "learning_rate": 2.945888133372343e-06,
      "loss": 0.0294,
      "step": 727
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.921526768030892,
      "learning_rate": 2.9408460707021114e-06,
      "loss": 0.1625,
      "step": 728
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.797477748058581,
      "learning_rate": 2.93580215662336e-06,
      "loss": 0.2835,
      "step": 729
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.7187894931266885,
      "learning_rate": 2.930756412318869e-06,
      "loss": 0.0789,
      "step": 730
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.4511765726658745,
      "learning_rate": 2.925708858979106e-06,
      "loss": 0.052,
      "step": 731
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.717261840262642,
      "learning_rate": 2.9206595178021337e-06,
      "loss": 0.5213,
      "step": 732
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.412682681220795,
      "learning_rate": 2.9156084099935256e-06,
      "loss": 0.2256,
      "step": 733
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.429462148257445,
      "learning_rate": 2.9105555567662724e-06,
      "loss": 0.2526,
      "step": 734
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.758202267029308,
      "learning_rate": 2.9055009793406972e-06,
      "loss": 0.2356,
      "step": 735
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.453784973534456,
      "learning_rate": 2.900444698944362e-06,
      "loss": 0.2218,
      "step": 736
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.2798803960298,
      "learning_rate": 2.895386736811982e-06,
      "loss": 0.411,
      "step": 737
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.179331242436376,
      "learning_rate": 2.8903271141853346e-06,
      "loss": 0.351,
      "step": 738
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.694703687587645,
      "learning_rate": 2.885265852313171e-06,
      "loss": 0.4684,
      "step": 739
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.552161224629805,
      "learning_rate": 2.8802029724511265e-06,
      "loss": 0.0449,
      "step": 740
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.193204040919566,
      "learning_rate": 2.8751384958616318e-06,
      "loss": 0.1399,
      "step": 741
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1645583263001535,
      "learning_rate": 2.8700724438138237e-06,
      "loss": 0.0175,
      "step": 742
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.745994284258654,
      "learning_rate": 2.8650048375834534e-06,
      "loss": 0.1841,
      "step": 743
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.503711114693864,
      "learning_rate": 2.8599356984528014e-06,
      "loss": 0.433,
      "step": 744
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.274015445144218,
      "learning_rate": 2.8548650477105843e-06,
      "loss": 0.2573,
      "step": 745
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.404104589760424,
      "learning_rate": 2.8497929066518686e-06,
      "loss": 0.1154,
      "step": 746
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.134366095017097,
      "learning_rate": 2.8447192965779778e-06,
      "loss": 0.1819,
      "step": 747
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.7412682843096325,
      "learning_rate": 2.8396442387964075e-06,
      "loss": 0.2815,
      "step": 748
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.955670728667691,
      "learning_rate": 2.8345677546207283e-06,
      "loss": 0.2243,
      "step": 749
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.948314863134504,
      "learning_rate": 2.829489865370507e-06,
      "loss": 0.3371,
      "step": 750
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.02287740711825,
      "learning_rate": 2.8244105923712074e-06,
      "loss": 0.4982,
      "step": 751
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6725793941111144,
      "learning_rate": 2.819329956954106e-06,
      "loss": 0.0248,
      "step": 752
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.534539813275994,
      "learning_rate": 2.8142479804562e-06,
      "loss": 0.279,
      "step": 753
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.853079625904447,
      "learning_rate": 2.809164684220121e-06,
      "loss": 0.2999,
      "step": 754
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.246319961517067,
      "learning_rate": 2.8040800895940395e-06,
      "loss": 0.485,
      "step": 755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.0004816702790924369,
      "learning_rate": 2.7989942179315832e-06,
      "loss": 0.0,
      "step": 756
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.5227088143646,
      "learning_rate": 2.7939070905917378e-06,
      "loss": 0.0986,
      "step": 757
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.858583930974236,
      "learning_rate": 2.7888187289387675e-06,
      "loss": 0.3007,
      "step": 758
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.512850116308536,
      "learning_rate": 2.7837291543421164e-06,
      "loss": 0.4299,
      "step": 759
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.078627982855776,
      "learning_rate": 2.778638388176324e-06,
      "loss": 0.0617,
      "step": 760
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.690378856516007,
      "learning_rate": 2.773546451820936e-06,
      "loss": 0.5397,
      "step": 761
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.9189220302761445,
      "learning_rate": 2.768453366660408e-06,
      "loss": 0.139,
      "step": 762
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.124384978713398,
      "learning_rate": 2.7633591540840243e-06,
      "loss": 0.0565,
      "step": 763
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.884950030030152,
      "learning_rate": 2.7582638354858023e-06,
      "loss": 0.5086,
      "step": 764
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.9719791964189675,
      "learning_rate": 2.753167432264404e-06,
      "loss": 0.153,
      "step": 765
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.941511201055096,
      "learning_rate": 2.748069965823047e-06,
      "loss": 0.5118,
      "step": 766
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.274503111374512,
      "learning_rate": 2.7429714575694145e-06,
      "loss": 0.1259,
      "step": 767
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.626365578898379,
      "learning_rate": 2.7378719289155637e-06,
      "loss": 0.0231,
      "step": 768
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.24206038669074,
      "learning_rate": 2.7327714012778385e-06,
      "loss": 0.2318,
      "step": 769
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.074652402506702,
      "learning_rate": 2.7276698960767774e-06,
      "loss": 0.0255,
      "step": 770
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2154239395202375,
      "learning_rate": 2.7225674347370248e-06,
      "loss": 0.0465,
      "step": 771
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.499591384985523,
      "learning_rate": 2.7174640386872394e-06,
      "loss": 0.0556,
      "step": 772
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.238796086510255,
      "learning_rate": 2.712359729360007e-06,
      "loss": 0.3998,
      "step": 773
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.01275841186851,
      "learning_rate": 2.707254528191749e-06,
      "loss": 0.1265,
      "step": 774
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.4573307606659505,
      "learning_rate": 2.70214845662263e-06,
      "loss": 0.439,
      "step": 775
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.171731537733631,
      "learning_rate": 2.697041536096472e-06,
      "loss": 0.2573,
      "step": 776
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.5856258730395885,
      "learning_rate": 2.691933788060662e-06,
      "loss": 0.2674,
      "step": 777
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.56001199506063,
      "learning_rate": 2.686825233966061e-06,
      "loss": 0.4671,
      "step": 778
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.760523751639689,
      "learning_rate": 2.6817158952669177e-06,
      "loss": 0.3438,
      "step": 779
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.095872121035828,
      "learning_rate": 2.6766057934207734e-06,
      "loss": 0.1716,
      "step": 780
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.796274818557782,
      "learning_rate": 2.6714949498883745e-06,
      "loss": 0.6519,
      "step": 781
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.858535994838844,
      "learning_rate": 2.6663833861335857e-06,
      "loss": 0.2606,
      "step": 782
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.965167711461031,
      "learning_rate": 2.6612711236232915e-06,
      "loss": 0.1159,
      "step": 783
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.785247889961181,
      "learning_rate": 2.6561581838273137e-06,
      "loss": 0.297,
      "step": 784
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.4581581251424875,
      "learning_rate": 2.651044588218318e-06,
      "loss": 0.2434,
      "step": 785
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.397183452172526,
      "learning_rate": 2.645930358271725e-06,
      "loss": 0.0306,
      "step": 786
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.3505465939643,
      "learning_rate": 2.640815515465618e-06,
      "loss": 0.7295,
      "step": 787
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.947761630417348,
      "learning_rate": 2.635700081280655e-06,
      "loss": 0.1265,
      "step": 788
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.494343570607112,
      "learning_rate": 2.6305840771999764e-06,
      "loss": 0.2842,
      "step": 789
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.981692874105951,
      "learning_rate": 2.6254675247091182e-06,
      "loss": 0.2263,
      "step": 790
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.334599604098417,
      "learning_rate": 2.6203504452959174e-06,
      "loss": 0.0601,
      "step": 791
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.1997284732058215,
      "learning_rate": 2.615232860450425e-06,
      "loss": 0.3632,
      "step": 792
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.6590461409884,
      "learning_rate": 2.610114791664814e-06,
      "loss": 0.278,
      "step": 793
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9268985235479245,
      "learning_rate": 2.6049962604332913e-06,
      "loss": 0.0039,
      "step": 794
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.657219060849721,
      "learning_rate": 2.599877288252003e-06,
      "loss": 0.1487,
      "step": 795
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.163294966566469,
      "learning_rate": 2.59475789661895e-06,
      "loss": 0.0782,
      "step": 796
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.695923990878327,
      "learning_rate": 2.589638107033894e-06,
      "loss": 0.1973,
      "step": 797
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.506726364024847,
      "learning_rate": 2.5845179409982667e-06,
      "loss": 0.2817,
      "step": 798
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.582582410043921,
      "learning_rate": 2.579397420015081e-06,
      "loss": 0.5584,
      "step": 799
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1432966569019563,
      "learning_rate": 2.574276565588843e-06,
      "loss": 0.0564,
      "step": 800
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.198366328815605,
      "learning_rate": 2.569155399225456e-06,
      "loss": 0.1426,
      "step": 801
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.369378538308984,
      "learning_rate": 2.5640339424321346e-06,
      "loss": 0.1068,
      "step": 802
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.453453950333066,
      "learning_rate": 2.5589122167173137e-06,
      "loss": 0.2165,
      "step": 803
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.850692533814966,
      "learning_rate": 2.553790243590556e-06,
      "loss": 0.535,
      "step": 804
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6188975654486732,
      "learning_rate": 2.548668044562465e-06,
      "loss": 0.0219,
      "step": 805
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.189834620904021,
      "learning_rate": 2.5435456411445912e-06,
      "loss": 0.2881,
      "step": 806
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.7281350070918187,
      "learning_rate": 2.5384230548493467e-06,
      "loss": 0.1371,
      "step": 807
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.953050679031332,
      "learning_rate": 2.533300307189906e-06,
      "loss": 0.3427,
      "step": 808
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.44563726176564,
      "learning_rate": 2.5281774196801274e-06,
      "loss": 0.2655,
      "step": 809
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.097336518648792,
      "learning_rate": 2.5230544138344513e-06,
      "loss": 0.2631,
      "step": 810
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.291798186983118,
      "learning_rate": 2.5179313111678203e-06,
      "loss": 0.1324,
      "step": 811
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.519882570952501,
      "learning_rate": 2.5128081331955783e-06,
      "loss": 0.11,
      "step": 812
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.156809089959282,
      "learning_rate": 2.50768490143339e-06,
      "loss": 0.5701,
      "step": 813
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.1893533360395345,
      "learning_rate": 2.5025616373971416e-06,
      "loss": 0.1588,
      "step": 814
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.055942862625886,
      "learning_rate": 2.497438362602859e-06,
      "loss": 0.0073,
      "step": 815
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.216517887397222,
      "learning_rate": 2.492315098566612e-06,
      "loss": 0.0435,
      "step": 816
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.571209889528808,
      "learning_rate": 2.487191866804422e-06,
      "loss": 0.104,
      "step": 817
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.88362440674566,
      "learning_rate": 2.482068688832181e-06,
      "loss": 0.2601,
      "step": 818
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.985565904274485,
      "learning_rate": 2.476945586165549e-06,
      "loss": 0.1496,
      "step": 819
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.572992281599657,
      "learning_rate": 2.471822580319874e-06,
      "loss": 0.1586,
      "step": 820
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.37683127501101,
      "learning_rate": 2.466699692810095e-06,
      "loss": 0.1533,
      "step": 821
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.5080739274471355,
      "learning_rate": 2.461576945150655e-06,
      "loss": 0.1901,
      "step": 822
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.258736181047285,
      "learning_rate": 2.456454358855409e-06,
      "loss": 0.2367,
      "step": 823
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.601914275166003,
      "learning_rate": 2.4513319554375363e-06,
      "loss": 0.4312,
      "step": 824
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.299126385864762,
      "learning_rate": 2.446209756409445e-06,
      "loss": 0.151,
      "step": 825
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.878069082785739,
      "learning_rate": 2.4410877832826876e-06,
      "loss": 0.5147,
      "step": 826
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.740606129145274,
      "learning_rate": 2.435966057567866e-06,
      "loss": 0.0242,
      "step": 827
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.954831580420314,
      "learning_rate": 2.4308446007745452e-06,
      "loss": 0.1391,
      "step": 828
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.0008022396197248448,
      "learning_rate": 2.425723434411158e-06,
      "loss": 0.0,
      "step": 829
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.70913423911258,
      "learning_rate": 2.4206025799849196e-06,
      "loss": 0.18,
      "step": 830
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.48847202621159,
      "learning_rate": 2.415482059001734e-06,
      "loss": 0.4761,
      "step": 831
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17612393672201163,
      "learning_rate": 2.4103618929661072e-06,
      "loss": 0.0007,
      "step": 832
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.466116937604331,
      "learning_rate": 2.4052421033810503e-06,
      "loss": 0.0284,
      "step": 833
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.58109403101183,
      "learning_rate": 2.4001227117479983e-06,
      "loss": 0.2474,
      "step": 834
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.597120022029906,
      "learning_rate": 2.3950037395667096e-06,
      "loss": 0.1241,
      "step": 835
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.072654765424927,
      "learning_rate": 2.3898852083351874e-06,
      "loss": 0.4918,
      "step": 836
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.72258925675377,
      "learning_rate": 2.3847671395495757e-06,
      "loss": 0.2196,
      "step": 837
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.889633793878003,
      "learning_rate": 2.379649554704084e-06,
      "loss": 0.2769,
      "step": 838
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.786338422639018,
      "learning_rate": 2.3745324752908826e-06,
      "loss": 0.1242,
      "step": 839
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.816598791234943,
      "learning_rate": 2.369415922800025e-06,
      "loss": 0.4444,
      "step": 840
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.793592620683167,
      "learning_rate": 2.364299918719346e-06,
      "loss": 0.2588,
      "step": 841
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.715642366127424,
      "learning_rate": 2.3591844845343828e-06,
      "loss": 0.2249,
      "step": 842
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.19279245322333,
      "learning_rate": 2.3540696417282755e-06,
      "loss": 0.3364,
      "step": 843
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.0007639544783230961,
      "learning_rate": 2.3489554117816827e-06,
      "loss": 0.0,
      "step": 844
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.660356689196195,
      "learning_rate": 2.343841816172687e-06,
      "loss": 0.3094,
      "step": 845
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.056742231194947,
      "learning_rate": 2.3387288763767097e-06,
      "loss": 0.7192,
      "step": 846
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.180834670631299,
      "learning_rate": 2.333616613866415e-06,
      "loss": 0.2911,
      "step": 847
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.732843707320632,
      "learning_rate": 2.328505050111626e-06,
      "loss": 0.1936,
      "step": 848
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.767254551541562,
      "learning_rate": 2.3233942065792274e-06,
      "loss": 0.5851,
      "step": 849
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.880672994500611,
      "learning_rate": 2.3182841047330836e-06,
      "loss": 0.046,
      "step": 850
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.668434354045177,
      "learning_rate": 2.3131747660339396e-06,
      "loss": 0.4546,
      "step": 851
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.395906460539295,
      "learning_rate": 2.3080662119393395e-06,
      "loss": 0.4166,
      "step": 852
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.432345874353768,
      "learning_rate": 2.3029584639035288e-06,
      "loss": 0.1,
      "step": 853
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.045690886127331,
      "learning_rate": 2.297851543377371e-06,
      "loss": 0.1253,
      "step": 854
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.64455022521076,
      "learning_rate": 2.292745471808252e-06,
      "loss": 0.2786,
      "step": 855
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.774363274597841,
      "learning_rate": 2.287640270639994e-06,
      "loss": 0.0335,
      "step": 856
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.037660170601043,
      "learning_rate": 2.2825359613127615e-06,
      "loss": 0.568,
      "step": 857
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.294832489960329,
      "learning_rate": 2.2774325652629765e-06,
      "loss": 0.1654,
      "step": 858
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.81186666081957,
      "learning_rate": 2.272330103923224e-06,
      "loss": 0.2634,
      "step": 859
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.888207706105006,
      "learning_rate": 2.2672285987221628e-06,
      "loss": 0.2884,
      "step": 860
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.885442044280387,
      "learning_rate": 2.262128071084437e-06,
      "loss": 0.4424,
      "step": 861
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.511127609345163,
      "learning_rate": 2.257028542430587e-06,
      "loss": 0.2785,
      "step": 862
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.313982951761123,
      "learning_rate": 2.2519300341769535e-06,
      "loss": 0.28,
      "step": 863
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.148900685130503,
      "learning_rate": 2.246832567735597e-06,
      "loss": 0.3623,
      "step": 864
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.176549725061074,
      "learning_rate": 2.2417361645141985e-06,
      "loss": 0.5615,
      "step": 865
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.953518111097834,
      "learning_rate": 2.2366408459159765e-06,
      "loss": 0.1704,
      "step": 866
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.298203798642748,
      "learning_rate": 2.2315466333395927e-06,
      "loss": 0.332,
      "step": 867
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5129001621401583,
      "learning_rate": 2.2264535481790656e-06,
      "loss": 0.0281,
      "step": 868
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.387135269350836,
      "learning_rate": 2.2213616118236762e-06,
      "loss": 0.1011,
      "step": 869
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.360321199475546,
      "learning_rate": 2.2162708456578853e-06,
      "loss": 0.2468,
      "step": 870
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.904042956829158,
      "learning_rate": 2.2111812710612333e-06,
      "loss": 0.2183,
      "step": 871
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.389410119218008,
      "learning_rate": 2.2060929094082635e-06,
      "loss": 0.3332,
      "step": 872
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.497967951531669,
      "learning_rate": 2.2010057820684176e-06,
      "loss": 0.1991,
      "step": 873
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.38982233765663,
      "learning_rate": 2.1959199104059613e-06,
      "loss": 0.3669,
      "step": 874
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.546885487128155,
      "learning_rate": 2.19083531577988e-06,
      "loss": 0.1043,
      "step": 875
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.0072057299867465,
      "learning_rate": 2.185752019543801e-06,
      "loss": 0.1262,
      "step": 876
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.264415746255334,
      "learning_rate": 2.1806700430458948e-06,
      "loss": 0.4108,
      "step": 877
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.680273126283689,
      "learning_rate": 2.175589407628794e-06,
      "loss": 0.3585,
      "step": 878
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.24682536561072,
      "learning_rate": 2.1705101346294936e-06,
      "loss": 0.182,
      "step": 879
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.567777951456993,
      "learning_rate": 2.165432245379273e-06,
      "loss": 0.4158,
      "step": 880
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.282528482598164,
      "learning_rate": 2.1603557612035937e-06,
      "loss": 0.3149,
      "step": 881
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.062430245293939,
      "learning_rate": 2.1552807034220226e-06,
      "loss": 0.4366,
      "step": 882
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.655764076820983,
      "learning_rate": 2.1502070933481318e-06,
      "loss": 0.4819,
      "step": 883
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.216308349794778,
      "learning_rate": 2.1451349522894165e-06,
      "loss": 0.9684,
      "step": 884
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.114028446249881,
      "learning_rate": 2.1400643015471994e-06,
      "loss": 0.2168,
      "step": 885
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.703920909788037,
      "learning_rate": 2.1349951624165474e-06,
      "loss": 0.1802,
      "step": 886
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.930618860044399,
      "learning_rate": 2.129927556186177e-06,
      "loss": 0.2779,
      "step": 887
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.278586493962798,
      "learning_rate": 2.1248615041383686e-06,
      "loss": 0.371,
      "step": 888
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9533216622884786,
      "learning_rate": 2.119797027548874e-06,
      "loss": 0.044,
      "step": 889
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.96798633445337,
      "learning_rate": 2.11473414768683e-06,
      "loss": 0.5775,
      "step": 890
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.909277109667161,
      "learning_rate": 2.1096728858146667e-06,
      "loss": 0.257,
      "step": 891
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.980777056728199,
      "learning_rate": 2.1046132631880194e-06,
      "loss": 0.517,
      "step": 892
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.918104101716663,
      "learning_rate": 2.099555301055639e-06,
      "loss": 0.1367,
      "step": 893
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.490316356227466,
      "learning_rate": 2.094499020659304e-06,
      "loss": 0.2168,
      "step": 894
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.4285994981500245,
      "learning_rate": 2.0894444432337284e-06,
      "loss": 0.0505,
      "step": 895
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.161394177245647,
      "learning_rate": 2.084391590006476e-06,
      "loss": 0.1563,
      "step": 896
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.159789963097649,
      "learning_rate": 2.079340482197867e-06,
      "loss": 0.3148,
      "step": 897
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.475321550380632,
      "learning_rate": 2.0742911410208956e-06,
      "loss": 0.0655,
      "step": 898
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.764324996208217,
      "learning_rate": 2.0692435876811317e-06,
      "loss": 0.2755,
      "step": 899
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.186111448520437,
      "learning_rate": 2.0641978433766412e-06,
      "loss": 0.2011,
      "step": 900
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.118794580444385,
      "learning_rate": 2.0591539292978894e-06,
      "loss": 0.4267,
      "step": 901
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.822696471046605,
      "learning_rate": 2.054111866627658e-06,
      "loss": 0.4114,
      "step": 902
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.864310872771928,
      "learning_rate": 2.049071676540951e-06,
      "loss": 0.076,
      "step": 903
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.003151320510557,
      "learning_rate": 2.04403338020491e-06,
      "loss": 0.1848,
      "step": 904
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.574367764465379,
      "learning_rate": 2.038996998778722e-06,
      "loss": 0.262,
      "step": 905
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9392739043718448,
      "learning_rate": 2.033962553413534e-06,
      "loss": 0.0052,
      "step": 906
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.344892697001689,
      "learning_rate": 2.0289300652523594e-06,
      "loss": 0.4998,
      "step": 907
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.67647400658845,
      "learning_rate": 2.0238995554299946e-06,
      "loss": 0.4042,
      "step": 908
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.499106865128995,
      "learning_rate": 2.0188710450729255e-06,
      "loss": 0.3047,
      "step": 909
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.004581641090867,
      "learning_rate": 2.0138445552992432e-06,
      "loss": 0.0607,
      "step": 910
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.094376039941312,
      "learning_rate": 2.0088201072185497e-06,
      "loss": 0.1673,
      "step": 911
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.014596434640352,
      "learning_rate": 2.0037977219318776e-06,
      "loss": 0.5188,
      "step": 912
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.863010301389402,
      "learning_rate": 1.99877742053159e-06,
      "loss": 0.3261,
      "step": 913
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.981489215198552,
      "learning_rate": 1.993759224101305e-06,
      "loss": 0.4534,
      "step": 914
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.07740233584443,
      "learning_rate": 1.9887431537157944e-06,
      "loss": 0.1136,
      "step": 915
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.884234146148189,
      "learning_rate": 1.9837292304409074e-06,
      "loss": 0.408,
      "step": 916
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.787966447683212,
      "learning_rate": 1.9787174753334694e-06,
      "loss": 0.2243,
      "step": 917
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.283351185380925,
      "learning_rate": 1.9737079094412072e-06,
      "loss": 0.5424,
      "step": 918
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.1475586669305855,
      "learning_rate": 1.9687005538026464e-06,
      "loss": 0.3028,
      "step": 919
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.465485347328667,
      "learning_rate": 1.9636954294470375e-06,
      "loss": 0.0826,
      "step": 920
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.5577608676894794,
      "learning_rate": 1.958692557394254e-06,
      "loss": 0.1947,
      "step": 921
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.472700512289778,
      "learning_rate": 1.9536919586547153e-06,
      "loss": 0.2325,
      "step": 922
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.082228503645407,
      "learning_rate": 1.94869365422929e-06,
      "loss": 0.3296,
      "step": 923
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.010116351545602,
      "learning_rate": 1.9436976651092143e-06,
      "loss": 0.3629,
      "step": 924
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.9070867023831153,
      "learning_rate": 1.9387040122759984e-06,
      "loss": 0.1464,
      "step": 925
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.330271265084205,
      "learning_rate": 1.933712716701343e-06,
      "loss": 0.231,
      "step": 926
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.8114288327506,
      "learning_rate": 1.9287237993470475e-06,
      "loss": 0.1546,
      "step": 927
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.41754775266865,
      "learning_rate": 1.923737281164924e-06,
      "loss": 0.452,
      "step": 928
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.241115988554974,
      "learning_rate": 1.918753183096709e-06,
      "loss": 0.5664,
      "step": 929
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.345251874881729,
      "learning_rate": 1.913771526073976e-06,
      "loss": 0.4855,
      "step": 930
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.155355885045427,
      "learning_rate": 1.9087923310180453e-06,
      "loss": 0.8178,
      "step": 931
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.797336220701351,
      "learning_rate": 1.9038156188398992e-06,
      "loss": 0.0949,
      "step": 932
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.868256940025731,
      "learning_rate": 1.898841410440092e-06,
      "loss": 0.1998,
      "step": 933
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.704476568722761,
      "learning_rate": 1.8938697267086636e-06,
      "loss": 0.4457,
      "step": 934
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.702762627819754,
      "learning_rate": 1.88890058852505e-06,
      "loss": 0.1097,
      "step": 935
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.170947504972868,
      "learning_rate": 1.8839340167579977e-06,
      "loss": 0.2688,
      "step": 936
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.126022050873371,
      "learning_rate": 1.878970032265475e-06,
      "loss": 0.1048,
      "step": 937
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.95862194146853,
      "learning_rate": 1.8740086558945842e-06,
      "loss": 0.0675,
      "step": 938
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.157331297607431,
      "learning_rate": 1.8690499084814739e-06,
      "loss": 0.7983,
      "step": 939
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.726465731825782,
      "learning_rate": 1.8640938108512538e-06,
      "loss": 0.435,
      "step": 940
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.614194874872148,
      "learning_rate": 1.8591403838179026e-06,
      "loss": 0.4425,
      "step": 941
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.407712529097981,
      "learning_rate": 1.854189648184186e-06,
      "loss": 0.1933,
      "step": 942
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.040178261408725,
      "learning_rate": 1.8492416247415645e-06,
      "loss": 0.3822,
      "step": 943
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.0005694511845410909,
      "learning_rate": 1.8442963342701107e-06,
      "loss": 0.0,
      "step": 944
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.39729231837916,
      "learning_rate": 1.8393537975384168e-06,
      "loss": 0.1104,
      "step": 945
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.878587674652728,
      "learning_rate": 1.8344140353035128e-06,
      "loss": 0.43,
      "step": 946
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.129325901826996,
      "learning_rate": 1.829477068310775e-06,
      "loss": 0.198,
      "step": 947
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.821412527888683,
      "learning_rate": 1.8245429172938414e-06,
      "loss": 0.3571,
      "step": 948
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.974929134305073,
      "learning_rate": 1.8196116029745231e-06,
      "loss": 0.1794,
      "step": 949
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.072969063944583,
      "learning_rate": 1.8146831460627198e-06,
      "loss": 0.048,
      "step": 950
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.224159657607188,
      "learning_rate": 1.8097575672563278e-06,
      "loss": 0.1226,
      "step": 951
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4196871179088537,
      "learning_rate": 1.8048348872411608e-06,
      "loss": 0.0074,
      "step": 952
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.3432469661818,
      "learning_rate": 1.7999151266908531e-06,
      "loss": 0.3635,
      "step": 953
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.432677106164474,
      "learning_rate": 1.7949983062667842e-06,
      "loss": 0.1209,
      "step": 954
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.328754336856798,
      "learning_rate": 1.7900844466179797e-06,
      "loss": 0.3935,
      "step": 955
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.881259913864808,
      "learning_rate": 1.7851735683810373e-06,
      "loss": 0.3537,
      "step": 956
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.165188039190037,
      "learning_rate": 1.7802656921800277e-06,
      "loss": 0.3275,
      "step": 957
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.835526773375352,
      "learning_rate": 1.7753608386264196e-06,
      "loss": 0.2289,
      "step": 958
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.3085698649896145,
      "learning_rate": 1.770459028318982e-06,
      "loss": 0.1853,
      "step": 959
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.779465188491945,
      "learning_rate": 1.7655602818437095e-06,
      "loss": 0.2324,
      "step": 960
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.877593082809007,
      "learning_rate": 1.7606646197737227e-06,
      "loss": 0.6645,
      "step": 961
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.89453414924489,
      "learning_rate": 1.755772062669196e-06,
      "loss": 0.309,
      "step": 962
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.595836569907511,
      "learning_rate": 1.7508826310772566e-06,
      "loss": 0.2078,
      "step": 963
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.8634597227955005,
      "learning_rate": 1.7459963455319124e-06,
      "loss": 0.1937,
      "step": 964
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.245447224042685,
      "learning_rate": 1.7411132265539537e-06,
      "loss": 0.6336,
      "step": 965
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.032942504630441,
      "learning_rate": 1.7362332946508759e-06,
      "loss": 0.2036,
      "step": 966
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8539538815018344,
      "learning_rate": 1.7313565703167868e-06,
      "loss": 0.0123,
      "step": 967
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.1530133564179605,
      "learning_rate": 1.7264830740323257e-06,
      "loss": 0.2706,
      "step": 968
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.977211506591645,
      "learning_rate": 1.7216128262645734e-06,
      "loss": 0.4502,
      "step": 969
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.862819632224306,
      "learning_rate": 1.71674584746697e-06,
      "loss": 0.1013,
      "step": 970
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.729901662394336,
      "learning_rate": 1.711882158079225e-06,
      "loss": 0.4949,
      "step": 971
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.062131326696847,
      "learning_rate": 1.7070217785272354e-06,
      "loss": 0.1355,
      "step": 972
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.303964181266942,
      "learning_rate": 1.7021647292229954e-06,
      "loss": 0.2781,
      "step": 973
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.80831740080856,
      "learning_rate": 1.6973110305645164e-06,
      "loss": 0.2575,
      "step": 974
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.2783490734077185,
      "learning_rate": 1.6924607029357354e-06,
      "loss": 0.2815,
      "step": 975
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.9908680492974415,
      "learning_rate": 1.6876137667064339e-06,
      "loss": 0.1786,
      "step": 976
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.487313985604553,
      "learning_rate": 1.6827702422321496e-06,
      "loss": 0.1503,
      "step": 977
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.370661082599542,
      "learning_rate": 1.6779301498540934e-06,
      "loss": 0.2291,
      "step": 978
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.715340173901825,
      "learning_rate": 1.6730935098990602e-06,
      "loss": 0.409,
      "step": 979
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.987489498576722,
      "learning_rate": 1.6682603426793487e-06,
      "loss": 0.1897,
      "step": 980
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.626683784056619,
      "learning_rate": 1.6634306684926705e-06,
      "loss": 0.4759,
      "step": 981
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.641636799157488,
      "learning_rate": 1.65860450762207e-06,
      "loss": 0.128,
      "step": 982
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.557916312549223,
      "learning_rate": 1.6537818803358347e-06,
      "loss": 0.5851,
      "step": 983
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.318635669498663,
      "learning_rate": 1.6489628068874142e-06,
      "loss": 0.3414,
      "step": 984
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.349500410473203,
      "learning_rate": 1.644147307515331e-06,
      "loss": 0.4838,
      "step": 985
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.401575695026681,
      "learning_rate": 1.6393354024431e-06,
      "loss": 0.2328,
      "step": 986
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.015267879083828,
      "learning_rate": 1.6345271118791395e-06,
      "loss": 0.5914,
      "step": 987
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.086201016237776,
      "learning_rate": 1.629722456016689e-06,
      "loss": 0.2998,
      "step": 988
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.831802813844422,
      "learning_rate": 1.6249214550337222e-06,
      "loss": 0.5583,
      "step": 989
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.586155488548567,
      "learning_rate": 1.6201241290928654e-06,
      "loss": 0.3286,
      "step": 990
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.2517639655007406,
      "learning_rate": 1.615330498341309e-06,
      "loss": 0.0454,
      "step": 991
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.182363781326082,
      "learning_rate": 1.6105405829107268e-06,
      "loss": 0.2463,
      "step": 992
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.643108634794028,
      "learning_rate": 1.6057544029171863e-06,
      "loss": 0.1361,
      "step": 993
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.297687659474982,
      "learning_rate": 1.6009719784610724e-06,
      "loss": 0.1918,
      "step": 994
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.361747728058531,
      "learning_rate": 1.5961933296269923e-06,
      "loss": 0.1998,
      "step": 995
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.086942306227849,
      "learning_rate": 1.5914184764837024e-06,
      "loss": 0.3141,
      "step": 996
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.080349820561993,
      "learning_rate": 1.5866474390840126e-06,
      "loss": 0.4342,
      "step": 997
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.043675544297867,
      "learning_rate": 1.5818802374647134e-06,
      "loss": 0.3817,
      "step": 998
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.109757995249827,
      "learning_rate": 1.577116891646482e-06,
      "loss": 0.268,
      "step": 999
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.481416440050833,
      "learning_rate": 1.5723574216338066e-06,
      "loss": 0.38,
      "step": 1000
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.905021454641723,
      "learning_rate": 1.5676018474148935e-06,
      "loss": 0.1244,
      "step": 1001
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.724723826161714,
      "learning_rate": 1.562850188961593e-06,
      "loss": 0.1855,
      "step": 1002
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.249846586037759,
      "learning_rate": 1.5581024662293047e-06,
      "loss": 0.3089,
      "step": 1003
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.341712961328778,
      "learning_rate": 1.5533586991569065e-06,
      "loss": 0.421,
      "step": 1004
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.275293590489044,
      "learning_rate": 1.5486189076666575e-06,
      "loss": 0.3399,
      "step": 1005
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.811066252920472,
      "learning_rate": 1.5438831116641244e-06,
      "loss": 0.1119,
      "step": 1006
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7495788339894816,
      "learning_rate": 1.5391513310380923e-06,
      "loss": 0.0152,
      "step": 1007
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.7866886672666995,
      "learning_rate": 1.5344235856604845e-06,
      "loss": 0.4424,
      "step": 1008
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.584903037571111,
      "learning_rate": 1.5296998953862755e-06,
      "loss": 0.3718,
      "step": 1009
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.780438211958827,
      "learning_rate": 1.5249802800534125e-06,
      "loss": 0.2445,
      "step": 1010
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.219854055575279,
      "learning_rate": 1.5202647594827269e-06,
      "loss": 0.4316,
      "step": 1011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.696767437583493,
      "learning_rate": 1.515553353477855e-06,
      "loss": 0.0192,
      "step": 1012
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.291649685729808,
      "learning_rate": 1.5108460818251513e-06,
      "loss": 0.2354,
      "step": 1013
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.03699664828297,
      "learning_rate": 1.5061429642936107e-06,
      "loss": 0.3573,
      "step": 1014
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.268248008257986,
      "learning_rate": 1.5014440206347785e-06,
      "loss": 0.3276,
      "step": 1015
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.071068126333296,
      "learning_rate": 1.4967492705826735e-06,
      "loss": 0.3007,
      "step": 1016
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.901553240320484,
      "learning_rate": 1.4920587338537018e-06,
      "loss": 0.3938,
      "step": 1017
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.879127234614124,
      "learning_rate": 1.4873724301465753e-06,
      "loss": 0.2296,
      "step": 1018
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.506181191264519,
      "learning_rate": 1.4826903791422282e-06,
      "loss": 0.3104,
      "step": 1019
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.972314581527378,
      "learning_rate": 1.4780126005037354e-06,
      "loss": 0.3961,
      "step": 1020
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.687599265102159,
      "learning_rate": 1.4733391138762277e-06,
      "loss": 0.1626,
      "step": 1021
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.415780175474357,
      "learning_rate": 1.4686699388868137e-06,
      "loss": 0.2941,
      "step": 1022
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.9160262643309744,
      "learning_rate": 1.4640050951444912e-06,
      "loss": 0.184,
      "step": 1023
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.384766762389278,
      "learning_rate": 1.4593446022400715e-06,
      "loss": 0.4718,
      "step": 1024
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.417072068656092,
      "learning_rate": 1.4546884797460904e-06,
      "loss": 0.2465,
      "step": 1025
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.750760446374365,
      "learning_rate": 1.450036747216732e-06,
      "loss": 0.1225,
      "step": 1026
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.143795399162357,
      "learning_rate": 1.4453894241877433e-06,
      "loss": 0.2569,
      "step": 1027
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46836918620952056,
      "learning_rate": 1.4407465301763534e-06,
      "loss": 0.0021,
      "step": 1028
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.7521595560981025,
      "learning_rate": 1.4361080846811888e-06,
      "loss": 0.4665,
      "step": 1029
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.3502201691201714,
      "learning_rate": 1.4314741071821955e-06,
      "loss": 0.0774,
      "step": 1030
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.693049466974612,
      "learning_rate": 1.4268446171405554e-06,
      "loss": 0.3174,
      "step": 1031
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.8976653716194125,
      "learning_rate": 1.4222196339986053e-06,
      "loss": 0.0902,
      "step": 1032
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.141925703130366,
      "learning_rate": 1.4175991771797508e-06,
      "loss": 0.2603,
      "step": 1033
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.170304074227231,
      "learning_rate": 1.4129832660883933e-06,
      "loss": 0.1593,
      "step": 1034
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.355323181931178,
      "learning_rate": 1.4083719201098404e-06,
      "loss": 0.5368,
      "step": 1035
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.152163596326409,
      "learning_rate": 1.4037651586102297e-06,
      "loss": 0.5907,
      "step": 1036
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.3126463264105706,
      "learning_rate": 1.399163000936443e-06,
      "loss": 0.4369,
      "step": 1037
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.408593700628791,
      "learning_rate": 1.3945654664160296e-06,
      "loss": 0.2632,
      "step": 1038
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.204173732093424,
      "learning_rate": 1.389972574357123e-06,
      "loss": 0.2789,
      "step": 1039
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.896821068441229,
      "learning_rate": 1.38538434404836e-06,
      "loss": 0.5113,
      "step": 1040
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.707353959285309,
      "learning_rate": 1.3808007947587976e-06,
      "loss": 0.479,
      "step": 1041
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.706670038052496,
      "learning_rate": 1.3762219457378357e-06,
      "loss": 0.112,
      "step": 1042
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.500034332196811,
      "learning_rate": 1.3716478162151347e-06,
      "loss": 0.5281,
      "step": 1043
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.725800717258658,
      "learning_rate": 1.3670784254005354e-06,
      "loss": 0.2327,
      "step": 1044
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.471210424454569,
      "learning_rate": 1.3625137924839749e-06,
      "loss": 0.3952,
      "step": 1045
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.302821779321099,
      "learning_rate": 1.3579539366354106e-06,
      "loss": 0.1637,
      "step": 1046
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.267206605363414,
      "learning_rate": 1.353398877004738e-06,
      "loss": 0.1717,
      "step": 1047
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7598349983453683,
      "learning_rate": 1.34884863272171e-06,
      "loss": 0.081,
      "step": 1048
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.177234804334449,
      "learning_rate": 1.3443032228958547e-06,
      "loss": 0.2947,
      "step": 1049
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.667959310461245,
      "learning_rate": 1.3397626666163998e-06,
      "loss": 0.2487,
      "step": 1050
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.705428392307878,
      "learning_rate": 1.335226982952188e-06,
      "loss": 0.2478,
      "step": 1051
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.654495820641348,
      "learning_rate": 1.3306961909516e-06,
      "loss": 0.4952,
      "step": 1052
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.797367179793667,
      "learning_rate": 1.3261703096424695e-06,
      "loss": 0.2861,
      "step": 1053
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.183331640373762,
      "learning_rate": 1.3216493580320137e-06,
      "loss": 0.422,
      "step": 1054
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.15635336794597313,
      "learning_rate": 1.3171333551067389e-06,
      "loss": 0.0006,
      "step": 1055
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.11514306747898,
      "learning_rate": 1.3126223198323752e-06,
      "loss": 0.1062,
      "step": 1056
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.600095470778333,
      "learning_rate": 1.3081162711537838e-06,
      "loss": 0.2,
      "step": 1057
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.592780659662964,
      "learning_rate": 1.303615227994891e-06,
      "loss": 0.3045,
      "step": 1058
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.706069987427142,
      "learning_rate": 1.2991192092585952e-06,
      "loss": 0.0946,
      "step": 1059
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.087550409990484,
      "learning_rate": 1.2946282338266985e-06,
      "loss": 0.1687,
      "step": 1060
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.968857050230302,
      "learning_rate": 1.290142320559818e-06,
      "loss": 0.1613,
      "step": 1061
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.776313165488766,
      "learning_rate": 1.2856614882973178e-06,
      "loss": 0.2875,
      "step": 1062
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.584041967573621,
      "learning_rate": 1.2811857558572168e-06,
      "loss": 0.3282,
      "step": 1063
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.6335259506077886,
      "learning_rate": 1.2767151420361218e-06,
      "loss": 0.1407,
      "step": 1064
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.5209474413436777,
      "learning_rate": 1.272249665609137e-06,
      "loss": 0.038,
      "step": 1065
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.837792554441897,
      "learning_rate": 1.2677893453297996e-06,
      "loss": 0.402,
      "step": 1066
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.327216196721906,
      "learning_rate": 1.2633341999299845e-06,
      "loss": 0.2392,
      "step": 1067
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.619910646674139,
      "learning_rate": 1.2588842481198382e-06,
      "loss": 0.0695,
      "step": 1068
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.905789547444904,
      "learning_rate": 1.2544395085876954e-06,
      "loss": 0.3835,
      "step": 1069
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.42389883359606,
      "learning_rate": 1.2500000000000007e-06,
      "loss": 0.1295,
      "step": 1070
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.812823830867673,
      "learning_rate": 1.2455657410012283e-06,
      "loss": 0.2931,
      "step": 1071
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.961542239838846,
      "learning_rate": 1.2411367502138095e-06,
      "loss": 0.1907,
      "step": 1072
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.89326203511695,
      "learning_rate": 1.2367130462380483e-06,
      "loss": 0.3032,
      "step": 1073
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.777073854745678,
      "learning_rate": 1.232294647652048e-06,
      "loss": 0.406,
      "step": 1074
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.2234632289856515,
      "learning_rate": 1.2278815730116275e-06,
      "loss": 0.2681,
      "step": 1075
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.129733470841275,
      "learning_rate": 1.22347384085025e-06,
      "loss": 0.3404,
      "step": 1076
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.599542614506364,
      "learning_rate": 1.219071469678941e-06,
      "loss": 0.5208,
      "step": 1077
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.4150575309625335,
      "learning_rate": 1.2146744779862126e-06,
      "loss": 0.0678,
      "step": 1078
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.487797177853558,
      "learning_rate": 1.2102828842379824e-06,
      "loss": 0.5726,
      "step": 1079
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.296861066400527,
      "learning_rate": 1.2058967068775002e-06,
      "loss": 0.1804,
      "step": 1080
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.9092885202242647,
      "learning_rate": 1.2015159643252689e-06,
      "loss": 0.0314,
      "step": 1081
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.357779129026744,
      "learning_rate": 1.1971406749789673e-06,
      "loss": 0.3348,
      "step": 1082
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.648957892199807,
      "learning_rate": 1.19277085721337e-06,
      "loss": 0.2089,
      "step": 1083
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.937832111279206,
      "learning_rate": 1.1884065293802756e-06,
      "loss": 0.2238,
      "step": 1084
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.877264357515748,
      "learning_rate": 1.1840477098084258e-06,
      "loss": 0.19,
      "step": 1085
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.515347844137308,
      "learning_rate": 1.17969441680343e-06,
      "loss": 0.8935,
      "step": 1086
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.696434632574428,
      "learning_rate": 1.1753466686476855e-06,
      "loss": 0.2842,
      "step": 1087
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.987193427469957,
      "learning_rate": 1.1710044836003057e-06,
      "loss": 0.1777,
      "step": 1088
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.658585954525592,
      "learning_rate": 1.1666678798970397e-06,
      "loss": 0.1025,
      "step": 1089
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.038527358294637,
      "learning_rate": 1.1623368757501977e-06,
      "loss": 0.3914,
      "step": 1090
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.6766430354346022,
      "learning_rate": 1.1580114893485715e-06,
      "loss": 0.0806,
      "step": 1091
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.800865948460224,
      "learning_rate": 1.1536917388573618e-06,
      "loss": 0.2416,
      "step": 1092
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.253488562335028,
      "learning_rate": 1.1493776424181005e-06,
      "loss": 0.1593,
      "step": 1093
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.327816350497619,
      "learning_rate": 1.1450692181485749e-06,
      "loss": 0.2859,
      "step": 1094
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.577937617213323,
      "learning_rate": 1.1407664841427472e-06,
      "loss": 0.3745,
      "step": 1095
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.353078800566941,
      "learning_rate": 1.1364694584706884e-06,
      "loss": 0.0767,
      "step": 1096
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.082330177423406,
      "learning_rate": 1.132178159178491e-06,
      "loss": 0.1001,
      "step": 1097
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.730500037193049,
      "learning_rate": 1.1278926042882026e-06,
      "loss": 0.1931,
      "step": 1098
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.870542930090928,
      "learning_rate": 1.123612811797742e-06,
      "loss": 0.3888,
      "step": 1099
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.670458170659172,
      "learning_rate": 1.1193387996808329e-06,
      "loss": 0.5273,
      "step": 1100
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.121059484534487,
      "learning_rate": 1.115070585886918e-06,
      "loss": 0.3433,
      "step": 1101
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.3361939779315914,
      "learning_rate": 1.1108081883410943e-06,
      "loss": 0.2923,
      "step": 1102
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.036367317732274,
      "learning_rate": 1.1065516249440265e-06,
      "loss": 0.1008,
      "step": 1103
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.021726372650959,
      "learning_rate": 1.1023009135718845e-06,
      "loss": 0.2031,
      "step": 1104
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6551065857841114,
      "learning_rate": 1.0980560720762556e-06,
      "loss": 0.0375,
      "step": 1105
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.2919719813776895,
      "learning_rate": 1.0938171182840792e-06,
      "loss": 0.0966,
      "step": 1106
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1069531135936748,
      "learning_rate": 1.089584069997567e-06,
      "loss": 0.0144,
      "step": 1107
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.236674792754401,
      "learning_rate": 1.085356944994131e-06,
      "loss": 0.2038,
      "step": 1108
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.046522261862831,
      "learning_rate": 1.0811357610263037e-06,
      "loss": 0.668,
      "step": 1109
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.421638051316306,
      "learning_rate": 1.0769205358216707e-06,
      "loss": 0.2997,
      "step": 1110
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.741120204247029,
      "learning_rate": 1.0727112870827918e-06,
      "loss": 0.3144,
      "step": 1111
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.016605785664773,
      "learning_rate": 1.068508032487128e-06,
      "loss": 0.401,
      "step": 1112
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.595881546946371,
      "learning_rate": 1.064310789686965e-06,
      "loss": 0.3715,
      "step": 1113
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.424384341419906,
      "learning_rate": 1.0601195763093427e-06,
      "loss": 0.3372,
      "step": 1114
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0735853129567308,
      "learning_rate": 1.0559344099559793e-06,
      "loss": 0.0263,
      "step": 1115
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.952186810915065,
      "learning_rate": 1.051755308203198e-06,
      "loss": 0.1453,
      "step": 1116
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.662348844397025,
      "learning_rate": 1.0475822886018498e-06,
      "loss": 0.463,
      "step": 1117
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.960529518414667,
      "learning_rate": 1.0434153686772455e-06,
      "loss": 0.1973,
      "step": 1118
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.361360520820744,
      "learning_rate": 1.0392545659290789e-06,
      "loss": 0.636,
      "step": 1119
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.177106971974251,
      "learning_rate": 1.0350998978313537e-06,
      "loss": 0.279,
      "step": 1120
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.457132739720128,
      "learning_rate": 1.030951381832308e-06,
      "loss": 0.2955,
      "step": 1121
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.235605617186508,
      "learning_rate": 1.0268090353543456e-06,
      "loss": 0.2572,
      "step": 1122
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.151599967441302,
      "learning_rate": 1.0226728757939595e-06,
      "loss": 0.0566,
      "step": 1123
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.394817590565157,
      "learning_rate": 1.0185429205216602e-06,
      "loss": 0.1488,
      "step": 1124
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.705416058881438,
      "learning_rate": 1.0144191868819003e-06,
      "loss": 0.2971,
      "step": 1125
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1250939067959833,
      "learning_rate": 1.0103016921930057e-06,
      "loss": 0.0226,
      "step": 1126
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.0763281983266255,
      "learning_rate": 1.0061904537470998e-06,
      "loss": 0.1321,
      "step": 1127
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.034939353263933,
      "learning_rate": 1.002085488810033e-06,
      "loss": 0.249,
      "step": 1128
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.826931982444494,
      "learning_rate": 9.979868146213068e-07,
      "loss": 0.2878,
      "step": 1129
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.233577656155282,
      "learning_rate": 9.938944483940053e-07,
      "loss": 0.1311,
      "step": 1130
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.94281067924806,
      "learning_rate": 9.898084073147207e-07,
      "loss": 0.2794,
      "step": 1131
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.594481521596385,
      "learning_rate": 9.85728708543483e-07,
      "loss": 0.0774,
      "step": 1132
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.939104036942747,
      "learning_rate": 9.816553692136834e-07,
      "loss": 0.1812,
      "step": 1133
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.760392348110935,
      "learning_rate": 9.775884064320087e-07,
      "loss": 0.6073,
      "step": 1134
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.599541168447486,
      "learning_rate": 9.735278372783648e-07,
      "loss": 0.3746,
      "step": 1135
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.925234952987741,
      "learning_rate": 9.694736788058079e-07,
      "loss": 0.5354,
      "step": 1136
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.800757410226256,
      "learning_rate": 9.654259480404676e-07,
      "loss": 0.4464,
      "step": 1137
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.06714963281593,
      "learning_rate": 9.613846619814854e-07,
      "loss": 0.4664,
      "step": 1138
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.352241255651302,
      "learning_rate": 9.573498376009313e-07,
      "loss": 0.3984,
      "step": 1139
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9775548562255114,
      "learning_rate": 9.533214918437422e-07,
      "loss": 0.0233,
      "step": 1140
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.0371188240697355,
      "learning_rate": 9.492996416276434e-07,
      "loss": 0.1628,
      "step": 1141
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.430928255416815,
      "learning_rate": 9.452843038430862e-07,
      "loss": 0.2736,
      "step": 1142
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.058750841673109,
      "learning_rate": 9.412754953531664e-07,
      "loss": 0.185,
      "step": 1143
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.536225634086424,
      "learning_rate": 9.37273232993563e-07,
      "loss": 0.4172,
      "step": 1144
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8043015928598556,
      "learning_rate": 9.332775335724592e-07,
      "loss": 0.068,
      "step": 1145
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.47882306020341,
      "learning_rate": 9.292884138704819e-07,
      "loss": 0.6243,
      "step": 1146
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.199609217918214,
      "learning_rate": 9.253058906406195e-07,
      "loss": 0.2345,
      "step": 1147
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.124477043304045,
      "learning_rate": 9.213299806081607e-07,
      "loss": 0.4456,
      "step": 1148
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.799650402053632,
      "learning_rate": 9.1736070047062e-07,
      "loss": 0.0988,
      "step": 1149
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.82518089846187,
      "learning_rate": 9.133980668976697e-07,
      "loss": 0.784,
      "step": 1150
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.680766398264065,
      "learning_rate": 9.094420965310658e-07,
      "loss": 0.5107,
      "step": 1151
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.337185397056643,
      "learning_rate": 9.054928059845836e-07,
      "loss": 0.4432,
      "step": 1152
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.6635458302526525,
      "learning_rate": 9.015502118439451e-07,
      "loss": 0.2868,
      "step": 1153
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.673698372677808,
      "learning_rate": 8.976143306667493e-07,
      "loss": 0.2073,
      "step": 1154
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.00671527825118,
      "learning_rate": 8.936851789824017e-07,
      "loss": 0.2681,
      "step": 1155
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.609538912713343,
      "learning_rate": 8.897627732920474e-07,
      "loss": 0.3573,
      "step": 1156
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.5452974896995,
      "learning_rate": 8.858471300685012e-07,
      "loss": 0.1631,
      "step": 1157
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.728287337495833,
      "learning_rate": 8.81938265756177e-07,
      "loss": 0.3244,
      "step": 1158
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.948301837050428,
      "learning_rate": 8.780361967710185e-07,
      "loss": 0.2076,
      "step": 1159
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.697940838422666,
      "learning_rate": 8.741409395004324e-07,
      "loss": 0.4055,
      "step": 1160
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.494988672663058,
      "learning_rate": 8.702525103032186e-07,
      "loss": 0.1384,
      "step": 1161
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.4240950291053736,
      "learning_rate": 8.663709255095018e-07,
      "loss": 0.2086,
      "step": 1162
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.601571532946843,
      "learning_rate": 8.624962014206597e-07,
      "loss": 0.2748,
      "step": 1163
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.106685456662032,
      "learning_rate": 8.586283543092603e-07,
      "loss": 0.4134,
      "step": 1164
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.1297395238872205,
      "learning_rate": 8.547674004189894e-07,
      "loss": 0.0915,
      "step": 1165
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.037661273197978,
      "learning_rate": 8.509133559645846e-07,
      "loss": 0.4811,
      "step": 1166
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0356577108462715,
      "learning_rate": 8.470662371317631e-07,
      "loss": 0.0052,
      "step": 1167
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.276048622813399,
      "learning_rate": 8.4322606007716e-07,
      "loss": 0.1627,
      "step": 1168
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.769225514553384,
      "learning_rate": 8.393928409282561e-07,
      "loss": 0.1517,
      "step": 1169
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.967266437640687,
      "learning_rate": 8.35566595783312e-07,
      "loss": 0.2066,
      "step": 1170
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.0004566797260357606,
      "learning_rate": 8.317473407112969e-07,
      "loss": 0.0,
      "step": 1171
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.675421685386072,
      "learning_rate": 8.279350917518278e-07,
      "loss": 0.4631,
      "step": 1172
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.45604089883548277,
      "learning_rate": 8.241298649150958e-07,
      "loss": 0.0021,
      "step": 1173
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.14980425182308,
      "learning_rate": 8.203316761818034e-07,
      "loss": 0.4811,
      "step": 1174
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.678894611062049,
      "learning_rate": 8.165405415030917e-07,
      "loss": 0.4395,
      "step": 1175
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.299243869516143,
      "learning_rate": 8.12756476800483e-07,
      "loss": 0.3727,
      "step": 1176
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.287321525902303,
      "learning_rate": 8.08979497965802e-07,
      "loss": 0.6829,
      "step": 1177
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.9881196704349815,
      "learning_rate": 8.052096208611193e-07,
      "loss": 0.2384,
      "step": 1178
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.170632388841534,
      "learning_rate": 8.014468613186768e-07,
      "loss": 0.1246,
      "step": 1179
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.046362177932173,
      "learning_rate": 7.976912351408303e-07,
      "loss": 0.0351,
      "step": 1180
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.544281876161677,
      "learning_rate": 7.939427580999717e-07,
      "loss": 0.1978,
      "step": 1181
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.758983942348827,
      "learning_rate": 7.902014459384744e-07,
      "loss": 0.5287,
      "step": 1182
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.402694326680793,
      "learning_rate": 7.864673143686161e-07,
      "loss": 0.2269,
      "step": 1183
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.817453094051748,
      "learning_rate": 7.827403790725246e-07,
      "loss": 0.442,
      "step": 1184
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.753924888818989,
      "learning_rate": 7.790206557021001e-07,
      "loss": 0.3795,
      "step": 1185
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.122092842096281,
      "learning_rate": 7.753081598789594e-07,
      "loss": 0.5175,
      "step": 1186
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9394371474948677,
      "learning_rate": 7.716029071943612e-07,
      "loss": 0.074,
      "step": 1187
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.743111097388839,
      "learning_rate": 7.67904913209152e-07,
      "loss": 0.098,
      "step": 1188
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.447947915439228,
      "learning_rate": 7.642141934536876e-07,
      "loss": 0.6088,
      "step": 1189
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.857754913865471,
      "learning_rate": 7.605307634277781e-07,
      "loss": 0.2789,
      "step": 1190
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.365997250666482,
      "learning_rate": 7.568546386006185e-07,
      "loss": 0.416,
      "step": 1191
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.114306837360714,
      "learning_rate": 7.531858344107246e-07,
      "loss": 0.1902,
      "step": 1192
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.05534797356418,
      "learning_rate": 7.49524366265866e-07,
      "loss": 0.2576,
      "step": 1193
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.82133989566553,
      "learning_rate": 7.458702495430056e-07,
      "loss": 0.5049,
      "step": 1194
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.067195500885777,
      "learning_rate": 7.422234995882322e-07,
      "loss": 0.4976,
      "step": 1195
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.689589378096578,
      "learning_rate": 7.385841317166966e-07,
      "loss": 0.0803,
      "step": 1196
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.724191150500358,
      "learning_rate": 7.349521612125459e-07,
      "loss": 0.463,
      "step": 1197
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.423637742971244,
      "learning_rate": 7.313276033288624e-07,
      "loss": 0.1105,
      "step": 1198
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.449954705391146,
      "learning_rate": 7.277104732875972e-07,
      "loss": 0.6319,
      "step": 1199
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.051548343337302,
      "learning_rate": 7.241007862795079e-07,
      "loss": 0.3577,
      "step": 1200
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.683512576226657,
      "learning_rate": 7.20498557464091e-07,
      "loss": 0.2195,
      "step": 1201
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.019963904602504,
      "learning_rate": 7.169038019695241e-07,
      "loss": 0.4715,
      "step": 1202
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.181379257507884,
      "learning_rate": 7.133165348925978e-07,
      "loss": 0.0928,
      "step": 1203
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.0344719180629065,
      "learning_rate": 7.097367712986548e-07,
      "loss": 0.3156,
      "step": 1204
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.661127000627731,
      "learning_rate": 7.061645262215233e-07,
      "loss": 0.2014,
      "step": 1205
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.183429667209468,
      "learning_rate": 7.025998146634591e-07,
      "loss": 0.5082,
      "step": 1206
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.344470354343886,
      "learning_rate": 6.990426515950783e-07,
      "loss": 0.2123,
      "step": 1207
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.637031065390516,
      "learning_rate": 6.954930519552963e-07,
      "loss": 0.1589,
      "step": 1208
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.474302645914507,
      "learning_rate": 6.919510306512633e-07,
      "loss": 0.4279,
      "step": 1209
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.846024747295774,
      "learning_rate": 6.884166025583044e-07,
      "loss": 0.0696,
      "step": 1210
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.980983065903735,
      "learning_rate": 6.848897825198556e-07,
      "loss": 0.4559,
      "step": 1211
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.823689641676835,
      "learning_rate": 6.813705853474014e-07,
      "loss": 0.2294,
      "step": 1212
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.549338958230435,
      "learning_rate": 6.778590258204116e-07,
      "loss": 0.2122,
      "step": 1213
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.953447316651441,
      "learning_rate": 6.743551186862817e-07,
      "loss": 0.3647,
      "step": 1214
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.248103223405018,
      "learning_rate": 6.708588786602699e-07,
      "loss": 0.0772,
      "step": 1215
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.828220798566685,
      "learning_rate": 6.673703204254348e-07,
      "loss": 0.0694,
      "step": 1216
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.406936996195398,
      "learning_rate": 6.63889458632572e-07,
      "loss": 0.0508,
      "step": 1217
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.291232593926028,
      "learning_rate": 6.604163079001586e-07,
      "loss": 0.025,
      "step": 1218
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.0449736215474985,
      "learning_rate": 6.569508828142837e-07,
      "loss": 0.3029,
      "step": 1219
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.924293142026417,
      "learning_rate": 6.534931979285947e-07,
      "loss": 0.098,
      "step": 1220
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.687484130832513,
      "learning_rate": 6.500432677642282e-07,
      "loss": 0.179,
      "step": 1221
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.767871309214235,
      "learning_rate": 6.4660110680976e-07,
      "loss": 0.2453,
      "step": 1222
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.2186521589547255,
      "learning_rate": 6.431667295211314e-07,
      "loss": 0.2563,
      "step": 1223
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.201440316942234,
      "learning_rate": 6.397401503215992e-07,
      "loss": 0.4153,
      "step": 1224
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.83563811773922,
      "learning_rate": 6.363213836016669e-07,
      "loss": 0.1926,
      "step": 1225
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.833525273233217,
      "learning_rate": 6.329104437190333e-07,
      "loss": 0.6962,
      "step": 1226
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.720222658745461,
      "learning_rate": 6.295073449985223e-07,
      "loss": 0.1187,
      "step": 1227
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.635566701042897,
      "learning_rate": 6.261121017320299e-07,
      "loss": 0.0345,
      "step": 1228
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.417512253686738,
      "learning_rate": 6.227247281784613e-07,
      "loss": 0.6206,
      "step": 1229
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.748078384893074,
      "learning_rate": 6.193452385636722e-07,
      "loss": 0.3458,
      "step": 1230
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.378567064084603,
      "learning_rate": 6.159736470804059e-07,
      "loss": 0.3271,
      "step": 1231
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.69959905009879,
      "learning_rate": 6.126099678882389e-07,
      "loss": 0.1606,
      "step": 1232
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.7809707916834636,
      "learning_rate": 6.092542151135175e-07,
      "loss": 0.0598,
      "step": 1233
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.208261040951747,
      "learning_rate": 6.059064028493003e-07,
      "loss": 0.4187,
      "step": 1234
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.573923104458301,
      "learning_rate": 6.025665451552968e-07,
      "loss": 0.2647,
      "step": 1235
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.77186642419758,
      "learning_rate": 5.992346560578113e-07,
      "loss": 0.5419,
      "step": 1236
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.235602118287411,
      "learning_rate": 5.959107495496827e-07,
      "loss": 0.167,
      "step": 1237
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.377726654266931,
      "learning_rate": 5.925948395902253e-07,
      "loss": 0.6169,
      "step": 1238
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.714411412601687,
      "learning_rate": 5.892869401051698e-07,
      "loss": 0.2701,
      "step": 1239
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.317109492874942,
      "learning_rate": 5.859870649866065e-07,
      "loss": 0.2369,
      "step": 1240
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.32657318782199934,
      "learning_rate": 5.826952280929258e-07,
      "loss": 0.0016,
      "step": 1241
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.860547408747393,
      "learning_rate": 5.794114432487605e-07,
      "loss": 0.3061,
      "step": 1242
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.334092626948106,
      "learning_rate": 5.761357242449259e-07,
      "loss": 0.4243,
      "step": 1243
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.1639392827395785,
      "learning_rate": 5.728680848383653e-07,
      "loss": 0.0566,
      "step": 1244
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.430033036950494,
      "learning_rate": 5.696085387520894e-07,
      "loss": 0.4149,
      "step": 1245
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.595394906496898,
      "learning_rate": 5.663570996751208e-07,
      "loss": 0.5662,
      "step": 1246
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.530556294636503,
      "learning_rate": 5.631137812624329e-07,
      "loss": 0.4349,
      "step": 1247
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.6481565455714176,
      "learning_rate": 5.598785971348969e-07,
      "loss": 0.2056,
      "step": 1248
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.342931344183399,
      "learning_rate": 5.566515608792223e-07,
      "loss": 0.1623,
      "step": 1249
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9183746423272177,
      "learning_rate": 5.534326860479008e-07,
      "loss": 0.0167,
      "step": 1250
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.021548676371724,
      "learning_rate": 5.502219861591471e-07,
      "loss": 0.2745,
      "step": 1251
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.3786650780300524,
      "learning_rate": 5.470194746968451e-07,
      "loss": 0.2425,
      "step": 1252
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.863226342561528,
      "learning_rate": 5.4382516511049e-07,
      "loss": 0.1711,
      "step": 1253
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.759145686774258,
      "learning_rate": 5.406390708151321e-07,
      "loss": 0.2674,
      "step": 1254
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.065181493095607,
      "learning_rate": 5.374612051913186e-07,
      "loss": 0.2833,
      "step": 1255
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.418198046471073,
      "learning_rate": 5.342915815850402e-07,
      "loss": 0.5178,
      "step": 1256
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.266771113769888,
      "learning_rate": 5.311302133076742e-07,
      "loss": 0.4868,
      "step": 1257
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.99996042232007,
      "learning_rate": 5.279771136359279e-07,
      "loss": 0.1837,
      "step": 1258
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.302677632642498,
      "learning_rate": 5.248322958117815e-07,
      "loss": 0.4099,
      "step": 1259
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.450149987939859,
      "learning_rate": 5.216957730424382e-07,
      "loss": 0.097,
      "step": 1260
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.49409622655994,
      "learning_rate": 5.185675585002597e-07,
      "loss": 0.6324,
      "step": 1261
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.4665371301709,
      "learning_rate": 5.154476653227203e-07,
      "loss": 0.4794,
      "step": 1262
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9993792527445464,
      "learning_rate": 5.123361066123428e-07,
      "loss": 0.0241,
      "step": 1263
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.411799871323337,
      "learning_rate": 5.092328954366535e-07,
      "loss": 0.2649,
      "step": 1264
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.083115909122426,
      "learning_rate": 5.061380448281166e-07,
      "loss": 0.2333,
      "step": 1265
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.19014179037132,
      "learning_rate": 5.030515677840883e-07,
      "loss": 0.3651,
      "step": 1266
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.638652424899343,
      "learning_rate": 4.999734772667552e-07,
      "loss": 0.2034,
      "step": 1267
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.191679142748107,
      "learning_rate": 4.969037862030881e-07,
      "loss": 0.237,
      "step": 1268
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.194300449828385,
      "learning_rate": 4.938425074847778e-07,
      "loss": 0.1435,
      "step": 1269
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5156697334434965,
      "learning_rate": 4.907896539681884e-07,
      "loss": 0.0223,
      "step": 1270
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.079149802142939,
      "learning_rate": 4.877452384743012e-07,
      "loss": 0.1507,
      "step": 1271
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.633573884459923,
      "learning_rate": 4.847092737886608e-07,
      "loss": 0.0838,
      "step": 1272
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.387581648270514,
      "learning_rate": 4.816817726613188e-07,
      "loss": 0.6913,
      "step": 1273
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.890780607922828,
      "learning_rate": 4.786627478067852e-07,
      "loss": 0.1232,
      "step": 1274
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.01637350149555,
      "learning_rate": 4.756522119039714e-07,
      "loss": 0.0432,
      "step": 1275
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.7069533293012995,
      "learning_rate": 4.72650177596139e-07,
      "loss": 0.4121,
      "step": 1276
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.7918383163520435,
      "learning_rate": 4.6965665749084345e-07,
      "loss": 0.2983,
      "step": 1277
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.976018663982369,
      "learning_rate": 4.666716641598851e-07,
      "loss": 0.2853,
      "step": 1278
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.863794212405065,
      "learning_rate": 4.6369521013925475e-07,
      "loss": 0.1324,
      "step": 1279
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.5537468542506465,
      "learning_rate": 4.607273079290811e-07,
      "loss": 0.1497,
      "step": 1280
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.5258704512695465,
      "learning_rate": 4.577679699935755e-07,
      "loss": 0.1564,
      "step": 1281
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.0004484928355647459,
      "learning_rate": 4.548172087609856e-07,
      "loss": 0.0,
      "step": 1282
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.357004540107548,
      "learning_rate": 4.5187503662353747e-07,
      "loss": 0.762,
      "step": 1283
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.270852264114231,
      "learning_rate": 4.4894146593738724e-07,
      "loss": 0.1886,
      "step": 1284
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.921765523026006,
      "learning_rate": 4.4601650902256597e-07,
      "loss": 0.2278,
      "step": 1285
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.427307754627865,
      "learning_rate": 4.4310017816293155e-07,
      "loss": 0.3,
      "step": 1286
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.4904721702882675,
      "learning_rate": 4.401924856061146e-07,
      "loss": 0.236,
      "step": 1287
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.381259395583696,
      "learning_rate": 4.3729344356346767e-07,
      "loss": 0.4531,
      "step": 1288
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.15602227578895,
      "learning_rate": 4.344030642100133e-07,
      "loss": 0.2468,
      "step": 1289
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.471011146663745,
      "learning_rate": 4.315213596843945e-07,
      "loss": 0.1772,
      "step": 1290
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.316799443579021,
      "learning_rate": 4.2864834208882267e-07,
      "loss": 0.2099,
      "step": 1291
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.078974411948371,
      "learning_rate": 4.2578402348902756e-07,
      "loss": 0.1977,
      "step": 1292
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.23347062128574,
      "learning_rate": 4.229284159142044e-07,
      "loss": 0.5377,
      "step": 1293
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3464634610544234,
      "learning_rate": 4.2008153135696587e-07,
      "loss": 0.0015,
      "step": 1294
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.8637455037130986,
      "learning_rate": 4.1724338177329156e-07,
      "loss": 0.2514,
      "step": 1295
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.861907247690541,
      "learning_rate": 4.144139790824764e-07,
      "loss": 0.3169,
      "step": 1296
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.6535236097512325,
      "learning_rate": 4.1159333516707976e-07,
      "loss": 0.4802,
      "step": 1297
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.452392528931751,
      "learning_rate": 4.0878146187288084e-07,
      "loss": 0.2476,
      "step": 1298
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.366285225902557,
      "learning_rate": 4.0597837100882053e-07,
      "loss": 0.6126,
      "step": 1299
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.993415967346264,
      "learning_rate": 4.031840743469606e-07,
      "loss": 0.5295,
      "step": 1300
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.177072080073364,
      "learning_rate": 4.0039858362242555e-07,
      "loss": 0.1871,
      "step": 1301
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.172281990635541,
      "learning_rate": 3.976219105333634e-07,
      "loss": 0.2052,
      "step": 1302
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.308371222855513,
      "learning_rate": 3.9485406674088595e-07,
      "loss": 0.439,
      "step": 1303
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.711694558520636,
      "learning_rate": 3.920950638690277e-07,
      "loss": 0.1565,
      "step": 1304
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.772361122057014,
      "learning_rate": 3.893449135046923e-07,
      "loss": 0.1727,
      "step": 1305
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.311852449332666,
      "learning_rate": 3.8660362719760867e-07,
      "loss": 0.0844,
      "step": 1306
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.42096406614009,
      "learning_rate": 3.838712164602762e-07,
      "loss": 0.4988,
      "step": 1307
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.279107938448386,
      "learning_rate": 3.811476927679228e-07,
      "loss": 0.2964,
      "step": 1308
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.5713009457264016,
      "learning_rate": 3.7843306755845024e-07,
      "loss": 0.045,
      "step": 1309
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.283329875372001,
      "learning_rate": 3.7572735223239426e-07,
      "loss": 0.4163,
      "step": 1310
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.128545682207385,
      "learning_rate": 3.7303055815286733e-07,
      "loss": 0.0763,
      "step": 1311
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.274153767095931,
      "learning_rate": 3.703426966455187e-07,
      "loss": 0.2464,
      "step": 1312
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.929415107371386,
      "learning_rate": 3.6766377899848243e-07,
      "loss": 0.1365,
      "step": 1313
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.654715832431283,
      "learning_rate": 3.649938164623326e-07,
      "loss": 0.4255,
      "step": 1314
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.689330493828162,
      "learning_rate": 3.623328202500323e-07,
      "loss": 0.1858,
      "step": 1315
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.308800548081129,
      "learning_rate": 3.5968080153689104e-07,
      "loss": 0.3296,
      "step": 1316
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.070902662836929,
      "learning_rate": 3.570377714605158e-07,
      "loss": 0.6856,
      "step": 1317
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.784315791308768,
      "learning_rate": 3.544037411207638e-07,
      "loss": 0.1901,
      "step": 1318
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.281675661930889,
      "learning_rate": 3.517787215796953e-07,
      "loss": 0.4394,
      "step": 1319
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.117864071979914,
      "learning_rate": 3.4916272386152946e-07,
      "loss": 0.135,
      "step": 1320
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.175714917027768,
      "learning_rate": 3.4655575895259667e-07,
      "loss": 0.1428,
      "step": 1321
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.587351220783437,
      "learning_rate": 3.4395783780129254e-07,
      "loss": 0.2707,
      "step": 1322
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.910050257101179,
      "learning_rate": 3.413689713180307e-07,
      "loss": 0.2505,
      "step": 1323
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.394795817201763,
      "learning_rate": 3.387891703751994e-07,
      "loss": 0.078,
      "step": 1324
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.0004353534679147159,
      "learning_rate": 3.362184458071141e-07,
      "loss": 0.0,
      "step": 1325
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.491810446132425,
      "learning_rate": 3.3365680840997337e-07,
      "loss": 0.1125,
      "step": 1326
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.685452270769086,
      "learning_rate": 3.3110426894181066e-07,
      "loss": 0.1605,
      "step": 1327
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.849547096392178,
      "learning_rate": 3.2856083812245253e-07,
      "loss": 0.1821,
      "step": 1328
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.796367621110237,
      "learning_rate": 3.260265266334725e-07,
      "loss": 0.4107,
      "step": 1329
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.590596534229808,
      "learning_rate": 3.235013451181457e-07,
      "loss": 0.2966,
      "step": 1330
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.824323365637058,
      "learning_rate": 3.2098530418140315e-07,
      "loss": 0.5485,
      "step": 1331
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.16455123387599,
      "learning_rate": 3.184784143897895e-07,
      "loss": 0.2093,
      "step": 1332
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.05220657340829,
      "learning_rate": 3.1598068627141763e-07,
      "loss": 0.4038,
      "step": 1333
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.689539388690154,
      "learning_rate": 3.1349213031592495e-07,
      "loss": 0.6675,
      "step": 1334
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.255588350843405,
      "learning_rate": 3.11012756974427e-07,
      "loss": 0.3487,
      "step": 1335
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.194256163169376,
      "learning_rate": 3.0854257665947683e-07,
      "loss": 0.1817,
      "step": 1336
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.814087204001677,
      "learning_rate": 3.0608159974501924e-07,
      "loss": 0.4353,
      "step": 1337
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.552531707961687,
      "learning_rate": 3.036298365663487e-07,
      "loss": 0.1674,
      "step": 1338
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.6705039203821888,
      "learning_rate": 3.01187297420063e-07,
      "loss": 0.0878,
      "step": 1339
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8658706426632623,
      "learning_rate": 2.987539925640248e-07,
      "loss": 0.1105,
      "step": 1340
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.168866083198873,
      "learning_rate": 2.96329932217313e-07,
      "loss": 0.1791,
      "step": 1341
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.387063374685278,
      "learning_rate": 2.93915126560185e-07,
      "loss": 0.3495,
      "step": 1342
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.837495637156543,
      "learning_rate": 2.915095857340289e-07,
      "loss": 0.2139,
      "step": 1343
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.393526360801382,
      "learning_rate": 2.891133198413268e-07,
      "loss": 0.1317,
      "step": 1344
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.024247616112522,
      "learning_rate": 2.8672633894560596e-07,
      "loss": 0.3229,
      "step": 1345
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.983989116070319,
      "learning_rate": 2.843486530714026e-07,
      "loss": 0.2376,
      "step": 1346
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.596305260268161,
      "learning_rate": 2.8198027220421363e-07,
      "loss": 0.1078,
      "step": 1347
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.16937889834316,
      "learning_rate": 2.7962120629046183e-07,
      "loss": 0.2107,
      "step": 1348
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.141138584746491,
      "learning_rate": 2.772714652374467e-07,
      "loss": 0.0773,
      "step": 1349
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.2918029500925226,
      "learning_rate": 2.7493105891330837e-07,
      "loss": 0.1872,
      "step": 1350
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.1827283560670367,
      "learning_rate": 2.725999971469839e-07,
      "loss": 0.0751,
      "step": 1351
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.441762520178824,
      "learning_rate": 2.7027828972816664e-07,
      "loss": 0.2989,
      "step": 1352
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.823169472203725,
      "learning_rate": 2.679659464072631e-07,
      "loss": 0.3266,
      "step": 1353
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.302283482499232,
      "learning_rate": 2.656629768953553e-07,
      "loss": 0.0994,
      "step": 1354
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.46057154526848,
      "learning_rate": 2.6336939086415796e-07,
      "loss": 0.0911,
      "step": 1355
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.783404304589278,
      "learning_rate": 2.610851979459786e-07,
      "loss": 0.1883,
      "step": 1356
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.0405895669428435,
      "learning_rate": 2.588104077336751e-07,
      "loss": 0.1434,
      "step": 1357
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.417202477974949,
      "learning_rate": 2.565450297806191e-07,
      "loss": 0.1391,
      "step": 1358
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.845751729631906,
      "learning_rate": 2.5428907360065256e-07,
      "loss": 0.1827,
      "step": 1359
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.130814264428655,
      "learning_rate": 2.520425486680503e-07,
      "loss": 0.1775,
      "step": 1360
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.1435033560399415,
      "learning_rate": 2.4980546441747744e-07,
      "loss": 0.2421,
      "step": 1361
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.788641070421076,
      "learning_rate": 2.4757783024395244e-07,
      "loss": 0.2225,
      "step": 1362
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.5390928925601,
      "learning_rate": 2.4535965550280594e-07,
      "loss": 0.6407,
      "step": 1363
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.059805486333964,
      "learning_rate": 2.431509495096435e-07,
      "loss": 0.4669,
      "step": 1364
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6004940303788455,
      "learning_rate": 2.4095172154030235e-07,
      "loss": 0.003,
      "step": 1365
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.662855339672104,
      "learning_rate": 2.3876198083081754e-07,
      "loss": 0.6122,
      "step": 1366
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.7828968677475965,
      "learning_rate": 2.365817365773801e-07,
      "loss": 0.0665,
      "step": 1367
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.590139202591782,
      "learning_rate": 2.3441099793629906e-07,
      "loss": 0.2537,
      "step": 1368
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.344419525552738,
      "learning_rate": 2.3224977402396183e-07,
      "loss": 0.5967,
      "step": 1369
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.027213468080827,
      "learning_rate": 2.300980739167988e-07,
      "loss": 0.208,
      "step": 1370
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.5308681724236,
      "learning_rate": 2.2795590665124267e-07,
      "loss": 0.5625,
      "step": 1371
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.268451056817928,
      "learning_rate": 2.2582328122369212e-07,
      "loss": 0.0789,
      "step": 1372
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.426635759577424,
      "learning_rate": 2.2370020659047144e-07,
      "loss": 0.2975,
      "step": 1373
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.1845259110910495,
      "learning_rate": 2.2158669166779667e-07,
      "loss": 0.4297,
      "step": 1374
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.98179038071042,
      "learning_rate": 2.1948274533173525e-07,
      "loss": 0.0769,
      "step": 1375
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.546048715383937,
      "learning_rate": 2.1738837641817e-07,
      "loss": 0.3055,
      "step": 1376
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.827645781107163,
      "learning_rate": 2.1530359372276126e-07,
      "loss": 0.1171,
      "step": 1377
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.874737023494315,
      "learning_rate": 2.1322840600091098e-07,
      "loss": 0.3757,
      "step": 1378
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.133585553764287,
      "learning_rate": 2.111628219677253e-07,
      "loss": 0.2789,
      "step": 1379
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.140339423715327,
      "learning_rate": 2.0910685029797789e-07,
      "loss": 0.5185,
      "step": 1380
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.114861671762458,
      "learning_rate": 2.070604996260725e-07,
      "loss": 0.3173,
      "step": 1381
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.469462517975171,
      "learning_rate": 2.0502377854601103e-07,
      "loss": 0.0963,
      "step": 1382
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.584093389202246,
      "learning_rate": 2.0299669561135048e-07,
      "loss": 0.2258,
      "step": 1383
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.381083678617386,
      "learning_rate": 2.0097925933517393e-07,
      "loss": 0.272,
      "step": 1384
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.530355729933298,
      "learning_rate": 1.9897147819004841e-07,
      "loss": 0.5881,
      "step": 1385
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.201646932808746,
      "learning_rate": 1.9697336060799626e-07,
      "loss": 0.1984,
      "step": 1386
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.683744231636336,
      "learning_rate": 1.9498491498045334e-07,
      "loss": 0.6364,
      "step": 1387
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.771182640695547,
      "learning_rate": 1.9300614965823805e-07,
      "loss": 0.2412,
      "step": 1388
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.323466703328956,
      "learning_rate": 1.9103707295151262e-07,
      "loss": 0.3189,
      "step": 1389
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.568747026640621,
      "learning_rate": 1.8907769312975332e-07,
      "loss": 0.2135,
      "step": 1390
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.360316701237373,
      "learning_rate": 1.8712801842171004e-07,
      "loss": 0.2523,
      "step": 1391
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.498390533688088,
      "learning_rate": 1.851880570153755e-07,
      "loss": 0.3119,
      "step": 1392
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.627047154510003,
      "learning_rate": 1.8325781705794964e-07,
      "loss": 0.4537,
      "step": 1393
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.374041146573841,
      "learning_rate": 1.8133730665580608e-07,
      "loss": 0.5308,
      "step": 1394
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.223755755977993,
      "learning_rate": 1.7942653387445607e-07,
      "loss": 0.0912,
      "step": 1395
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.3063778035395694,
      "learning_rate": 1.775255067385173e-07,
      "loss": 0.3447,
      "step": 1396
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.159462956884823,
      "learning_rate": 1.7563423323167878e-07,
      "loss": 0.1981,
      "step": 1397
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.2887937338614,
      "learning_rate": 1.737527212966672e-07,
      "loss": 0.3705,
      "step": 1398
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.325701214063891,
      "learning_rate": 1.7188097883521353e-07,
      "loss": 0.2369,
      "step": 1399
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.599735145094215,
      "learning_rate": 1.700190137080207e-07,
      "loss": 0.5197,
      "step": 1400
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.436528436964425,
      "learning_rate": 1.6816683373473004e-07,
      "loss": 0.1849,
      "step": 1401
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.051643573420177,
      "learning_rate": 1.6632444669388876e-07,
      "loss": 0.4636,
      "step": 1402
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.92276421063471,
      "learning_rate": 1.6449186032291565e-07,
      "loss": 0.1783,
      "step": 1403
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.982641032089617,
      "learning_rate": 1.6266908231807127e-07,
      "loss": 0.2434,
      "step": 1404
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.621919828314278,
      "learning_rate": 1.608561203344239e-07,
      "loss": 0.6195,
      "step": 1405
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.4191879999621,
      "learning_rate": 1.5905298198581776e-07,
      "loss": 0.175,
      "step": 1406
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.64018099825841,
      "learning_rate": 1.5725967484484046e-07,
      "loss": 0.1374,
      "step": 1407
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.074503375724342,
      "learning_rate": 1.5547620644279255e-07,
      "loss": 0.0964,
      "step": 1408
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2040563206383803,
      "learning_rate": 1.537025842696549e-07,
      "loss": 0.182,
      "step": 1409
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.622362701128079,
      "learning_rate": 1.5193881577405729e-07,
      "loss": 0.4963,
      "step": 1410
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.350747438634668,
      "learning_rate": 1.50184908363247e-07,
      "loss": 0.4012,
      "step": 1411
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.583122423660912,
      "learning_rate": 1.484408694030587e-07,
      "loss": 0.3722,
      "step": 1412
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.387553764444533,
      "learning_rate": 1.467067062178823e-07,
      "loss": 0.509,
      "step": 1413
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.379003382480915,
      "learning_rate": 1.449824260906335e-07,
      "loss": 0.0959,
      "step": 1414
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.58051937852108,
      "learning_rate": 1.4326803626272117e-07,
      "loss": 0.6395,
      "step": 1415
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.478441945670635,
      "learning_rate": 1.415635439340185e-07,
      "loss": 0.0697,
      "step": 1416
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.289528582503555,
      "learning_rate": 1.398689562628333e-07,
      "loss": 0.2054,
      "step": 1417
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.4945406361767555,
      "learning_rate": 1.38184280365877e-07,
      "loss": 0.1162,
      "step": 1418
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.735258541311191,
      "learning_rate": 1.365095233182326e-07,
      "loss": 0.4807,
      "step": 1419
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.960743334504654,
      "learning_rate": 1.3484469215333084e-07,
      "loss": 0.7189,
      "step": 1420
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.132947164624221,
      "learning_rate": 1.3318979386291397e-07,
      "loss": 0.1765,
      "step": 1421
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.45488103524444,
      "learning_rate": 1.315448353970114e-07,
      "loss": 0.4651,
      "step": 1422
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.637253170654683,
      "learning_rate": 1.2990982366390654e-07,
      "loss": 0.381,
      "step": 1423
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.641859083620049,
      "learning_rate": 1.2828476553011227e-07,
      "loss": 0.3657,
      "step": 1424
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.08632240806161,
      "learning_rate": 1.2666966782033745e-07,
      "loss": 0.2428,
      "step": 1425
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.529403523482693,
      "learning_rate": 1.2506453731746194e-07,
      "loss": 0.3485,
      "step": 1426
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.941044135798286,
      "learning_rate": 1.234693807625048e-07,
      "loss": 0.1049,
      "step": 1427
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.211934553733774,
      "learning_rate": 1.2188420485460066e-07,
      "loss": 0.1333,
      "step": 1428
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.444159083171612,
      "learning_rate": 1.2030901625096596e-07,
      "loss": 0.2546,
      "step": 1429
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5884628545617323,
      "learning_rate": 1.1874382156687514e-07,
      "loss": 0.1518,
      "step": 1430
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.819229890357049,
      "learning_rate": 1.1718862737563063e-07,
      "loss": 0.023,
      "step": 1431
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.415873738252984,
      "learning_rate": 1.1564344020853768e-07,
      "loss": 0.0783,
      "step": 1432
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.482058478727181,
      "learning_rate": 1.1410826655487317e-07,
      "loss": 0.2457,
      "step": 1433
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.2669632566614215,
      "learning_rate": 1.1258311286186208e-07,
      "loss": 0.1778,
      "step": 1434
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.789782357086632,
      "learning_rate": 1.1106798553464804e-07,
      "loss": 0.3006,
      "step": 1435
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.8420653331100634,
      "learning_rate": 1.0956289093626837e-07,
      "loss": 0.0192,
      "step": 1436
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.920763287609952,
      "learning_rate": 1.0806783538762439e-07,
      "loss": 0.5136,
      "step": 1437
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.284644404278689,
      "learning_rate": 1.0658282516745833e-07,
      "loss": 0.4975,
      "step": 1438
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.4656066580148925,
      "learning_rate": 1.0510786651232452e-07,
      "loss": 0.1322,
      "step": 1439
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.819073569639567,
      "learning_rate": 1.0364296561656467e-07,
      "loss": 0.3233,
      "step": 1440
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.849850979859642,
      "learning_rate": 1.0218812863228012e-07,
      "loss": 0.4625,
      "step": 1441
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.049652983015644,
      "learning_rate": 1.007433616693082e-07,
      "loss": 0.4446,
      "step": 1442
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.74265450375623,
      "learning_rate": 9.930867079519485e-08,
      "loss": 0.1227,
      "step": 1443
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.054313874560784,
      "learning_rate": 9.788406203517037e-08,
      "loss": 0.1747,
      "step": 1444
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.792405371817055,
      "learning_rate": 9.646954137212228e-08,
      "loss": 0.2176,
      "step": 1445
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.561878641120704,
      "learning_rate": 9.506511474657282e-08,
      "loss": 0.316,
      "step": 1446
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1851026529201485,
      "learning_rate": 9.367078805665147e-08,
      "loss": 0.0072,
      "step": 1447
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1354173594798254,
      "learning_rate": 9.22865671580725e-08,
      "loss": 0.0725,
      "step": 1448
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.275467093519736,
      "learning_rate": 9.091245786410768e-08,
      "loss": 0.0572,
      "step": 1449
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.923831706187367,
      "learning_rate": 8.954846594556449e-08,
      "loss": 0.3651,
      "step": 1450
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.066185069201036,
      "learning_rate": 8.819459713076073e-08,
      "loss": 0.154,
      "step": 1451
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.499823624520053,
      "learning_rate": 8.68508571054999e-08,
      "loss": 0.3934,
      "step": 1452
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.946462512481981,
      "learning_rate": 8.551725151304785e-08,
      "loss": 0.2607,
      "step": 1453
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.833471712034125,
      "learning_rate": 8.419378595411004e-08,
      "loss": 0.0076,
      "step": 1454
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.882422769917423,
      "learning_rate": 8.288046598680627e-08,
      "loss": 0.3481,
      "step": 1455
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.331843030054485,
      "learning_rate": 8.157729712664875e-08,
      "loss": 0.0064,
      "step": 1456
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.300152060288706,
      "learning_rate": 8.028428484651796e-08,
      "loss": 0.2812,
      "step": 1457
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.189792540121143,
      "learning_rate": 7.900143457664017e-08,
      "loss": 0.2562,
      "step": 1458
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.755755322269527,
      "learning_rate": 7.772875170456495e-08,
      "loss": 0.3462,
      "step": 1459
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1676605736626782,
      "learning_rate": 7.64662415751416e-08,
      "loss": 0.0096,
      "step": 1460
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.141204075945238,
      "learning_rate": 7.521390949049717e-08,
      "loss": 0.3239,
      "step": 1461
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.666948685453614,
      "learning_rate": 7.397176071001544e-08,
      "loss": 0.2252,
      "step": 1462
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.45920295196315,
      "learning_rate": 7.273980045031188e-08,
      "loss": 0.154,
      "step": 1463
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.196136062989676,
      "learning_rate": 7.151803388521505e-08,
      "loss": 0.234,
      "step": 1464
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.441134121468979,
      "learning_rate": 7.030646614574199e-08,
      "loss": 0.1867,
      "step": 1465
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.762645172171149,
      "learning_rate": 6.910510232007977e-08,
      "loss": 0.0168,
      "step": 1466
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.86077782686222,
      "learning_rate": 6.791394745356089e-08,
      "loss": 0.1645,
      "step": 1467
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.621629466253201,
      "learning_rate": 6.673300654864406e-08,
      "loss": 0.0686,
      "step": 1468
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5898943494099643,
      "learning_rate": 6.556228456489233e-08,
      "loss": 0.0032,
      "step": 1469
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.963063655882533,
      "learning_rate": 6.440178641895389e-08,
      "loss": 0.0059,
      "step": 1470
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.663932851425676,
      "learning_rate": 6.325151698453852e-08,
      "loss": 0.2666,
      "step": 1471
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.834146038128758,
      "learning_rate": 6.211148109239978e-08,
      "loss": 0.3064,
      "step": 1472
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.1838422673162645,
      "learning_rate": 6.09816835303137e-08,
      "loss": 0.4441,
      "step": 1473
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.167225050068681,
      "learning_rate": 5.986212904305871e-08,
      "loss": 0.1062,
      "step": 1474
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.862076673737142,
      "learning_rate": 5.875282233239493e-08,
      "loss": 0.1184,
      "step": 1475
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.9312355121471,
      "learning_rate": 5.7653768057045757e-08,
      "loss": 0.4186,
      "step": 1476
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.775789233193873,
      "learning_rate": 5.65649708326782e-08,
      "loss": 0.378,
      "step": 1477
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.458891085405919,
      "learning_rate": 5.548643523188235e-08,
      "loss": 0.1608,
      "step": 1478
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.640605271422711,
      "learning_rate": 5.441816578415276e-08,
      "loss": 0.1665,
      "step": 1479
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.092384884911356,
      "learning_rate": 5.336016697586988e-08,
      "loss": 0.031,
      "step": 1480
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.562558463142998,
      "learning_rate": 5.231244325028084e-08,
      "loss": 0.2326,
      "step": 1481
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.635782122855447,
      "learning_rate": 5.127499900748123e-08,
      "loss": 0.4579,
      "step": 1482
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.747869084579012,
      "learning_rate": 5.024783860439475e-08,
      "loss": 0.2569,
      "step": 1483
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.104292742151748,
      "learning_rate": 4.9230966354757993e-08,
      "loss": 0.2344,
      "step": 1484
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.30196364860068,
      "learning_rate": 4.8224386529099884e-08,
      "loss": 0.6728,
      "step": 1485
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.122846330168349,
      "learning_rate": 4.7228103354724495e-08,
      "loss": 0.0706,
      "step": 1486
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8690891040515463,
      "learning_rate": 4.624212101569353e-08,
      "loss": 0.0244,
      "step": 1487
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.090607139712084,
      "learning_rate": 4.526644365280858e-08,
      "loss": 0.4098,
      "step": 1488
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.405627515560764,
      "learning_rate": 4.4301075363593904e-08,
      "loss": 0.0104,
      "step": 1489
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.910626432785334,
      "learning_rate": 4.3346020202278684e-08,
      "loss": 0.5296,
      "step": 1490
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.903497069569218,
      "learning_rate": 4.240128217978062e-08,
      "loss": 0.2886,
      "step": 1491
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.4777925410258,
      "learning_rate": 4.14668652636882e-08,
      "loss": 0.4299,
      "step": 1492
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.492501643081704,
      "learning_rate": 4.054277337824597e-08,
      "loss": 0.3325,
      "step": 1493
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.85073144090915,
      "learning_rate": 3.9629010404335644e-08,
      "loss": 0.8091,
      "step": 1494
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.058855314744182,
      "learning_rate": 3.8725580179460874e-08,
      "loss": 0.2288,
      "step": 1495
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.458801124661521,
      "learning_rate": 3.7832486497731966e-08,
      "loss": 0.8825,
      "step": 1496
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.825964370262293,
      "learning_rate": 3.6949733109848395e-08,
      "loss": 0.503,
      "step": 1497
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.12181850647075,
      "learning_rate": 3.6077323723084366e-08,
      "loss": 0.4041,
      "step": 1498
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2882274474099714,
      "learning_rate": 3.5215262001272176e-08,
      "loss": 0.0011,
      "step": 1499
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4677384634071937,
      "learning_rate": 3.436355156478749e-08,
      "loss": 0.0287,
      "step": 1500
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.411555520706232,
      "learning_rate": 3.352219599053408e-08,
      "loss": 0.4804,
      "step": 1501
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.000344902297967,
      "learning_rate": 3.269119881192912e-08,
      "loss": 0.1797,
      "step": 1502
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.25677804791622,
      "learning_rate": 3.187056351888651e-08,
      "loss": 0.4743,
      "step": 1503
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.984594286179355,
      "learning_rate": 3.1060293557805824e-08,
      "loss": 0.2057,
      "step": 1504
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.64017732367631,
      "learning_rate": 3.0260392331553366e-08,
      "loss": 0.6345,
      "step": 1505
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.0176756282493145,
      "learning_rate": 2.9470863199451972e-08,
      "loss": 0.1084,
      "step": 1506
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.171223511896896,
      "learning_rate": 2.869170947726374e-08,
      "loss": 0.1464,
      "step": 1507
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22352443384374662,
      "learning_rate": 2.7922934437178695e-08,
      "loss": 0.0011,
      "step": 1508
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.335905865286441,
      "learning_rate": 2.7164541307798665e-08,
      "loss": 0.2648,
      "step": 1509
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.508021440558775,
      "learning_rate": 2.6416533274125912e-08,
      "loss": 0.2862,
      "step": 1510
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.653577688527514,
      "learning_rate": 2.5678913477547306e-08,
      "loss": 0.2058,
      "step": 1511
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.143165399483983,
      "learning_rate": 2.4951685015824623e-08,
      "loss": 0.5874,
      "step": 1512
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.632141991094312,
      "learning_rate": 2.4234850943077315e-08,
      "loss": 0.1479,
      "step": 1513
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.087911243360334,
      "learning_rate": 2.3528414269772814e-08,
      "loss": 0.1874,
      "step": 1514
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.65698482595607,
      "learning_rate": 2.2832377962713203e-08,
      "loss": 0.1166,
      "step": 1515
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.331473964333673,
      "learning_rate": 2.214674494502217e-08,
      "loss": 0.2949,
      "step": 1516
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.426507600259947,
      "learning_rate": 2.1471518096132516e-08,
      "loss": 0.356,
      "step": 1517
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.82471331939393,
      "learning_rate": 2.0806700251775057e-08,
      "loss": 0.0352,
      "step": 1518
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.090778699395952,
      "learning_rate": 2.0152294203966138e-08,
      "loss": 0.1704,
      "step": 1519
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.500984410789352,
      "learning_rate": 1.9508302700995962e-08,
      "loss": 0.4922,
      "step": 1520
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.001297694218992,
      "learning_rate": 1.8874728447417223e-08,
      "loss": 0.1259,
      "step": 1521
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.76889352497473,
      "learning_rate": 1.8251574104032888e-08,
      "loss": 0.515,
      "step": 1522
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.53008155227246,
      "learning_rate": 1.7638842287887036e-08,
      "loss": 0.27,
      "step": 1523
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.476618315397019,
      "learning_rate": 1.7036535572251267e-08,
      "loss": 0.1063,
      "step": 1524
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.55209631847243,
      "learning_rate": 1.6444656486615808e-08,
      "loss": 0.2422,
      "step": 1525
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.867984124840309,
      "learning_rate": 1.5863207516678138e-08,
      "loss": 0.1624,
      "step": 1526
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.1059014902612425,
      "learning_rate": 1.529219110433272e-08,
      "loss": 0.3255,
      "step": 1527
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.414717310045153,
      "learning_rate": 1.4731609647661016e-08,
      "loss": 0.1094,
      "step": 1528
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.043296477508314,
      "learning_rate": 1.4181465500920367e-08,
      "loss": 0.4796,
      "step": 1529
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.695897536717045,
      "learning_rate": 1.3641760974535401e-08,
      "loss": 0.2226,
      "step": 1530
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.754830163275214,
      "learning_rate": 1.3112498335087764e-08,
      "loss": 0.2284,
      "step": 1531
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.52310592813784,
      "learning_rate": 1.2593679805306402e-08,
      "loss": 0.1476,
      "step": 1532
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.055509285255124,
      "learning_rate": 1.2085307564058679e-08,
      "loss": 0.2849,
      "step": 1533
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.4433155642266255,
      "learning_rate": 1.1587383746340386e-08,
      "loss": 0.3427,
      "step": 1534
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.3836416728327485,
      "learning_rate": 1.1099910443268248e-08,
      "loss": 0.1598,
      "step": 1535
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.742945415788494,
      "learning_rate": 1.062288970206965e-08,
      "loss": 0.5362,
      "step": 1536
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.7552970039360805,
      "learning_rate": 1.0156323526074874e-08,
      "loss": 0.2596,
      "step": 1537
    },
    {
      "epoch": 0.97,
      "grad_norm": 13.226891285414418,
      "learning_rate": 9.700213874708208e-09,
      "loss": 0.2647,
      "step": 1538
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.866419983302873,
      "learning_rate": 9.25456266348046e-09,
      "loss": 0.5231,
      "step": 1539
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.8747296968119205,
      "learning_rate": 8.819371763980345e-09,
      "loss": 0.317,
      "step": 1540
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.722838679159709,
      "learning_rate": 8.394643003865887e-09,
      "loss": 0.2588,
      "step": 1541
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.71384759193691,
      "learning_rate": 7.98037816685887e-09,
      "loss": 0.4926,
      "step": 1542
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.012437327607516,
      "learning_rate": 7.576578992735118e-09,
      "loss": 0.271,
      "step": 1543
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.027609864810419,
      "learning_rate": 7.183247177318109e-09,
      "loss": 0.4248,
      "step": 1544
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.0435917735961375,
      "learning_rate": 6.800384372472324e-09,
      "loss": 0.1569,
      "step": 1545
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.182820387831553,
      "learning_rate": 6.427992186095744e-09,
      "loss": 0.4146,
      "step": 1546
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.89663938016722,
      "learning_rate": 6.066072182112914e-09,
      "loss": 0.2914,
      "step": 1547
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.133212442174154,
      "learning_rate": 5.714625880469116e-09,
      "loss": 0.2168,
      "step": 1548
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.547254608489343,
      "learning_rate": 5.37365475712287e-09,
      "loss": 0.1037,
      "step": 1549
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.261510396954706,
      "learning_rate": 5.043160244040945e-09,
      "loss": 0.2477,
      "step": 1550
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.857257490108165,
      "learning_rate": 4.7231437291916925e-09,
      "loss": 0.4704,
      "step": 1551
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.4681728673351517,
      "learning_rate": 4.413606556538941e-09,
      "loss": 0.0619,
      "step": 1552
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.85391270018635,
      "learning_rate": 4.114550026037279e-09,
      "loss": 0.4576,
      "step": 1553
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.272525022661298,
      "learning_rate": 3.8259753936256695e-09,
      "loss": 0.2374,
      "step": 1554
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.5000498766692116,
      "learning_rate": 3.5478838712227347e-09,
      "loss": 0.1446,
      "step": 1555
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.323387387005008,
      "learning_rate": 3.280276626721479e-09,
      "loss": 0.3869,
      "step": 1556
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.132127968766267,
      "learning_rate": 3.0231547839842945e-09,
      "loss": 0.1073,
      "step": 1557
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.999095530684547,
      "learning_rate": 2.776519422838242e-09,
      "loss": 0.2723,
      "step": 1558
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.005699025015711,
      "learning_rate": 2.540371579070611e-09,
      "loss": 0.0171,
      "step": 1559
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0890057804875521,
      "learning_rate": 2.3147122444250327e-09,
      "loss": 0.0127,
      "step": 1560
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.036589003677193,
      "learning_rate": 2.099542366596208e-09,
      "loss": 0.1716,
      "step": 1561
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.120532145022686,
      "learning_rate": 1.8948628492271303e-09,
      "loss": 0.5481,
      "step": 1562
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.691719695873141,
      "learning_rate": 1.7006745519043688e-09,
      "loss": 0.2047,
      "step": 1563
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.280277986881785,
      "learning_rate": 1.5169782901547358e-09,
      "loss": 0.4618,
      "step": 1564
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.3709332827737795,
      "learning_rate": 1.3437748354425128e-09,
      "loss": 0.225,
      "step": 1565
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.936250939097428,
      "learning_rate": 1.1810649151647313e-09,
      "loss": 0.311,
      "step": 1566
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.997272777622928,
      "learning_rate": 1.028849212649785e-09,
      "loss": 0.155,
      "step": 1567
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.109573979224292,
      "learning_rate": 8.871283671535446e-10,
      "loss": 0.4534,
      "step": 1568
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.087829519186801,
      "learning_rate": 7.559029738571366e-10,
      "loss": 0.0842,
      "step": 1569
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.349230359770883,
      "learning_rate": 6.351735838638906e-10,
      "loss": 0.1848,
      "step": 1570
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.150807272414549,
      "learning_rate": 5.24940704197674e-10,
      "loss": 0.0204,
      "step": 1571
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.855499829003662,
      "learning_rate": 4.2520479780011613e-10,
      "loss": 0.4247,
      "step": 1572
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.755197566434629,
      "learning_rate": 3.35966283529221e-10,
      "loss": 0.3989,
      "step": 1573
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.300312027179169,
      "learning_rate": 2.572255361577014e-10,
      "loss": 0.3569,
      "step": 1574
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.64896707185098,
      "learning_rate": 1.8898288637048122e-10,
      "loss": 0.3549,
      "step": 1575
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.788016744652564,
      "learning_rate": 1.3123862076441785e-10,
      "loss": 0.47,
      "step": 1576
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.3169950589427435,
      "learning_rate": 8.399298184663674e-11,
      "loss": 0.3157,
      "step": 1577
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.822387161701134,
      "learning_rate": 4.724616803286619e-11,
      "loss": 0.2321,
      "step": 1578
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1010913788331604,
      "learning_rate": 2.0998333647714863e-11,
      "loss": 0.0112,
      "step": 1579
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.70557614229169,
      "learning_rate": 5.249588923561533e-12,
      "loss": 0.3462,
      "step": 1580
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.500391512720971,
      "learning_rate": 0.0,
      "loss": 0.1648,
      "step": 1581
    },
    {
      "epoch": 1.0,
      "step": 1581,
      "total_flos": 331147936948224.0,
      "train_loss": 0.295211889783912,
      "train_runtime": 24740.4507,
      "train_samples_per_second": 0.511,
      "train_steps_per_second": 0.064
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1581,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 331147936948224.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}