{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.7389756231169544,
  "eval_steps": 500,
  "global_step": 20000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001369487811558477,
      "grad_norm": 196.0,
      "learning_rate": 0.0,
      "loss": 13.9057,
      "step": 1
    },
    {
      "epoch": 0.0002738975623116954,
      "grad_norm": 198.0,
      "learning_rate": 4.5e-07,
      "loss": 13.8992,
      "step": 2
    },
    {
      "epoch": 0.0004108463434675431,
      "grad_norm": 196.0,
      "learning_rate": 9e-07,
      "loss": 13.8783,
      "step": 3
    },
    {
      "epoch": 0.0005477951246233908,
      "grad_norm": 194.0,
      "learning_rate": 1.35e-06,
      "loss": 13.8561,
      "step": 4
    },
    {
      "epoch": 0.0006847439057792385,
      "grad_norm": 198.0,
      "learning_rate": 1.8e-06,
      "loss": 13.8868,
      "step": 5
    },
    {
      "epoch": 0.0008216926869350862,
      "grad_norm": 195.0,
      "learning_rate": 2.25e-06,
      "loss": 13.8643,
      "step": 6
    },
    {
      "epoch": 0.000958641468090934,
      "grad_norm": 194.0,
      "learning_rate": 2.7e-06,
      "loss": 13.7948,
      "step": 7
    },
    {
      "epoch": 0.0010955902492467817,
      "grad_norm": 186.0,
      "learning_rate": 3.15e-06,
      "loss": 13.6762,
      "step": 8
    },
    {
      "epoch": 0.0012325390304026294,
      "grad_norm": 182.0,
      "learning_rate": 3.6e-06,
      "loss": 13.6216,
      "step": 9
    },
    {
      "epoch": 0.001369487811558477,
      "grad_norm": 180.0,
      "learning_rate": 4.049999999999999e-06,
      "loss": 13.5441,
      "step": 10
    },
    {
      "epoch": 0.0015064365927143248,
      "grad_norm": 157.0,
      "learning_rate": 4.5e-06,
      "loss": 13.3211,
      "step": 11
    },
    {
      "epoch": 0.0016433853738701725,
      "grad_norm": 133.0,
      "learning_rate": 4.949999999999999e-06,
      "loss": 13.1497,
      "step": 12
    },
    {
      "epoch": 0.0017803341550260202,
      "grad_norm": 100.0,
      "learning_rate": 5.4e-06,
      "loss": 12.9686,
      "step": 13
    },
    {
      "epoch": 0.001917282936181868,
      "grad_norm": 79.5,
      "learning_rate": 5.85e-06,
      "loss": 12.8665,
      "step": 14
    },
    {
      "epoch": 0.002054231717337716,
      "grad_norm": 60.5,
      "learning_rate": 6.3e-06,
      "loss": 12.7644,
      "step": 15
    },
    {
      "epoch": 0.0021911804984935633,
      "grad_norm": 41.0,
      "learning_rate": 6.75e-06,
      "loss": 12.6897,
      "step": 16
    },
    {
      "epoch": 0.0023281292796494112,
      "grad_norm": 30.125,
      "learning_rate": 7.2e-06,
      "loss": 12.6356,
      "step": 17
    },
    {
      "epoch": 0.0024650780608052587,
      "grad_norm": 26.875,
      "learning_rate": 7.65e-06,
      "loss": 12.5841,
      "step": 18
    },
    {
      "epoch": 0.0026020268419611066,
      "grad_norm": 30.875,
      "learning_rate": 8.099999999999999e-06,
      "loss": 12.5515,
      "step": 19
    },
    {
      "epoch": 0.002738975623116954,
      "grad_norm": 33.0,
      "learning_rate": 8.55e-06,
      "loss": 12.4829,
      "step": 20
    },
    {
      "epoch": 0.002875924404272802,
      "grad_norm": 34.5,
      "learning_rate": 9e-06,
      "loss": 12.4316,
      "step": 21
    },
    {
      "epoch": 0.0030128731854286495,
      "grad_norm": 32.5,
      "learning_rate": 9.450000000000001e-06,
      "loss": 12.3322,
      "step": 22
    },
    {
      "epoch": 0.0031498219665844975,
      "grad_norm": 29.125,
      "learning_rate": 9.899999999999998e-06,
      "loss": 12.2473,
      "step": 23
    },
    {
      "epoch": 0.003286770747740345,
      "grad_norm": 23.375,
      "learning_rate": 1.035e-05,
      "loss": 12.121,
      "step": 24
    },
    {
      "epoch": 0.003423719528896193,
      "grad_norm": 17.875,
      "learning_rate": 1.08e-05,
      "loss": 12.031,
      "step": 25
    },
    {
      "epoch": 0.0035606683100520404,
      "grad_norm": 18.125,
      "learning_rate": 1.125e-05,
      "loss": 11.9719,
      "step": 26
    },
    {
      "epoch": 0.0036976170912078883,
      "grad_norm": 24.375,
      "learning_rate": 1.17e-05,
      "loss": 11.9073,
      "step": 27
    },
    {
      "epoch": 0.003834565872363736,
      "grad_norm": 25.875,
      "learning_rate": 1.2149999999999999e-05,
      "loss": 11.8401,
      "step": 28
    },
    {
      "epoch": 0.003971514653519584,
      "grad_norm": 23.875,
      "learning_rate": 1.26e-05,
      "loss": 11.7595,
      "step": 29
    },
    {
      "epoch": 0.004108463434675432,
      "grad_norm": 18.375,
      "learning_rate": 1.305e-05,
      "loss": 11.6638,
      "step": 30
    },
    {
      "epoch": 0.004245412215831279,
      "grad_norm": 13.3125,
      "learning_rate": 1.35e-05,
      "loss": 11.5798,
      "step": 31
    },
    {
      "epoch": 0.004382360996987127,
      "grad_norm": 11.8125,
      "learning_rate": 1.395e-05,
      "loss": 11.462,
      "step": 32
    },
    {
      "epoch": 0.0045193097781429745,
      "grad_norm": 10.8125,
      "learning_rate": 1.44e-05,
      "loss": 11.3614,
      "step": 33
    },
    {
      "epoch": 0.0046562585592988225,
      "grad_norm": 8.9375,
      "learning_rate": 1.485e-05,
      "loss": 11.2572,
      "step": 34
    },
    {
      "epoch": 0.00479320734045467,
      "grad_norm": 8.1875,
      "learning_rate": 1.53e-05,
      "loss": 11.1713,
      "step": 35
    },
    {
      "epoch": 0.0049301561216105174,
      "grad_norm": 6.9375,
      "learning_rate": 1.575e-05,
      "loss": 11.0894,
      "step": 36
    },
    {
      "epoch": 0.005067104902766365,
      "grad_norm": 5.40625,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 11.0244,
      "step": 37
    },
    {
      "epoch": 0.005204053683922213,
      "grad_norm": 4.8125,
      "learning_rate": 1.6649999999999998e-05,
      "loss": 10.9753,
      "step": 38
    },
    {
      "epoch": 0.005341002465078061,
      "grad_norm": 4.21875,
      "learning_rate": 1.71e-05,
      "loss": 10.8844,
      "step": 39
    },
    {
      "epoch": 0.005477951246233908,
      "grad_norm": 3.96875,
      "learning_rate": 1.755e-05,
      "loss": 10.8384,
      "step": 40
    },
    {
      "epoch": 0.005614900027389756,
      "grad_norm": 3.703125,
      "learning_rate": 1.8e-05,
      "loss": 10.7818,
      "step": 41
    },
    {
      "epoch": 0.005751848808545604,
      "grad_norm": 2.875,
      "learning_rate": 1.845e-05,
      "loss": 10.7277,
      "step": 42
    },
    {
      "epoch": 0.005888797589701452,
      "grad_norm": 2.4375,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 10.6632,
      "step": 43
    },
    {
      "epoch": 0.006025746370857299,
      "grad_norm": 2.234375,
      "learning_rate": 1.935e-05,
      "loss": 10.6237,
      "step": 44
    },
    {
      "epoch": 0.006162695152013147,
      "grad_norm": 2.859375,
      "learning_rate": 1.9799999999999997e-05,
      "loss": 10.57,
      "step": 45
    },
    {
      "epoch": 0.006299643933168995,
      "grad_norm": 1.8984375,
      "learning_rate": 2.0249999999999998e-05,
      "loss": 10.5182,
      "step": 46
    },
    {
      "epoch": 0.006436592714324843,
      "grad_norm": 1.796875,
      "learning_rate": 2.07e-05,
      "loss": 10.4806,
      "step": 47
    },
    {
      "epoch": 0.00657354149548069,
      "grad_norm": 1.7890625,
      "learning_rate": 2.115e-05,
      "loss": 10.4212,
      "step": 48
    },
    {
      "epoch": 0.006710490276636538,
      "grad_norm": 1.7890625,
      "learning_rate": 2.16e-05,
      "loss": 10.3629,
      "step": 49
    },
    {
      "epoch": 0.006847439057792386,
      "grad_norm": 1.828125,
      "learning_rate": 2.205e-05,
      "loss": 10.2941,
      "step": 50
    },
    {
      "epoch": 0.006984387838948234,
      "grad_norm": 1.7421875,
      "learning_rate": 2.25e-05,
      "loss": 10.2314,
      "step": 51
    },
    {
      "epoch": 0.007121336620104081,
      "grad_norm": 1.7109375,
      "learning_rate": 2.295e-05,
      "loss": 10.1834,
      "step": 52
    },
    {
      "epoch": 0.007258285401259929,
      "grad_norm": 1.65625,
      "learning_rate": 2.34e-05,
      "loss": 10.0932,
      "step": 53
    },
    {
      "epoch": 0.007395234182415777,
      "grad_norm": 1.59375,
      "learning_rate": 2.385e-05,
      "loss": 10.0131,
      "step": 54
    },
    {
      "epoch": 0.0075321829635716245,
      "grad_norm": 1.609375,
      "learning_rate": 2.4299999999999998e-05,
      "loss": 9.9487,
      "step": 55
    },
    {
      "epoch": 0.007669131744727472,
      "grad_norm": 1.5625,
      "learning_rate": 2.475e-05,
      "loss": 9.8628,
      "step": 56
    },
    {
      "epoch": 0.0078060805258833195,
      "grad_norm": 1.4375,
      "learning_rate": 2.52e-05,
      "loss": 9.7863,
      "step": 57
    },
    {
      "epoch": 0.007943029307039167,
      "grad_norm": 1.6484375,
      "learning_rate": 2.565e-05,
      "loss": 9.7262,
      "step": 58
    },
    {
      "epoch": 0.008079978088195015,
      "grad_norm": 1.3046875,
      "learning_rate": 2.61e-05,
      "loss": 9.6532,
      "step": 59
    },
    {
      "epoch": 0.008216926869350863,
      "grad_norm": 1.25,
      "learning_rate": 2.6549999999999998e-05,
      "loss": 9.5901,
      "step": 60
    },
    {
      "epoch": 0.008353875650506711,
      "grad_norm": 1.2265625,
      "learning_rate": 2.7e-05,
      "loss": 9.5115,
      "step": 61
    },
    {
      "epoch": 0.008490824431662557,
      "grad_norm": 1.3359375,
      "learning_rate": 2.745e-05,
      "loss": 9.4599,
      "step": 62
    },
    {
      "epoch": 0.008627773212818405,
      "grad_norm": 1.1796875,
      "learning_rate": 2.79e-05,
      "loss": 9.4067,
      "step": 63
    },
    {
      "epoch": 0.008764721993974253,
      "grad_norm": 0.94921875,
      "learning_rate": 2.8349999999999998e-05,
      "loss": 9.3618,
      "step": 64
    },
    {
      "epoch": 0.008901670775130101,
      "grad_norm": 1.03125,
      "learning_rate": 2.88e-05,
      "loss": 9.2991,
      "step": 65
    },
    {
      "epoch": 0.009038619556285949,
      "grad_norm": 0.875,
      "learning_rate": 2.925e-05,
      "loss": 9.269,
      "step": 66
    },
    {
      "epoch": 0.009175568337441797,
      "grad_norm": 1.21875,
      "learning_rate": 2.97e-05,
      "loss": 9.2197,
      "step": 67
    },
    {
      "epoch": 0.009312517118597645,
      "grad_norm": 0.7109375,
      "learning_rate": 3.015e-05,
      "loss": 9.2047,
      "step": 68
    },
    {
      "epoch": 0.009449465899753493,
      "grad_norm": 0.80078125,
      "learning_rate": 3.06e-05,
      "loss": 9.1676,
      "step": 69
    },
    {
      "epoch": 0.00958641468090934,
      "grad_norm": 0.67578125,
      "learning_rate": 3.105e-05,
      "loss": 9.1279,
      "step": 70
    },
    {
      "epoch": 0.009723363462065187,
      "grad_norm": 0.6171875,
      "learning_rate": 3.15e-05,
      "loss": 9.1303,
      "step": 71
    },
    {
      "epoch": 0.009860312243221035,
      "grad_norm": 0.5703125,
      "learning_rate": 3.195e-05,
      "loss": 9.1105,
      "step": 72
    },
    {
      "epoch": 0.009997261024376883,
      "grad_norm": 0.625,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 9.0798,
      "step": 73
    },
    {
      "epoch": 0.01013420980553273,
      "grad_norm": 0.53515625,
      "learning_rate": 3.285e-05,
      "loss": 9.0347,
      "step": 74
    },
    {
      "epoch": 0.010271158586688579,
      "grad_norm": 0.55859375,
      "learning_rate": 3.3299999999999996e-05,
      "loss": 9.0461,
      "step": 75
    },
    {
      "epoch": 0.010408107367844427,
      "grad_norm": 0.5234375,
      "learning_rate": 3.375e-05,
      "loss": 9.0455,
      "step": 76
    },
    {
      "epoch": 0.010545056149000274,
      "grad_norm": 0.486328125,
      "learning_rate": 3.42e-05,
      "loss": 9.0306,
      "step": 77
    },
    {
      "epoch": 0.010682004930156122,
      "grad_norm": 0.490234375,
      "learning_rate": 3.465e-05,
      "loss": 9.022,
      "step": 78
    },
    {
      "epoch": 0.010818953711311969,
      "grad_norm": 0.5234375,
      "learning_rate": 3.51e-05,
      "loss": 9.0264,
      "step": 79
    },
    {
      "epoch": 0.010955902492467817,
      "grad_norm": 0.47265625,
      "learning_rate": 3.555e-05,
      "loss": 9.0006,
      "step": 80
    },
    {
      "epoch": 0.011092851273623664,
      "grad_norm": 0.48828125,
      "learning_rate": 3.6e-05,
      "loss": 8.9984,
      "step": 81
    },
    {
      "epoch": 0.011229800054779512,
      "grad_norm": 0.6328125,
      "learning_rate": 3.645e-05,
      "loss": 8.9881,
      "step": 82
    },
    {
      "epoch": 0.01136674883593536,
      "grad_norm": 0.87890625,
      "learning_rate": 3.69e-05,
      "loss": 8.9739,
      "step": 83
    },
    {
      "epoch": 0.011503697617091208,
      "grad_norm": 0.609375,
      "learning_rate": 3.735e-05,
      "loss": 8.9778,
      "step": 84
    },
    {
      "epoch": 0.011640646398247056,
      "grad_norm": 0.74609375,
      "learning_rate": 3.7800000000000004e-05,
      "loss": 8.9903,
      "step": 85
    },
    {
      "epoch": 0.011777595179402904,
      "grad_norm": 1.6875,
      "learning_rate": 3.825e-05,
      "loss": 8.9727,
      "step": 86
    },
    {
      "epoch": 0.01191454396055875,
      "grad_norm": 2.109375,
      "learning_rate": 3.87e-05,
      "loss": 8.9958,
      "step": 87
    },
    {
      "epoch": 0.012051492741714598,
      "grad_norm": 0.71484375,
      "learning_rate": 3.9149999999999996e-05,
      "loss": 8.9847,
      "step": 88
    },
    {
      "epoch": 0.012188441522870446,
      "grad_norm": 1.8515625,
      "learning_rate": 3.9599999999999994e-05,
      "loss": 8.9563,
      "step": 89
    },
    {
      "epoch": 0.012325390304026294,
      "grad_norm": 2.0,
      "learning_rate": 4.005e-05,
      "loss": 8.9727,
      "step": 90
    },
    {
      "epoch": 0.012462339085182142,
      "grad_norm": 0.69140625,
      "learning_rate": 4.0499999999999995e-05,
      "loss": 8.9688,
      "step": 91
    },
    {
      "epoch": 0.01259928786633799,
      "grad_norm": 3.40625,
      "learning_rate": 4.095e-05,
      "loss": 8.9711,
      "step": 92
    },
    {
      "epoch": 0.012736236647493838,
      "grad_norm": 2.203125,
      "learning_rate": 4.14e-05,
      "loss": 8.9566,
      "step": 93
    },
    {
      "epoch": 0.012873185428649686,
      "grad_norm": 1.5078125,
      "learning_rate": 4.185e-05,
      "loss": 8.9424,
      "step": 94
    },
    {
      "epoch": 0.013010134209805532,
      "grad_norm": 4.0625,
      "learning_rate": 4.23e-05,
      "loss": 8.9627,
      "step": 95
    },
    {
      "epoch": 0.01314708299096138,
      "grad_norm": 1.1171875,
      "learning_rate": 4.275e-05,
      "loss": 8.9445,
      "step": 96
    },
    {
      "epoch": 0.013284031772117228,
      "grad_norm": 4.4375,
      "learning_rate": 4.32e-05,
      "loss": 8.9509,
      "step": 97
    },
    {
      "epoch": 0.013420980553273076,
      "grad_norm": 2.90625,
      "learning_rate": 4.365e-05,
      "loss": 8.9389,
      "step": 98
    },
    {
      "epoch": 0.013557929334428924,
      "grad_norm": 11.125,
      "learning_rate": 4.41e-05,
      "loss": 9.0136,
      "step": 99
    },
    {
      "epoch": 0.013694878115584772,
      "grad_norm": 9.0,
      "learning_rate": 4.455e-05,
      "loss": 9.0097,
      "step": 100
    },
    {
      "epoch": 0.01383182689674062,
      "grad_norm": 2.296875,
      "learning_rate": 4.5e-05,
      "loss": 8.9368,
      "step": 101
    },
    {
      "epoch": 0.013968775677896467,
      "grad_norm": 7.4375,
      "learning_rate": 4.545e-05,
      "loss": 8.9928,
      "step": 102
    },
    {
      "epoch": 0.014105724459052315,
      "grad_norm": 6.8125,
      "learning_rate": 4.59e-05,
      "loss": 8.979,
      "step": 103
    },
    {
      "epoch": 0.014242673240208161,
      "grad_norm": 2.796875,
      "learning_rate": 4.6349999999999995e-05,
      "loss": 8.9547,
      "step": 104
    },
    {
      "epoch": 0.01437962202136401,
      "grad_norm": 3.484375,
      "learning_rate": 4.68e-05,
      "loss": 8.959,
      "step": 105
    },
    {
      "epoch": 0.014516570802519857,
      "grad_norm": 3.890625,
      "learning_rate": 4.7249999999999997e-05,
      "loss": 8.9766,
      "step": 106
    },
    {
      "epoch": 0.014653519583675705,
      "grad_norm": 2.734375,
      "learning_rate": 4.77e-05,
      "loss": 8.9405,
      "step": 107
    },
    {
      "epoch": 0.014790468364831553,
      "grad_norm": 1.1015625,
      "learning_rate": 4.815e-05,
      "loss": 8.9176,
      "step": 108
    },
    {
      "epoch": 0.014927417145987401,
      "grad_norm": 2.578125,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 8.938,
      "step": 109
    },
    {
      "epoch": 0.015064365927143249,
      "grad_norm": 2.828125,
      "learning_rate": 4.905e-05,
      "loss": 8.9491,
      "step": 110
    },
    {
      "epoch": 0.015201314708299097,
      "grad_norm": 1.6328125,
      "learning_rate": 4.95e-05,
      "loss": 8.8972,
      "step": 111
    },
    {
      "epoch": 0.015338263489454943,
      "grad_norm": 3.5625,
      "learning_rate": 4.995e-05,
      "loss": 8.9023,
      "step": 112
    },
    {
      "epoch": 0.015475212270610791,
      "grad_norm": 4.25,
      "learning_rate": 5.04e-05,
      "loss": 8.9124,
      "step": 113
    },
    {
      "epoch": 0.015612161051766639,
      "grad_norm": 1.265625,
      "learning_rate": 5.085e-05,
      "loss": 8.877,
      "step": 114
    },
    {
      "epoch": 0.01574910983292249,
      "grad_norm": 5.25,
      "learning_rate": 5.13e-05,
      "loss": 8.873,
      "step": 115
    },
    {
      "epoch": 0.015886058614078335,
      "grad_norm": 6.15625,
      "learning_rate": 5.175e-05,
      "loss": 8.864,
      "step": 116
    },
    {
      "epoch": 0.01602300739523418,
      "grad_norm": 1.7734375,
      "learning_rate": 5.22e-05,
      "loss": 8.8483,
      "step": 117
    },
    {
      "epoch": 0.01615995617639003,
      "grad_norm": 6.0,
      "learning_rate": 5.265e-05,
      "loss": 8.8419,
      "step": 118
    },
    {
      "epoch": 0.016296904957545877,
      "grad_norm": 6.78125,
      "learning_rate": 5.3099999999999996e-05,
      "loss": 8.8099,
      "step": 119
    },
    {
      "epoch": 0.016433853738701727,
      "grad_norm": 1.28125,
      "learning_rate": 5.3549999999999994e-05,
      "loss": 8.804,
      "step": 120
    },
    {
      "epoch": 0.016570802519857573,
      "grad_norm": 1.625,
      "learning_rate": 5.4e-05,
      "loss": 8.771,
      "step": 121
    },
    {
      "epoch": 0.016707751301013422,
      "grad_norm": 2.203125,
      "learning_rate": 5.4449999999999995e-05,
      "loss": 8.7765,
      "step": 122
    },
    {
      "epoch": 0.01684470008216927,
      "grad_norm": 1.2265625,
      "learning_rate": 5.49e-05,
      "loss": 8.7387,
      "step": 123
    },
    {
      "epoch": 0.016981648863325115,
      "grad_norm": 1.53125,
      "learning_rate": 5.535e-05,
      "loss": 8.707,
      "step": 124
    },
    {
      "epoch": 0.017118597644480964,
      "grad_norm": 1.8046875,
      "learning_rate": 5.58e-05,
      "loss": 8.6869,
      "step": 125
    },
    {
      "epoch": 0.01725554642563681,
      "grad_norm": 0.88671875,
      "learning_rate": 5.625e-05,
      "loss": 8.6471,
      "step": 126
    },
    {
      "epoch": 0.01739249520679266,
      "grad_norm": 0.94921875,
      "learning_rate": 5.6699999999999996e-05,
      "loss": 8.6021,
      "step": 127
    },
    {
      "epoch": 0.017529443987948506,
      "grad_norm": 1.140625,
      "learning_rate": 5.715e-05,
      "loss": 8.5724,
      "step": 128
    },
    {
      "epoch": 0.017666392769104356,
      "grad_norm": 1.2109375,
      "learning_rate": 5.76e-05,
      "loss": 8.5576,
      "step": 129
    },
    {
      "epoch": 0.017803341550260202,
      "grad_norm": 2.03125,
      "learning_rate": 5.805e-05,
      "loss": 8.4886,
      "step": 130
    },
    {
      "epoch": 0.017940290331416052,
      "grad_norm": 1.3515625,
      "learning_rate": 5.85e-05,
      "loss": 8.4582,
      "step": 131
    },
    {
      "epoch": 0.018077239112571898,
      "grad_norm": 2.96875,
      "learning_rate": 5.895e-05,
      "loss": 8.3878,
      "step": 132
    },
    {
      "epoch": 0.018214187893727744,
      "grad_norm": 5.3125,
      "learning_rate": 5.94e-05,
      "loss": 8.3508,
      "step": 133
    },
    {
      "epoch": 0.018351136674883594,
      "grad_norm": 2.640625,
      "learning_rate": 5.9850000000000005e-05,
      "loss": 8.2908,
      "step": 134
    },
    {
      "epoch": 0.01848808545603944,
      "grad_norm": 5.90625,
      "learning_rate": 6.03e-05,
      "loss": 8.2544,
      "step": 135
    },
    {
      "epoch": 0.01862503423719529,
      "grad_norm": 2.703125,
      "learning_rate": 6.075e-05,
      "loss": 8.2226,
      "step": 136
    },
    {
      "epoch": 0.018761983018351136,
      "grad_norm": 2.296875,
      "learning_rate": 6.12e-05,
      "loss": 8.1555,
      "step": 137
    },
    {
      "epoch": 0.018898931799506986,
      "grad_norm": 1.953125,
      "learning_rate": 6.165000000000001e-05,
      "loss": 8.1106,
      "step": 138
    },
    {
      "epoch": 0.019035880580662832,
      "grad_norm": 3.28125,
      "learning_rate": 6.21e-05,
      "loss": 8.0535,
      "step": 139
    },
    {
      "epoch": 0.01917282936181868,
      "grad_norm": 4.46875,
      "learning_rate": 6.255e-05,
      "loss": 8.0253,
      "step": 140
    },
    {
      "epoch": 0.019309778142974528,
      "grad_norm": 5.6875,
      "learning_rate": 6.3e-05,
      "loss": 8.0038,
      "step": 141
    },
    {
      "epoch": 0.019446726924130374,
      "grad_norm": 1.59375,
      "learning_rate": 6.345e-05,
      "loss": 7.9679,
      "step": 142
    },
    {
      "epoch": 0.019583675705286224,
      "grad_norm": 2.1875,
      "learning_rate": 6.39e-05,
      "loss": 7.9294,
      "step": 143
    },
    {
      "epoch": 0.01972062448644207,
      "grad_norm": 1.578125,
      "learning_rate": 6.434999999999999e-05,
      "loss": 7.8799,
      "step": 144
    },
    {
      "epoch": 0.01985757326759792,
      "grad_norm": 0.953125,
      "learning_rate": 6.479999999999999e-05,
      "loss": 7.858,
      "step": 145
    },
    {
      "epoch": 0.019994522048753766,
      "grad_norm": 1.46875,
      "learning_rate": 6.525e-05,
      "loss": 7.8392,
      "step": 146
    },
    {
      "epoch": 0.020131470829909615,
      "grad_norm": 1.125,
      "learning_rate": 6.57e-05,
      "loss": 7.8198,
      "step": 147
    },
    {
      "epoch": 0.02026841961106546,
      "grad_norm": 0.71875,
      "learning_rate": 6.615e-05,
      "loss": 7.7902,
      "step": 148
    },
    {
      "epoch": 0.020405368392221308,
      "grad_norm": 0.65625,
      "learning_rate": 6.659999999999999e-05,
      "loss": 7.7661,
      "step": 149
    },
    {
      "epoch": 0.020542317173377157,
      "grad_norm": 0.72265625,
      "learning_rate": 6.704999999999999e-05,
      "loss": 7.7613,
      "step": 150
    },
    {
      "epoch": 0.020679265954533003,
      "grad_norm": 0.71875,
      "learning_rate": 6.75e-05,
      "loss": 7.7461,
      "step": 151
    },
    {
      "epoch": 0.020816214735688853,
      "grad_norm": 0.498046875,
      "learning_rate": 6.795e-05,
      "loss": 7.7445,
      "step": 152
    },
    {
      "epoch": 0.0209531635168447,
      "grad_norm": 0.66796875,
      "learning_rate": 6.84e-05,
      "loss": 7.74,
      "step": 153
    },
    {
      "epoch": 0.02109011229800055,
      "grad_norm": 0.76171875,
      "learning_rate": 6.884999999999999e-05,
      "loss": 7.7104,
      "step": 154
    },
    {
      "epoch": 0.021227061079156395,
      "grad_norm": 0.44140625,
      "learning_rate": 6.93e-05,
      "loss": 7.7171,
      "step": 155
    },
    {
      "epoch": 0.021364009860312245,
      "grad_norm": 0.46484375,
      "learning_rate": 6.975e-05,
      "loss": 7.705,
      "step": 156
    },
    {
      "epoch": 0.02150095864146809,
      "grad_norm": 0.58203125,
      "learning_rate": 7.02e-05,
      "loss": 7.6932,
      "step": 157
    },
    {
      "epoch": 0.021637907422623937,
      "grad_norm": 0.4609375,
      "learning_rate": 7.065e-05,
      "loss": 7.6864,
      "step": 158
    },
    {
      "epoch": 0.021774856203779787,
      "grad_norm": 0.416015625,
      "learning_rate": 7.11e-05,
      "loss": 7.6698,
      "step": 159
    },
    {
      "epoch": 0.021911804984935633,
      "grad_norm": 0.56640625,
      "learning_rate": 7.155e-05,
      "loss": 7.6727,
      "step": 160
    },
    {
      "epoch": 0.022048753766091483,
      "grad_norm": 0.388671875,
      "learning_rate": 7.2e-05,
      "loss": 7.659,
      "step": 161
    },
    {
      "epoch": 0.02218570254724733,
      "grad_norm": 0.4140625,
      "learning_rate": 7.245e-05,
      "loss": 7.6782,
      "step": 162
    },
    {
      "epoch": 0.02232265132840318,
      "grad_norm": 0.578125,
      "learning_rate": 7.29e-05,
      "loss": 7.6583,
      "step": 163
    },
    {
      "epoch": 0.022459600109559025,
      "grad_norm": 0.412109375,
      "learning_rate": 7.335e-05,
      "loss": 7.6621,
      "step": 164
    },
    {
      "epoch": 0.02259654889071487,
      "grad_norm": 0.5234375,
      "learning_rate": 7.38e-05,
      "loss": 7.6505,
      "step": 165
    },
    {
      "epoch": 0.02273349767187072,
      "grad_norm": 0.396484375,
      "learning_rate": 7.425e-05,
      "loss": 7.6688,
      "step": 166
    },
    {
      "epoch": 0.022870446453026567,
      "grad_norm": 0.3984375,
      "learning_rate": 7.47e-05,
      "loss": 7.6718,
      "step": 167
    },
    {
      "epoch": 0.023007395234182416,
      "grad_norm": 0.4140625,
      "learning_rate": 7.515e-05,
      "loss": 7.661,
      "step": 168
    },
    {
      "epoch": 0.023144344015338263,
      "grad_norm": 0.3671875,
      "learning_rate": 7.560000000000001e-05,
      "loss": 7.6532,
      "step": 169
    },
    {
      "epoch": 0.023281292796494112,
      "grad_norm": 0.484375,
      "learning_rate": 7.605e-05,
      "loss": 7.6479,
      "step": 170
    },
    {
      "epoch": 0.02341824157764996,
      "grad_norm": 0.421875,
      "learning_rate": 7.65e-05,
      "loss": 7.6318,
      "step": 171
    },
    {
      "epoch": 0.023555190358805808,
      "grad_norm": 0.45703125,
      "learning_rate": 7.695e-05,
      "loss": 7.6348,
      "step": 172
    },
    {
      "epoch": 0.023692139139961654,
      "grad_norm": 0.421875,
      "learning_rate": 7.74e-05,
      "loss": 7.6213,
      "step": 173
    },
    {
      "epoch": 0.0238290879211175,
      "grad_norm": 0.46875,
      "learning_rate": 7.785e-05,
      "loss": 7.6418,
      "step": 174
    },
    {
      "epoch": 0.02396603670227335,
      "grad_norm": 0.4609375,
      "learning_rate": 7.829999999999999e-05,
      "loss": 7.6313,
      "step": 175
    },
    {
      "epoch": 0.024102985483429196,
      "grad_norm": 0.48046875,
      "learning_rate": 7.874999999999999e-05,
      "loss": 7.6247,
      "step": 176
    },
    {
      "epoch": 0.024239934264585046,
      "grad_norm": 0.400390625,
      "learning_rate": 7.919999999999999e-05,
      "loss": 7.6355,
      "step": 177
    },
    {
      "epoch": 0.024376883045740892,
      "grad_norm": 0.42578125,
      "learning_rate": 7.965e-05,
      "loss": 7.599,
      "step": 178
    },
    {
      "epoch": 0.024513831826896742,
      "grad_norm": 0.4453125,
      "learning_rate": 8.01e-05,
      "loss": 7.6115,
      "step": 179
    },
    {
      "epoch": 0.024650780608052588,
      "grad_norm": 0.3984375,
      "learning_rate": 8.054999999999999e-05,
      "loss": 7.5963,
      "step": 180
    },
    {
      "epoch": 0.024787729389208438,
      "grad_norm": 0.380859375,
      "learning_rate": 8.099999999999999e-05,
      "loss": 7.6176,
      "step": 181
    },
    {
      "epoch": 0.024924678170364284,
      "grad_norm": 0.5,
      "learning_rate": 8.145e-05,
      "loss": 7.5921,
      "step": 182
    },
    {
      "epoch": 0.02506162695152013,
      "grad_norm": 0.458984375,
      "learning_rate": 8.19e-05,
      "loss": 7.6013,
      "step": 183
    },
    {
      "epoch": 0.02519857573267598,
      "grad_norm": 0.67578125,
      "learning_rate": 8.235e-05,
      "loss": 7.6056,
      "step": 184
    },
    {
      "epoch": 0.025335524513831826,
      "grad_norm": 0.4296875,
      "learning_rate": 8.28e-05,
      "loss": 7.6033,
      "step": 185
    },
    {
      "epoch": 0.025472473294987676,
      "grad_norm": 0.53515625,
      "learning_rate": 8.324999999999999e-05,
      "loss": 7.6115,
      "step": 186
    },
    {
      "epoch": 0.025609422076143522,
      "grad_norm": 0.466796875,
      "learning_rate": 8.37e-05,
      "loss": 7.6033,
      "step": 187
    },
    {
      "epoch": 0.02574637085729937,
      "grad_norm": 0.73828125,
      "learning_rate": 8.415e-05,
      "loss": 7.5914,
      "step": 188
    },
    {
      "epoch": 0.025883319638455218,
      "grad_norm": 0.8515625,
      "learning_rate": 8.46e-05,
      "loss": 7.5743,
      "step": 189
    },
    {
      "epoch": 0.026020268419611064,
      "grad_norm": 0.49609375,
      "learning_rate": 8.505e-05,
      "loss": 7.5832,
      "step": 190
    },
    {
      "epoch": 0.026157217200766913,
      "grad_norm": 0.6953125,
      "learning_rate": 8.55e-05,
      "loss": 7.583,
      "step": 191
    },
    {
      "epoch": 0.02629416598192276,
      "grad_norm": 0.80859375,
      "learning_rate": 8.595e-05,
      "loss": 7.5877,
      "step": 192
    },
    {
      "epoch": 0.02643111476307861,
      "grad_norm": 0.58203125,
      "learning_rate": 8.64e-05,
      "loss": 7.5609,
      "step": 193
    },
    {
      "epoch": 0.026568063544234456,
      "grad_norm": 0.703125,
      "learning_rate": 8.685e-05,
      "loss": 7.5698,
      "step": 194
    },
    {
      "epoch": 0.026705012325390305,
      "grad_norm": 1.1640625,
      "learning_rate": 8.73e-05,
      "loss": 7.5508,
      "step": 195
    },
    {
      "epoch": 0.02684196110654615,
      "grad_norm": 1.1640625,
      "learning_rate": 8.775e-05,
      "loss": 7.5663,
      "step": 196
    },
    {
      "epoch": 0.026978909887702,
      "grad_norm": 0.8203125,
      "learning_rate": 8.82e-05,
      "loss": 7.565,
      "step": 197
    },
    {
      "epoch": 0.027115858668857847,
      "grad_norm": 1.125,
      "learning_rate": 8.865e-05,
      "loss": 7.5426,
      "step": 198
    },
    {
      "epoch": 0.027252807450013693,
      "grad_norm": 1.015625,
      "learning_rate": 8.91e-05,
      "loss": 7.5369,
      "step": 199
    },
    {
      "epoch": 0.027389756231169543,
      "grad_norm": 0.9375,
      "learning_rate": 8.955000000000001e-05,
      "loss": 7.5553,
      "step": 200
    },
    {
      "epoch": 0.02752670501232539,
      "grad_norm": 1.234375,
      "learning_rate": 9e-05,
      "loss": 7.525,
      "step": 201
    },
    {
      "epoch": 0.02766365379348124,
      "grad_norm": 2.953125,
      "learning_rate": 9.045e-05,
      "loss": 7.5365,
      "step": 202
    },
    {
      "epoch": 0.027800602574637085,
      "grad_norm": 6.625,
      "learning_rate": 9.09e-05,
      "loss": 7.5571,
      "step": 203
    },
    {
      "epoch": 0.027937551355792935,
      "grad_norm": 1.5859375,
      "learning_rate": 9.135e-05,
      "loss": 7.5367,
      "step": 204
    },
    {
      "epoch": 0.02807450013694878,
      "grad_norm": 12.25,
      "learning_rate": 9.18e-05,
      "loss": 7.5842,
      "step": 205
    },
    {
      "epoch": 0.02821144891810463,
      "grad_norm": 7.625,
      "learning_rate": 9.224999999999999e-05,
      "loss": 7.5458,
      "step": 206
    },
    {
      "epoch": 0.028348397699260477,
      "grad_norm": 11.1875,
      "learning_rate": 9.269999999999999e-05,
      "loss": 7.5844,
      "step": 207
    },
    {
      "epoch": 0.028485346480416323,
      "grad_norm": 5.125,
      "learning_rate": 9.314999999999999e-05,
      "loss": 7.5639,
      "step": 208
    },
    {
      "epoch": 0.028622295261572173,
      "grad_norm": 7.03125,
      "learning_rate": 9.36e-05,
      "loss": 7.5704,
      "step": 209
    },
    {
      "epoch": 0.02875924404272802,
      "grad_norm": 5.5,
      "learning_rate": 9.405e-05,
      "loss": 7.5644,
      "step": 210
    },
    {
      "epoch": 0.02889619282388387,
      "grad_norm": 2.90625,
      "learning_rate": 9.449999999999999e-05,
      "loss": 7.5353,
      "step": 211
    },
    {
      "epoch": 0.029033141605039715,
      "grad_norm": 5.9375,
      "learning_rate": 9.494999999999999e-05,
      "loss": 7.5404,
      "step": 212
    },
    {
      "epoch": 0.029170090386195564,
      "grad_norm": 5.0625,
      "learning_rate": 9.54e-05,
      "loss": 7.569,
      "step": 213
    },
    {
      "epoch": 0.02930703916735141,
      "grad_norm": 6.1875,
      "learning_rate": 9.585e-05,
      "loss": 7.5659,
      "step": 214
    },
    {
      "epoch": 0.029443987948507257,
      "grad_norm": 2.046875,
      "learning_rate": 9.63e-05,
      "loss": 7.55,
      "step": 215
    },
    {
      "epoch": 0.029580936729663106,
      "grad_norm": 4.1875,
      "learning_rate": 9.675e-05,
      "loss": 7.5451,
      "step": 216
    },
    {
      "epoch": 0.029717885510818953,
      "grad_norm": 3.09375,
      "learning_rate": 9.719999999999999e-05,
      "loss": 7.5574,
      "step": 217
    },
    {
      "epoch": 0.029854834291974802,
      "grad_norm": 2.71875,
      "learning_rate": 9.765e-05,
      "loss": 7.5368,
      "step": 218
    },
    {
      "epoch": 0.02999178307313065,
      "grad_norm": 2.59375,
      "learning_rate": 9.81e-05,
      "loss": 7.5194,
      "step": 219
    },
    {
      "epoch": 0.030128731854286498,
      "grad_norm": 3.765625,
      "learning_rate": 9.855e-05,
      "loss": 7.5142,
      "step": 220
    },
    {
      "epoch": 0.030265680635442344,
      "grad_norm": 2.25,
      "learning_rate": 9.9e-05,
      "loss": 7.5071,
      "step": 221
    },
    {
      "epoch": 0.030402629416598194,
      "grad_norm": 2.09375,
      "learning_rate": 9.944999999999999e-05,
      "loss": 7.5005,
      "step": 222
    },
    {
      "epoch": 0.03053957819775404,
      "grad_norm": 1.7890625,
      "learning_rate": 9.99e-05,
      "loss": 7.4989,
      "step": 223
    },
    {
      "epoch": 0.030676526978909886,
      "grad_norm": 1.515625,
      "learning_rate": 0.00010035,
      "loss": 7.4883,
      "step": 224
    },
    {
      "epoch": 0.030813475760065736,
      "grad_norm": 1.671875,
      "learning_rate": 0.0001008,
      "loss": 7.4967,
      "step": 225
    },
    {
      "epoch": 0.030950424541221582,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010125,
      "loss": 7.509,
      "step": 226
    },
    {
      "epoch": 0.031087373322377432,
      "grad_norm": 2.0625,
      "learning_rate": 0.0001017,
      "loss": 7.4841,
      "step": 227
    },
    {
      "epoch": 0.031224322103533278,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010215,
      "loss": 7.4966,
      "step": 228
    },
    {
      "epoch": 0.031361270884689124,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001026,
      "loss": 7.4609,
      "step": 229
    },
    {
      "epoch": 0.03149821966584498,
      "grad_norm": 2.09375,
      "learning_rate": 0.00010305,
      "loss": 7.4663,
      "step": 230
    },
    {
      "epoch": 0.031635168447000823,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001035,
      "loss": 7.4734,
      "step": 231
    },
    {
      "epoch": 0.03177211722815667,
      "grad_norm": 1.28125,
      "learning_rate": 0.00010395,
      "loss": 7.4865,
      "step": 232
    },
    {
      "epoch": 0.031909066009312516,
      "grad_norm": 0.875,
      "learning_rate": 0.0001044,
      "loss": 7.4516,
      "step": 233
    },
    {
      "epoch": 0.03204601479046836,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010485,
      "loss": 7.4508,
      "step": 234
    },
    {
      "epoch": 0.032182963571624215,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001053,
      "loss": 7.4599,
      "step": 235
    },
    {
      "epoch": 0.03231991235278006,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00010575,
      "loss": 7.4596,
      "step": 236
    },
    {
      "epoch": 0.03245686113393591,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010619999999999999,
      "loss": 7.4321,
      "step": 237
    },
    {
      "epoch": 0.032593809915091754,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010664999999999999,
      "loss": 7.4459,
      "step": 238
    },
    {
      "epoch": 0.03273075869624761,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010709999999999999,
      "loss": 7.4508,
      "step": 239
    },
    {
      "epoch": 0.03286770747740345,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010755,
      "loss": 7.4659,
      "step": 240
    },
    {
      "epoch": 0.0330046562585593,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000108,
      "loss": 7.4301,
      "step": 241
    },
    {
      "epoch": 0.033141605039715145,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010845,
      "loss": 7.4223,
      "step": 242
    },
    {
      "epoch": 0.03327855382087099,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010889999999999999,
      "loss": 7.4104,
      "step": 243
    },
    {
      "epoch": 0.033415502602026845,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010934999999999999,
      "loss": 7.4443,
      "step": 244
    },
    {
      "epoch": 0.03355245138318269,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001098,
      "loss": 7.4028,
      "step": 245
    },
    {
      "epoch": 0.03368940016433854,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011025,
      "loss": 7.445,
      "step": 246
    },
    {
      "epoch": 0.03382634894549438,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001107,
      "loss": 7.4077,
      "step": 247
    },
    {
      "epoch": 0.03396329772665023,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011114999999999999,
      "loss": 7.4005,
      "step": 248
    },
    {
      "epoch": 0.03410024650780608,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001116,
      "loss": 7.4164,
      "step": 249
    },
    {
      "epoch": 0.03423719528896193,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011205,
      "loss": 7.4165,
      "step": 250
    },
    {
      "epoch": 0.034374144070117775,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001125,
      "loss": 7.4154,
      "step": 251
    },
    {
      "epoch": 0.03451109285127362,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011295,
      "loss": 7.4106,
      "step": 252
    },
    {
      "epoch": 0.034648041632429474,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011339999999999999,
      "loss": 7.3941,
      "step": 253
    },
    {
      "epoch": 0.03478499041358532,
      "grad_norm": 0.90625,
      "learning_rate": 0.00011385,
      "loss": 7.4044,
      "step": 254
    },
    {
      "epoch": 0.03492193919474117,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001143,
      "loss": 7.4212,
      "step": 255
    },
    {
      "epoch": 0.03505888797589701,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011475,
      "loss": 7.3939,
      "step": 256
    },
    {
      "epoch": 0.03519583675705286,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001152,
      "loss": 7.3826,
      "step": 257
    },
    {
      "epoch": 0.03533278553820871,
      "grad_norm": 1.0,
      "learning_rate": 0.00011565,
      "loss": 7.3731,
      "step": 258
    },
    {
      "epoch": 0.03546973431936456,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001161,
      "loss": 7.3718,
      "step": 259
    },
    {
      "epoch": 0.035606683100520405,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00011655,
      "loss": 7.4059,
      "step": 260
    },
    {
      "epoch": 0.03574363188167625,
      "grad_norm": 0.65625,
      "learning_rate": 0.000117,
      "loss": 7.3723,
      "step": 261
    },
    {
      "epoch": 0.035880580662832104,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00011745,
      "loss": 7.3891,
      "step": 262
    },
    {
      "epoch": 0.03601752944398795,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001179,
      "loss": 7.3809,
      "step": 263
    },
    {
      "epoch": 0.036154478225143796,
      "grad_norm": 1.9921875,
      "learning_rate": 0.00011835,
      "loss": 7.3931,
      "step": 264
    },
    {
      "epoch": 0.03629142700629964,
      "grad_norm": 4.25,
      "learning_rate": 0.0001188,
      "loss": 7.4369,
      "step": 265
    },
    {
      "epoch": 0.03642837578745549,
      "grad_norm": 6.09375,
      "learning_rate": 0.00011925,
      "loss": 7.4306,
      "step": 266
    },
    {
      "epoch": 0.03656532456861134,
      "grad_norm": 1.03125,
      "learning_rate": 0.00011970000000000001,
      "loss": 7.3932,
      "step": 267
    },
    {
      "epoch": 0.03670227334976719,
      "grad_norm": 5.15625,
      "learning_rate": 0.00012015,
      "loss": 7.465,
      "step": 268
    },
    {
      "epoch": 0.036839222130923034,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001206,
      "loss": 7.3933,
      "step": 269
    },
    {
      "epoch": 0.03697617091207888,
      "grad_norm": 7.28125,
      "learning_rate": 0.00012105,
      "loss": 7.4613,
      "step": 270
    },
    {
      "epoch": 0.037113119693234733,
      "grad_norm": 3.625,
      "learning_rate": 0.0001215,
      "loss": 7.4243,
      "step": 271
    },
    {
      "epoch": 0.03725006847439058,
      "grad_norm": 7.40625,
      "learning_rate": 0.00012195000000000001,
      "loss": 7.5064,
      "step": 272
    },
    {
      "epoch": 0.037387017255546426,
      "grad_norm": 4.15625,
      "learning_rate": 0.0001224,
      "loss": 7.4799,
      "step": 273
    },
    {
      "epoch": 0.03752396603670227,
      "grad_norm": 4.3125,
      "learning_rate": 0.00012285,
      "loss": 7.4631,
      "step": 274
    },
    {
      "epoch": 0.03766091481785812,
      "grad_norm": 2.515625,
      "learning_rate": 0.00012330000000000002,
      "loss": 7.42,
      "step": 275
    },
    {
      "epoch": 0.03779786359901397,
      "grad_norm": 3.109375,
      "learning_rate": 0.00012375,
      "loss": 7.3992,
      "step": 276
    },
    {
      "epoch": 0.03793481238016982,
      "grad_norm": 4.34375,
      "learning_rate": 0.0001242,
      "loss": 7.4317,
      "step": 277
    },
    {
      "epoch": 0.038071761161325664,
      "grad_norm": 1.796875,
      "learning_rate": 0.00012465,
      "loss": 7.4234,
      "step": 278
    },
    {
      "epoch": 0.03820870994248151,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001251,
      "loss": 7.4179,
      "step": 279
    },
    {
      "epoch": 0.03834565872363736,
      "grad_norm": 1.1875,
      "learning_rate": 0.00012555000000000002,
      "loss": 7.4252,
      "step": 280
    },
    {
      "epoch": 0.03848260750479321,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000126,
      "loss": 7.3747,
      "step": 281
    },
    {
      "epoch": 0.038619556285949055,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012645,
      "loss": 7.3955,
      "step": 282
    },
    {
      "epoch": 0.0387565050671049,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001269,
      "loss": 7.3809,
      "step": 283
    },
    {
      "epoch": 0.03889345384826075,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012734999999999998,
      "loss": 7.3789,
      "step": 284
    },
    {
      "epoch": 0.0390304026294166,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001278,
      "loss": 7.3761,
      "step": 285
    },
    {
      "epoch": 0.03916735141057245,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012824999999999997,
      "loss": 7.3591,
      "step": 286
    },
    {
      "epoch": 0.03930430019172829,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012869999999999998,
      "loss": 7.3583,
      "step": 287
    },
    {
      "epoch": 0.03944124897288414,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012915,
      "loss": 7.3659,
      "step": 288
    },
    {
      "epoch": 0.039578197754039986,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012959999999999998,
      "loss": 7.3329,
      "step": 289
    },
    {
      "epoch": 0.03971514653519584,
      "grad_norm": 0.90625,
      "learning_rate": 0.00013005,
      "loss": 7.3426,
      "step": 290
    },
    {
      "epoch": 0.039852095316351685,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001305,
      "loss": 7.3585,
      "step": 291
    },
    {
      "epoch": 0.03998904409750753,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013094999999999998,
      "loss": 7.3359,
      "step": 292
    },
    {
      "epoch": 0.04012599287866338,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001314,
      "loss": 7.3577,
      "step": 293
    },
    {
      "epoch": 0.04026294165981923,
      "grad_norm": 0.921875,
      "learning_rate": 0.00013184999999999998,
      "loss": 7.3644,
      "step": 294
    },
    {
      "epoch": 0.04039989044097508,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001323,
      "loss": 7.337,
      "step": 295
    },
    {
      "epoch": 0.04053683922213092,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013275,
      "loss": 7.35,
      "step": 296
    },
    {
      "epoch": 0.04067378800328677,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00013319999999999999,
      "loss": 7.3006,
      "step": 297
    },
    {
      "epoch": 0.040810736784442615,
      "grad_norm": 0.875,
      "learning_rate": 0.00013365,
      "loss": 7.3397,
      "step": 298
    },
    {
      "epoch": 0.04094768556559847,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013409999999999998,
      "loss": 7.3118,
      "step": 299
    },
    {
      "epoch": 0.041084634346754315,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013455,
      "loss": 7.3419,
      "step": 300
    },
    {
      "epoch": 0.04122158312791016,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000135,
      "loss": 7.305,
      "step": 301
    },
    {
      "epoch": 0.04135853190906601,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00013544999999999999,
      "loss": 7.3169,
      "step": 302
    },
    {
      "epoch": 0.04149548069022186,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001359,
      "loss": 7.2975,
      "step": 303
    },
    {
      "epoch": 0.041632429471377706,
      "grad_norm": 1.109375,
      "learning_rate": 0.00013634999999999998,
      "loss": 7.3062,
      "step": 304
    },
    {
      "epoch": 0.04176937825253355,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001368,
      "loss": 7.3032,
      "step": 305
    },
    {
      "epoch": 0.0419063270336894,
      "grad_norm": 0.984375,
      "learning_rate": 0.00013725,
      "loss": 7.2905,
      "step": 306
    },
    {
      "epoch": 0.042043275814845245,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00013769999999999999,
      "loss": 7.2991,
      "step": 307
    },
    {
      "epoch": 0.0421802245960011,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00013815,
      "loss": 7.2832,
      "step": 308
    },
    {
      "epoch": 0.042317173377156944,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001386,
      "loss": 7.2954,
      "step": 309
    },
    {
      "epoch": 0.04245412215831279,
      "grad_norm": 1.953125,
      "learning_rate": 0.00013905,
      "loss": 7.2931,
      "step": 310
    },
    {
      "epoch": 0.04259107093946864,
      "grad_norm": 3.703125,
      "learning_rate": 0.0001395,
      "loss": 7.3623,
      "step": 311
    },
    {
      "epoch": 0.04272801972062449,
      "grad_norm": 6.3125,
      "learning_rate": 0.00013995,
      "loss": 7.4116,
      "step": 312
    },
    {
      "epoch": 0.042864968501780336,
      "grad_norm": 2.546875,
      "learning_rate": 0.0001404,
      "loss": 7.3347,
      "step": 313
    },
    {
      "epoch": 0.04300191728293618,
      "grad_norm": 4.375,
      "learning_rate": 0.00014085,
      "loss": 7.3447,
      "step": 314
    },
    {
      "epoch": 0.04313886606409203,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0001413,
      "loss": 7.2944,
      "step": 315
    },
    {
      "epoch": 0.043275814845247874,
      "grad_norm": 2.765625,
      "learning_rate": 0.00014175,
      "loss": 7.3247,
      "step": 316
    },
    {
      "epoch": 0.04341276362640373,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0001422,
      "loss": 7.3162,
      "step": 317
    },
    {
      "epoch": 0.043549712407559574,
      "grad_norm": 2.5,
      "learning_rate": 0.00014265,
      "loss": 7.305,
      "step": 318
    },
    {
      "epoch": 0.04368666118871542,
      "grad_norm": 2.140625,
      "learning_rate": 0.0001431,
      "loss": 7.3127,
      "step": 319
    },
    {
      "epoch": 0.043823609969871266,
      "grad_norm": 1.578125,
      "learning_rate": 0.00014355,
      "loss": 7.2805,
      "step": 320
    },
    {
      "epoch": 0.04396055875102712,
      "grad_norm": 2.296875,
      "learning_rate": 0.000144,
      "loss": 7.28,
      "step": 321
    },
    {
      "epoch": 0.044097507532182965,
      "grad_norm": 2.59375,
      "learning_rate": 0.00014445,
      "loss": 7.2999,
      "step": 322
    },
    {
      "epoch": 0.04423445631333881,
      "grad_norm": 3.1875,
      "learning_rate": 0.0001449,
      "loss": 7.2781,
      "step": 323
    },
    {
      "epoch": 0.04437140509449466,
      "grad_norm": 2.703125,
      "learning_rate": 0.00014535,
      "loss": 7.2669,
      "step": 324
    },
    {
      "epoch": 0.044508353875650504,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001458,
      "loss": 7.2577,
      "step": 325
    },
    {
      "epoch": 0.04464530265680636,
      "grad_norm": 1.578125,
      "learning_rate": 0.00014625,
      "loss": 7.2615,
      "step": 326
    },
    {
      "epoch": 0.0447822514379622,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001467,
      "loss": 7.2468,
      "step": 327
    },
    {
      "epoch": 0.04491920021911805,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014715,
      "loss": 7.2526,
      "step": 328
    },
    {
      "epoch": 0.045056149000273896,
      "grad_norm": 1.8125,
      "learning_rate": 0.0001476,
      "loss": 7.2539,
      "step": 329
    },
    {
      "epoch": 0.04519309778142974,
      "grad_norm": 1.390625,
      "learning_rate": 0.00014805,
      "loss": 7.2222,
      "step": 330
    },
    {
      "epoch": 0.045330046562585595,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001485,
      "loss": 7.2241,
      "step": 331
    },
    {
      "epoch": 0.04546699534374144,
      "grad_norm": 1.21875,
      "learning_rate": 0.00014895000000000002,
      "loss": 7.1853,
      "step": 332
    },
    {
      "epoch": 0.04560394412489729,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001494,
      "loss": 7.221,
      "step": 333
    },
    {
      "epoch": 0.045740892906053134,
      "grad_norm": 2.140625,
      "learning_rate": 0.00014985,
      "loss": 7.162,
      "step": 334
    },
    {
      "epoch": 0.04587784168720899,
      "grad_norm": 1.484375,
      "learning_rate": 0.0001503,
      "loss": 7.2107,
      "step": 335
    },
    {
      "epoch": 0.04601479046836483,
      "grad_norm": 2.109375,
      "learning_rate": 0.00015075,
      "loss": 7.2047,
      "step": 336
    },
    {
      "epoch": 0.04615173924952068,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00015120000000000002,
      "loss": 7.2089,
      "step": 337
    },
    {
      "epoch": 0.046288688030676525,
      "grad_norm": 1.546875,
      "learning_rate": 0.00015165,
      "loss": 7.2173,
      "step": 338
    },
    {
      "epoch": 0.04642563681183237,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001521,
      "loss": 7.1888,
      "step": 339
    },
    {
      "epoch": 0.046562585592988225,
      "grad_norm": 1.796875,
      "learning_rate": 0.00015255,
      "loss": 7.1815,
      "step": 340
    },
    {
      "epoch": 0.04669953437414407,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000153,
      "loss": 7.173,
      "step": 341
    },
    {
      "epoch": 0.04683648315529992,
      "grad_norm": 1.53125,
      "learning_rate": 0.00015345000000000002,
      "loss": 7.1503,
      "step": 342
    },
    {
      "epoch": 0.04697343193645576,
      "grad_norm": 1.953125,
      "learning_rate": 0.0001539,
      "loss": 7.1616,
      "step": 343
    },
    {
      "epoch": 0.047110380717611616,
      "grad_norm": 1.78125,
      "learning_rate": 0.00015435,
      "loss": 7.1626,
      "step": 344
    },
    {
      "epoch": 0.04724732949876746,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001548,
      "loss": 7.1728,
      "step": 345
    },
    {
      "epoch": 0.04738427827992331,
      "grad_norm": 1.328125,
      "learning_rate": 0.00015524999999999998,
      "loss": 7.1418,
      "step": 346
    },
    {
      "epoch": 0.047521227061079155,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001557,
      "loss": 7.1487,
      "step": 347
    },
    {
      "epoch": 0.047658175842235,
      "grad_norm": 1.71875,
      "learning_rate": 0.00015614999999999997,
      "loss": 7.1275,
      "step": 348
    },
    {
      "epoch": 0.047795124623390854,
      "grad_norm": 2.25,
      "learning_rate": 0.00015659999999999998,
      "loss": 7.159,
      "step": 349
    },
    {
      "epoch": 0.0479320734045467,
      "grad_norm": 2.875,
      "learning_rate": 0.00015705,
      "loss": 7.1375,
      "step": 350
    },
    {
      "epoch": 0.04806902218570255,
      "grad_norm": 4.03125,
      "learning_rate": 0.00015749999999999998,
      "loss": 7.1466,
      "step": 351
    },
    {
      "epoch": 0.04820597096685839,
      "grad_norm": 4.90625,
      "learning_rate": 0.00015795,
      "loss": 7.1484,
      "step": 352
    },
    {
      "epoch": 0.048342919748014246,
      "grad_norm": 1.65625,
      "learning_rate": 0.00015839999999999997,
      "loss": 7.1663,
      "step": 353
    },
    {
      "epoch": 0.04847986852917009,
      "grad_norm": 3.3125,
      "learning_rate": 0.00015884999999999999,
      "loss": 7.1589,
      "step": 354
    },
    {
      "epoch": 0.04861681731032594,
      "grad_norm": 3.875,
      "learning_rate": 0.0001593,
      "loss": 7.1078,
      "step": 355
    },
    {
      "epoch": 0.048753766091481784,
      "grad_norm": 1.515625,
      "learning_rate": 0.00015974999999999998,
      "loss": 7.094,
      "step": 356
    },
    {
      "epoch": 0.04889071487263763,
      "grad_norm": 2.828125,
      "learning_rate": 0.0001602,
      "loss": 7.175,
      "step": 357
    },
    {
      "epoch": 0.049027663653793484,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00016065,
      "loss": 7.116,
      "step": 358
    },
    {
      "epoch": 0.04916461243494933,
      "grad_norm": 2.28125,
      "learning_rate": 0.00016109999999999999,
      "loss": 7.0983,
      "step": 359
    },
    {
      "epoch": 0.049301561216105176,
      "grad_norm": 3.25,
      "learning_rate": 0.00016155,
      "loss": 7.1232,
      "step": 360
    },
    {
      "epoch": 0.04943850999726102,
      "grad_norm": 4.96875,
      "learning_rate": 0.00016199999999999998,
      "loss": 7.1203,
      "step": 361
    },
    {
      "epoch": 0.049575458778416875,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00016245,
      "loss": 7.107,
      "step": 362
    },
    {
      "epoch": 0.04971240755957272,
      "grad_norm": 3.625,
      "learning_rate": 0.0001629,
      "loss": 7.1032,
      "step": 363
    },
    {
      "epoch": 0.04984935634072857,
      "grad_norm": 3.296875,
      "learning_rate": 0.00016334999999999999,
      "loss": 7.0984,
      "step": 364
    },
    {
      "epoch": 0.049986305121884414,
      "grad_norm": 3.484375,
      "learning_rate": 0.0001638,
      "loss": 7.1367,
      "step": 365
    },
    {
      "epoch": 0.05012325390304026,
      "grad_norm": 2.296875,
      "learning_rate": 0.00016424999999999998,
      "loss": 7.0887,
      "step": 366
    },
    {
      "epoch": 0.05026020268419611,
      "grad_norm": 2.65625,
      "learning_rate": 0.0001647,
      "loss": 7.1159,
      "step": 367
    },
    {
      "epoch": 0.05039715146535196,
      "grad_norm": 4.09375,
      "learning_rate": 0.00016515,
      "loss": 7.1181,
      "step": 368
    },
    {
      "epoch": 0.050534100246507806,
      "grad_norm": 3.109375,
      "learning_rate": 0.0001656,
      "loss": 7.0935,
      "step": 369
    },
    {
      "epoch": 0.05067104902766365,
      "grad_norm": 2.984375,
      "learning_rate": 0.00016605,
      "loss": 7.0573,
      "step": 370
    },
    {
      "epoch": 0.0508079978088195,
      "grad_norm": 3.328125,
      "learning_rate": 0.00016649999999999998,
      "loss": 7.0839,
      "step": 371
    },
    {
      "epoch": 0.05094494658997535,
      "grad_norm": 3.4375,
      "learning_rate": 0.00016695,
      "loss": 7.0677,
      "step": 372
    },
    {
      "epoch": 0.0510818953711312,
      "grad_norm": 2.03125,
      "learning_rate": 0.0001674,
      "loss": 7.0694,
      "step": 373
    },
    {
      "epoch": 0.051218844152287044,
      "grad_norm": 1.40625,
      "learning_rate": 0.00016785,
      "loss": 7.0568,
      "step": 374
    },
    {
      "epoch": 0.05135579293344289,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001683,
      "loss": 7.0514,
      "step": 375
    },
    {
      "epoch": 0.05149274171459874,
      "grad_norm": 4.21875,
      "learning_rate": 0.00016875,
      "loss": 7.0549,
      "step": 376
    },
    {
      "epoch": 0.05162969049575459,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001692,
      "loss": 7.0422,
      "step": 377
    },
    {
      "epoch": 0.051766639276910435,
      "grad_norm": 4.09375,
      "learning_rate": 0.00016965,
      "loss": 7.0695,
      "step": 378
    },
    {
      "epoch": 0.05190358805806628,
      "grad_norm": 2.09375,
      "learning_rate": 0.0001701,
      "loss": 7.0432,
      "step": 379
    },
    {
      "epoch": 0.05204053683922213,
      "grad_norm": 2.734375,
      "learning_rate": 0.00017055,
      "loss": 7.0137,
      "step": 380
    },
    {
      "epoch": 0.05217748562037798,
      "grad_norm": 3.296875,
      "learning_rate": 0.000171,
      "loss": 7.0454,
      "step": 381
    },
    {
      "epoch": 0.05231443440153383,
      "grad_norm": 3.9375,
      "learning_rate": 0.00017145,
      "loss": 7.0433,
      "step": 382
    },
    {
      "epoch": 0.05245138318268967,
      "grad_norm": 2.140625,
      "learning_rate": 0.0001719,
      "loss": 7.0451,
      "step": 383
    },
    {
      "epoch": 0.05258833196384552,
      "grad_norm": 2.390625,
      "learning_rate": 0.00017235,
      "loss": 7.0138,
      "step": 384
    },
    {
      "epoch": 0.05272528074500137,
      "grad_norm": 2.703125,
      "learning_rate": 0.0001728,
      "loss": 7.0189,
      "step": 385
    },
    {
      "epoch": 0.05286222952615722,
      "grad_norm": 2.703125,
      "learning_rate": 0.00017325,
      "loss": 7.0061,
      "step": 386
    },
    {
      "epoch": 0.052999178307313065,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0001737,
      "loss": 6.9633,
      "step": 387
    },
    {
      "epoch": 0.05313612708846891,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00017415,
      "loss": 7.0036,
      "step": 388
    },
    {
      "epoch": 0.05327307586962476,
      "grad_norm": 1.984375,
      "learning_rate": 0.0001746,
      "loss": 6.9976,
      "step": 389
    },
    {
      "epoch": 0.05341002465078061,
      "grad_norm": 1.515625,
      "learning_rate": 0.00017505,
      "loss": 7.0004,
      "step": 390
    },
    {
      "epoch": 0.05354697343193646,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001755,
      "loss": 6.9977,
      "step": 391
    },
    {
      "epoch": 0.0536839222130923,
      "grad_norm": 2.15625,
      "learning_rate": 0.00017595,
      "loss": 6.9566,
      "step": 392
    },
    {
      "epoch": 0.05382087099424815,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0001764,
      "loss": 6.9667,
      "step": 393
    },
    {
      "epoch": 0.053957819775404,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00017685,
      "loss": 6.9632,
      "step": 394
    },
    {
      "epoch": 0.05409476855655985,
      "grad_norm": 2.125,
      "learning_rate": 0.0001773,
      "loss": 6.9626,
      "step": 395
    },
    {
      "epoch": 0.054231717337715694,
      "grad_norm": 1.796875,
      "learning_rate": 0.00017775,
      "loss": 6.9722,
      "step": 396
    },
    {
      "epoch": 0.05436866611887154,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001782,
      "loss": 6.9787,
      "step": 397
    },
    {
      "epoch": 0.05450561490002739,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00017865,
      "loss": 6.9582,
      "step": 398
    },
    {
      "epoch": 0.05464256368118324,
      "grad_norm": 2.4375,
      "learning_rate": 0.00017910000000000002,
      "loss": 6.9613,
      "step": 399
    },
    {
      "epoch": 0.054779512462339086,
      "grad_norm": 3.796875,
      "learning_rate": 0.00017955,
      "loss": 6.9717,
      "step": 400
    },
    {
      "epoch": 0.05491646124349493,
      "grad_norm": 3.0625,
      "learning_rate": 0.00018,
      "loss": 6.9433,
      "step": 401
    },
    {
      "epoch": 0.05505341002465078,
      "grad_norm": 3.421875,
      "learning_rate": 0.00018045,
      "loss": 6.9854,
      "step": 402
    },
    {
      "epoch": 0.05519035880580663,
      "grad_norm": 5.40625,
      "learning_rate": 0.0001809,
      "loss": 6.9832,
      "step": 403
    },
    {
      "epoch": 0.05532730758696248,
      "grad_norm": 2.234375,
      "learning_rate": 0.00018135000000000002,
      "loss": 6.9134,
      "step": 404
    },
    {
      "epoch": 0.055464256368118324,
      "grad_norm": 4.3125,
      "learning_rate": 0.0001818,
      "loss": 6.9525,
      "step": 405
    },
    {
      "epoch": 0.05560120514927417,
      "grad_norm": 2.75,
      "learning_rate": 0.00018225,
      "loss": 6.9314,
      "step": 406
    },
    {
      "epoch": 0.055738153930430016,
      "grad_norm": 2.921875,
      "learning_rate": 0.0001827,
      "loss": 6.9602,
      "step": 407
    },
    {
      "epoch": 0.05587510271158587,
      "grad_norm": 2.609375,
      "learning_rate": 0.00018314999999999998,
      "loss": 6.9738,
      "step": 408
    },
    {
      "epoch": 0.056012051492741716,
      "grad_norm": 2.984375,
      "learning_rate": 0.0001836,
      "loss": 6.9339,
      "step": 409
    },
    {
      "epoch": 0.05614900027389756,
      "grad_norm": 4.875,
      "learning_rate": 0.00018404999999999997,
      "loss": 6.9386,
      "step": 410
    },
    {
      "epoch": 0.05628594905505341,
      "grad_norm": 2.0625,
      "learning_rate": 0.00018449999999999999,
      "loss": 6.9322,
      "step": 411
    },
    {
      "epoch": 0.05642289783620926,
      "grad_norm": 4.6875,
      "learning_rate": 0.00018495,
      "loss": 6.9688,
      "step": 412
    },
    {
      "epoch": 0.05655984661736511,
      "grad_norm": 1.609375,
      "learning_rate": 0.00018539999999999998,
      "loss": 6.9453,
      "step": 413
    },
    {
      "epoch": 0.056696795398520954,
      "grad_norm": 2.46875,
      "learning_rate": 0.00018585,
      "loss": 6.9384,
      "step": 414
    },
    {
      "epoch": 0.0568337441796768,
      "grad_norm": 1.515625,
      "learning_rate": 0.00018629999999999997,
      "loss": 6.9052,
      "step": 415
    },
    {
      "epoch": 0.056970692960832646,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00018674999999999999,
      "loss": 6.9232,
      "step": 416
    },
    {
      "epoch": 0.0571076417419885,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0001872,
      "loss": 6.9229,
      "step": 417
    },
    {
      "epoch": 0.057244590523144345,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00018764999999999998,
      "loss": 6.9022,
      "step": 418
    },
    {
      "epoch": 0.05738153930430019,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001881,
      "loss": 6.8865,
      "step": 419
    },
    {
      "epoch": 0.05751848808545604,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00018854999999999998,
      "loss": 6.9306,
      "step": 420
    },
    {
      "epoch": 0.057655436866611884,
      "grad_norm": 1.5625,
      "learning_rate": 0.00018899999999999999,
      "loss": 6.9103,
      "step": 421
    },
    {
      "epoch": 0.05779238564776774,
      "grad_norm": 1.34375,
      "learning_rate": 0.00018945,
      "loss": 6.8836,
      "step": 422
    },
    {
      "epoch": 0.05792933442892358,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00018989999999999998,
      "loss": 6.8787,
      "step": 423
    },
    {
      "epoch": 0.05806628321007943,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019035,
      "loss": 6.8797,
      "step": 424
    },
    {
      "epoch": 0.058203231991235276,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001908,
      "loss": 6.8408,
      "step": 425
    },
    {
      "epoch": 0.05834018077239113,
      "grad_norm": 2.078125,
      "learning_rate": 0.00019125,
      "loss": 6.8772,
      "step": 426
    },
    {
      "epoch": 0.058477129553546975,
      "grad_norm": 2.625,
      "learning_rate": 0.0001917,
      "loss": 6.8563,
      "step": 427
    },
    {
      "epoch": 0.05861407833470282,
      "grad_norm": 3.53125,
      "learning_rate": 0.00019214999999999998,
      "loss": 6.8973,
      "step": 428
    },
    {
      "epoch": 0.05875102711585867,
      "grad_norm": 2.59375,
      "learning_rate": 0.0001926,
      "loss": 6.852,
      "step": 429
    },
    {
      "epoch": 0.05888797589701451,
      "grad_norm": 2.046875,
      "learning_rate": 0.00019305,
      "loss": 6.8884,
      "step": 430
    },
    {
      "epoch": 0.05902492467817037,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0001935,
      "loss": 6.8163,
      "step": 431
    },
    {
      "epoch": 0.05916187345932621,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019395,
      "loss": 6.843,
      "step": 432
    },
    {
      "epoch": 0.05929882224048206,
      "grad_norm": 2.25,
      "learning_rate": 0.00019439999999999998,
      "loss": 6.859,
      "step": 433
    },
    {
      "epoch": 0.059435771021637905,
      "grad_norm": 2.59375,
      "learning_rate": 0.00019485,
      "loss": 6.8694,
      "step": 434
    },
    {
      "epoch": 0.05957271980279376,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0001953,
      "loss": 6.8549,
      "step": 435
    },
    {
      "epoch": 0.059709668583949604,
      "grad_norm": 1.859375,
      "learning_rate": 0.00019575,
      "loss": 6.8143,
      "step": 436
    },
    {
      "epoch": 0.05984661736510545,
      "grad_norm": 1.875,
      "learning_rate": 0.0001962,
      "loss": 6.8309,
      "step": 437
    },
    {
      "epoch": 0.0599835661462613,
      "grad_norm": 1.9921875,
      "learning_rate": 0.00019664999999999998,
      "loss": 6.8168,
      "step": 438
    },
    {
      "epoch": 0.06012051492741714,
      "grad_norm": 3.0,
      "learning_rate": 0.0001971,
      "loss": 6.8048,
      "step": 439
    },
    {
      "epoch": 0.060257463708572996,
      "grad_norm": 2.734375,
      "learning_rate": 0.00019755,
      "loss": 6.816,
      "step": 440
    },
    {
      "epoch": 0.06039441248972884,
      "grad_norm": 2.421875,
      "learning_rate": 0.000198,
      "loss": 6.8355,
      "step": 441
    },
    {
      "epoch": 0.06053136127088469,
      "grad_norm": 3.171875,
      "learning_rate": 0.00019845,
      "loss": 6.8462,
      "step": 442
    },
    {
      "epoch": 0.060668310052040535,
      "grad_norm": 3.640625,
      "learning_rate": 0.00019889999999999998,
      "loss": 6.8594,
      "step": 443
    },
    {
      "epoch": 0.06080525883319639,
      "grad_norm": 2.875,
      "learning_rate": 0.00019935,
      "loss": 6.8125,
      "step": 444
    },
    {
      "epoch": 0.060942207614352234,
      "grad_norm": 4.6875,
      "learning_rate": 0.0001998,
      "loss": 6.8527,
      "step": 445
    },
    {
      "epoch": 0.06107915639550808,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00020025,
      "loss": 6.8233,
      "step": 446
    },
    {
      "epoch": 0.061216105176663926,
      "grad_norm": 4.15625,
      "learning_rate": 0.0002007,
      "loss": 6.8849,
      "step": 447
    },
    {
      "epoch": 0.06135305395781977,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00020115,
      "loss": 6.8225,
      "step": 448
    },
    {
      "epoch": 0.061490002738975626,
      "grad_norm": 2.90625,
      "learning_rate": 0.0002016,
      "loss": 6.8301,
      "step": 449
    },
    {
      "epoch": 0.06162695152013147,
      "grad_norm": 2.03125,
      "learning_rate": 0.00020205,
      "loss": 6.8767,
      "step": 450
    },
    {
      "epoch": 0.06176390030128732,
      "grad_norm": 2.71875,
      "learning_rate": 0.0002025,
      "loss": 6.8493,
      "step": 451
    },
    {
      "epoch": 0.061900849082443164,
      "grad_norm": 3.0,
      "learning_rate": 0.00020295,
      "loss": 6.8475,
      "step": 452
    },
    {
      "epoch": 0.06203779786359902,
      "grad_norm": 1.828125,
      "learning_rate": 0.0002034,
      "loss": 6.7851,
      "step": 453
    },
    {
      "epoch": 0.062174746644754864,
      "grad_norm": 1.953125,
      "learning_rate": 0.00020385,
      "loss": 6.7846,
      "step": 454
    },
    {
      "epoch": 0.06231169542591071,
      "grad_norm": 2.328125,
      "learning_rate": 0.0002043,
      "loss": 6.801,
      "step": 455
    },
    {
      "epoch": 0.062448644207066556,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00020475,
      "loss": 6.7718,
      "step": 456
    },
    {
      "epoch": 0.0625855929882224,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002052,
      "loss": 6.7815,
      "step": 457
    },
    {
      "epoch": 0.06272254176937825,
      "grad_norm": 1.75,
      "learning_rate": 0.00020565,
      "loss": 6.7649,
      "step": 458
    },
    {
      "epoch": 0.0628594905505341,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002061,
      "loss": 6.8004,
      "step": 459
    },
    {
      "epoch": 0.06299643933168995,
      "grad_norm": 2.0,
      "learning_rate": 0.00020655,
      "loss": 6.7749,
      "step": 460
    },
    {
      "epoch": 0.0631333881128458,
      "grad_norm": 1.25,
      "learning_rate": 0.000207,
      "loss": 6.7651,
      "step": 461
    },
    {
      "epoch": 0.06327033689400165,
      "grad_norm": 1.015625,
      "learning_rate": 0.00020745,
      "loss": 6.7844,
      "step": 462
    },
    {
      "epoch": 0.0634072856751575,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002079,
      "loss": 6.7551,
      "step": 463
    },
    {
      "epoch": 0.06354423445631334,
      "grad_norm": 1.59375,
      "learning_rate": 0.00020835,
      "loss": 6.6931,
      "step": 464
    },
    {
      "epoch": 0.06368118323746919,
      "grad_norm": 1.375,
      "learning_rate": 0.0002088,
      "loss": 6.7392,
      "step": 465
    },
    {
      "epoch": 0.06381813201862503,
      "grad_norm": 1.34375,
      "learning_rate": 0.00020925000000000002,
      "loss": 6.7079,
      "step": 466
    },
    {
      "epoch": 0.06395508079978088,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002097,
      "loss": 6.7312,
      "step": 467
    },
    {
      "epoch": 0.06409202958093672,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00021015,
      "loss": 6.7197,
      "step": 468
    },
    {
      "epoch": 0.06422897836209258,
      "grad_norm": 2.09375,
      "learning_rate": 0.0002106,
      "loss": 6.7653,
      "step": 469
    },
    {
      "epoch": 0.06436592714324843,
      "grad_norm": 2.671875,
      "learning_rate": 0.00021104999999999998,
      "loss": 6.7769,
      "step": 470
    },
    {
      "epoch": 0.06450287592440428,
      "grad_norm": 4.46875,
      "learning_rate": 0.0002115,
      "loss": 6.7981,
      "step": 471
    },
    {
      "epoch": 0.06463982470556012,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00021194999999999997,
      "loss": 6.6892,
      "step": 472
    },
    {
      "epoch": 0.06477677348671597,
      "grad_norm": 2.109375,
      "learning_rate": 0.00021239999999999999,
      "loss": 6.7418,
      "step": 473
    },
    {
      "epoch": 0.06491372226787182,
      "grad_norm": 2.109375,
      "learning_rate": 0.00021285,
      "loss": 6.7572,
      "step": 474
    },
    {
      "epoch": 0.06505067104902766,
      "grad_norm": 1.953125,
      "learning_rate": 0.00021329999999999998,
      "loss": 6.6923,
      "step": 475
    },
    {
      "epoch": 0.06518761983018351,
      "grad_norm": 1.46875,
      "learning_rate": 0.00021375,
      "loss": 6.7207,
      "step": 476
    },
    {
      "epoch": 0.06532456861133935,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00021419999999999998,
      "loss": 6.6732,
      "step": 477
    },
    {
      "epoch": 0.06546151739249521,
      "grad_norm": 1.453125,
      "learning_rate": 0.00021464999999999999,
      "loss": 6.7241,
      "step": 478
    },
    {
      "epoch": 0.06559846617365106,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002151,
      "loss": 6.6935,
      "step": 479
    },
    {
      "epoch": 0.0657354149548069,
      "grad_norm": 1.0625,
      "learning_rate": 0.00021554999999999998,
      "loss": 6.7233,
      "step": 480
    },
    {
      "epoch": 0.06587236373596275,
      "grad_norm": 1.578125,
      "learning_rate": 0.000216,
      "loss": 6.7025,
      "step": 481
    },
    {
      "epoch": 0.0660093125171186,
      "grad_norm": 1.625,
      "learning_rate": 0.00021644999999999998,
      "loss": 6.6841,
      "step": 482
    },
    {
      "epoch": 0.06614626129827444,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0002169,
      "loss": 6.653,
      "step": 483
    },
    {
      "epoch": 0.06628321007943029,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00021735,
      "loss": 6.6705,
      "step": 484
    },
    {
      "epoch": 0.06642015886058614,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00021779999999999998,
      "loss": 6.6958,
      "step": 485
    },
    {
      "epoch": 0.06655710764174198,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00021825,
      "loss": 6.709,
      "step": 486
    },
    {
      "epoch": 0.06669405642289783,
      "grad_norm": 1.265625,
      "learning_rate": 0.00021869999999999998,
      "loss": 6.6584,
      "step": 487
    },
    {
      "epoch": 0.06683100520405369,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00021915,
      "loss": 6.6605,
      "step": 488
    },
    {
      "epoch": 0.06696795398520954,
      "grad_norm": 1.53125,
      "learning_rate": 0.0002196,
      "loss": 6.6695,
      "step": 489
    },
    {
      "epoch": 0.06710490276636538,
      "grad_norm": 1.296875,
      "learning_rate": 0.00022004999999999998,
      "loss": 6.6881,
      "step": 490
    },
    {
      "epoch": 0.06724185154752123,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002205,
      "loss": 6.6444,
      "step": 491
    },
    {
      "epoch": 0.06737880032867707,
      "grad_norm": 1.7890625,
      "learning_rate": 0.00022095,
      "loss": 6.7068,
      "step": 492
    },
    {
      "epoch": 0.06751574910983292,
      "grad_norm": 2.171875,
      "learning_rate": 0.0002214,
      "loss": 6.6313,
      "step": 493
    },
    {
      "epoch": 0.06765269789098877,
      "grad_norm": 2.96875,
      "learning_rate": 0.00022185,
      "loss": 6.6916,
      "step": 494
    },
    {
      "epoch": 0.06778964667214461,
      "grad_norm": 4.59375,
      "learning_rate": 0.00022229999999999998,
      "loss": 6.6919,
      "step": 495
    },
    {
      "epoch": 0.06792659545330046,
      "grad_norm": 2.703125,
      "learning_rate": 0.00022275,
      "loss": 6.685,
      "step": 496
    },
    {
      "epoch": 0.06806354423445632,
      "grad_norm": 4.0625,
      "learning_rate": 0.0002232,
      "loss": 6.7247,
      "step": 497
    },
    {
      "epoch": 0.06820049301561217,
      "grad_norm": 2.859375,
      "learning_rate": 0.00022365,
      "loss": 6.6579,
      "step": 498
    },
    {
      "epoch": 0.06833744179676801,
      "grad_norm": 3.28125,
      "learning_rate": 0.0002241,
      "loss": 6.7159,
      "step": 499
    },
    {
      "epoch": 0.06847439057792386,
      "grad_norm": 2.40625,
      "learning_rate": 0.00022454999999999998,
      "loss": 6.6821,
      "step": 500
    },
    {
      "epoch": 0.0686113393590797,
      "grad_norm": 3.421875,
      "learning_rate": 0.000225,
      "loss": 6.7299,
      "step": 501
    },
    {
      "epoch": 0.06874828814023555,
      "grad_norm": 2.875,
      "learning_rate": 0.00022545,
      "loss": 6.6434,
      "step": 502
    },
    {
      "epoch": 0.0688852369213914,
      "grad_norm": 2.46875,
      "learning_rate": 0.0002259,
      "loss": 6.6986,
      "step": 503
    },
    {
      "epoch": 0.06902218570254724,
      "grad_norm": 1.671875,
      "learning_rate": 0.00022635,
      "loss": 6.6372,
      "step": 504
    },
    {
      "epoch": 0.06915913448370309,
      "grad_norm": 2.03125,
      "learning_rate": 0.00022679999999999998,
      "loss": 6.6797,
      "step": 505
    },
    {
      "epoch": 0.06929608326485895,
      "grad_norm": 2.03125,
      "learning_rate": 0.00022725,
      "loss": 6.6867,
      "step": 506
    },
    {
      "epoch": 0.0694330320460148,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002277,
      "loss": 6.6504,
      "step": 507
    },
    {
      "epoch": 0.06956998082717064,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00022815,
      "loss": 6.6616,
      "step": 508
    },
    {
      "epoch": 0.06970692960832649,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002286,
      "loss": 6.6579,
      "step": 509
    },
    {
      "epoch": 0.06984387838948233,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00022904999999999998,
      "loss": 6.6073,
      "step": 510
    },
    {
      "epoch": 0.06998082717063818,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002295,
      "loss": 6.627,
      "step": 511
    },
    {
      "epoch": 0.07011777595179403,
      "grad_norm": 1.09375,
      "learning_rate": 0.00022995,
      "loss": 6.6356,
      "step": 512
    },
    {
      "epoch": 0.07025472473294987,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002304,
      "loss": 6.6178,
      "step": 513
    },
    {
      "epoch": 0.07039167351410572,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00023085,
      "loss": 6.5817,
      "step": 514
    },
    {
      "epoch": 0.07052862229526158,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002313,
      "loss": 6.6116,
      "step": 515
    },
    {
      "epoch": 0.07066557107641742,
      "grad_norm": 1.40625,
      "learning_rate": 0.00023175,
      "loss": 6.6419,
      "step": 516
    },
    {
      "epoch": 0.07080251985757327,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002322,
      "loss": 6.6216,
      "step": 517
    },
    {
      "epoch": 0.07093946863872912,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00023265,
      "loss": 6.6035,
      "step": 518
    },
    {
      "epoch": 0.07107641741988496,
      "grad_norm": 1.453125,
      "learning_rate": 0.0002331,
      "loss": 6.5945,
      "step": 519
    },
    {
      "epoch": 0.07121336620104081,
      "grad_norm": 1.203125,
      "learning_rate": 0.00023355,
      "loss": 6.5565,
      "step": 520
    },
    {
      "epoch": 0.07135031498219666,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000234,
      "loss": 6.595,
      "step": 521
    },
    {
      "epoch": 0.0714872637633525,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00023445,
      "loss": 6.5658,
      "step": 522
    },
    {
      "epoch": 0.07162421254450835,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002349,
      "loss": 6.553,
      "step": 523
    },
    {
      "epoch": 0.07176116132566421,
      "grad_norm": 1.71875,
      "learning_rate": 0.00023535,
      "loss": 6.6382,
      "step": 524
    },
    {
      "epoch": 0.07189811010682005,
      "grad_norm": 2.34375,
      "learning_rate": 0.0002358,
      "loss": 6.5991,
      "step": 525
    },
    {
      "epoch": 0.0720350588879759,
      "grad_norm": 3.234375,
      "learning_rate": 0.00023625,
      "loss": 6.641,
      "step": 526
    },
    {
      "epoch": 0.07217200766913175,
      "grad_norm": 2.546875,
      "learning_rate": 0.0002367,
      "loss": 6.6043,
      "step": 527
    },
    {
      "epoch": 0.07230895645028759,
      "grad_norm": 1.7421875,
      "learning_rate": 0.00023715,
      "loss": 6.6034,
      "step": 528
    },
    {
      "epoch": 0.07244590523144344,
      "grad_norm": 2.640625,
      "learning_rate": 0.0002376,
      "loss": 6.5936,
      "step": 529
    },
    {
      "epoch": 0.07258285401259928,
      "grad_norm": 2.265625,
      "learning_rate": 0.00023805,
      "loss": 6.6054,
      "step": 530
    },
    {
      "epoch": 0.07271980279375513,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002385,
      "loss": 6.5595,
      "step": 531
    },
    {
      "epoch": 0.07285675157491098,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00023895,
      "loss": 6.5808,
      "step": 532
    },
    {
      "epoch": 0.07299370035606684,
      "grad_norm": 1.8125,
      "learning_rate": 0.00023940000000000002,
      "loss": 6.5629,
      "step": 533
    },
    {
      "epoch": 0.07313064913722268,
      "grad_norm": 1.9609375,
      "learning_rate": 0.00023985,
      "loss": 6.5787,
      "step": 534
    },
    {
      "epoch": 0.07326759791837853,
      "grad_norm": 1.453125,
      "learning_rate": 0.0002403,
      "loss": 6.5556,
      "step": 535
    },
    {
      "epoch": 0.07340454669953438,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00024075,
      "loss": 6.5743,
      "step": 536
    },
    {
      "epoch": 0.07354149548069022,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002412,
      "loss": 6.5802,
      "step": 537
    },
    {
      "epoch": 0.07367844426184607,
      "grad_norm": 1.328125,
      "learning_rate": 0.00024165000000000002,
      "loss": 6.5701,
      "step": 538
    },
    {
      "epoch": 0.07381539304300191,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002421,
      "loss": 6.5522,
      "step": 539
    },
    {
      "epoch": 0.07395234182415776,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00024255000000000001,
      "loss": 6.542,
      "step": 540
    },
    {
      "epoch": 0.0740892906053136,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000243,
      "loss": 6.564,
      "step": 541
    },
    {
      "epoch": 0.07422623938646947,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00024345,
      "loss": 6.5789,
      "step": 542
    },
    {
      "epoch": 0.07436318816762531,
      "grad_norm": 1.96875,
      "learning_rate": 0.00024390000000000002,
      "loss": 6.5397,
      "step": 543
    },
    {
      "epoch": 0.07450013694878116,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00024435,
      "loss": 6.5469,
      "step": 544
    },
    {
      "epoch": 0.074637085729937,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0002448,
      "loss": 6.5513,
      "step": 545
    },
    {
      "epoch": 0.07477403451109285,
      "grad_norm": 1.875,
      "learning_rate": 0.00024525,
      "loss": 6.531,
      "step": 546
    },
    {
      "epoch": 0.0749109832922487,
      "grad_norm": 1.796875,
      "learning_rate": 0.0002457,
      "loss": 6.549,
      "step": 547
    },
    {
      "epoch": 0.07504793207340454,
      "grad_norm": 1.71875,
      "learning_rate": 0.00024615,
      "loss": 6.5166,
      "step": 548
    },
    {
      "epoch": 0.07518488085456039,
      "grad_norm": 1.65625,
      "learning_rate": 0.00024660000000000003,
      "loss": 6.5732,
      "step": 549
    },
    {
      "epoch": 0.07532182963571624,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00024705,
      "loss": 6.5131,
      "step": 550
    },
    {
      "epoch": 0.0754587784168721,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0002475,
      "loss": 6.5579,
      "step": 551
    },
    {
      "epoch": 0.07559572719802794,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00024795000000000004,
      "loss": 6.599,
      "step": 552
    },
    {
      "epoch": 0.07573267597918379,
      "grad_norm": 2.421875,
      "learning_rate": 0.0002484,
      "loss": 6.5565,
      "step": 553
    },
    {
      "epoch": 0.07586962476033964,
      "grad_norm": 2.4375,
      "learning_rate": 0.00024885,
      "loss": 6.5131,
      "step": 554
    },
    {
      "epoch": 0.07600657354149548,
      "grad_norm": 2.359375,
      "learning_rate": 0.0002493,
      "loss": 6.5648,
      "step": 555
    },
    {
      "epoch": 0.07614352232265133,
      "grad_norm": 3.265625,
      "learning_rate": 0.00024975,
      "loss": 6.541,
      "step": 556
    },
    {
      "epoch": 0.07628047110380717,
      "grad_norm": 2.625,
      "learning_rate": 0.0002502,
      "loss": 6.5476,
      "step": 557
    },
    {
      "epoch": 0.07641741988496302,
      "grad_norm": 2.484375,
      "learning_rate": 0.00025065,
      "loss": 6.5467,
      "step": 558
    },
    {
      "epoch": 0.07655436866611887,
      "grad_norm": 2.453125,
      "learning_rate": 0.00025110000000000003,
      "loss": 6.5811,
      "step": 559
    },
    {
      "epoch": 0.07669131744727473,
      "grad_norm": 2.75,
      "learning_rate": 0.00025155,
      "loss": 6.5334,
      "step": 560
    },
    {
      "epoch": 0.07682826622843057,
      "grad_norm": 2.75,
      "learning_rate": 0.000252,
      "loss": 6.5257,
      "step": 561
    },
    {
      "epoch": 0.07696521500958642,
      "grad_norm": 2.65625,
      "learning_rate": 0.00025245000000000004,
      "loss": 6.5469,
      "step": 562
    },
    {
      "epoch": 0.07710216379074226,
      "grad_norm": 2.875,
      "learning_rate": 0.0002529,
      "loss": 6.5385,
      "step": 563
    },
    {
      "epoch": 0.07723911257189811,
      "grad_norm": 3.0625,
      "learning_rate": 0.00025334999999999995,
      "loss": 6.5733,
      "step": 564
    },
    {
      "epoch": 0.07737606135305396,
      "grad_norm": 2.515625,
      "learning_rate": 0.0002538,
      "loss": 6.5561,
      "step": 565
    },
    {
      "epoch": 0.0775130101342098,
      "grad_norm": 2.265625,
      "learning_rate": 0.00025425,
      "loss": 6.5551,
      "step": 566
    },
    {
      "epoch": 0.07764995891536565,
      "grad_norm": 2.515625,
      "learning_rate": 0.00025469999999999996,
      "loss": 6.5644,
      "step": 567
    },
    {
      "epoch": 0.0777869076965215,
      "grad_norm": 2.234375,
      "learning_rate": 0.00025515,
      "loss": 6.5283,
      "step": 568
    },
    {
      "epoch": 0.07792385647767734,
      "grad_norm": 1.921875,
      "learning_rate": 0.0002556,
      "loss": 6.4915,
      "step": 569
    },
    {
      "epoch": 0.0780608052588332,
      "grad_norm": 1.609375,
      "learning_rate": 0.00025604999999999996,
      "loss": 6.5079,
      "step": 570
    },
    {
      "epoch": 0.07819775403998905,
      "grad_norm": 1.328125,
      "learning_rate": 0.00025649999999999995,
      "loss": 6.5299,
      "step": 571
    },
    {
      "epoch": 0.0783347028211449,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00025695,
      "loss": 6.5197,
      "step": 572
    },
    {
      "epoch": 0.07847165160230074,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00025739999999999997,
      "loss": 6.492,
      "step": 573
    },
    {
      "epoch": 0.07860860038345659,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00025784999999999995,
      "loss": 6.5262,
      "step": 574
    },
    {
      "epoch": 0.07874554916461243,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002583,
      "loss": 6.513,
      "step": 575
    },
    {
      "epoch": 0.07888249794576828,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00025875,
      "loss": 6.5422,
      "step": 576
    },
    {
      "epoch": 0.07901944672692413,
      "grad_norm": 0.9375,
      "learning_rate": 0.00025919999999999996,
      "loss": 6.5102,
      "step": 577
    },
    {
      "epoch": 0.07915639550807997,
      "grad_norm": 1.0,
      "learning_rate": 0.00025965,
      "loss": 6.4726,
      "step": 578
    },
    {
      "epoch": 0.07929334428923583,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002601,
      "loss": 6.5342,
      "step": 579
    },
    {
      "epoch": 0.07943029307039168,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00026054999999999996,
      "loss": 6.5027,
      "step": 580
    },
    {
      "epoch": 0.07956724185154752,
      "grad_norm": 1.046875,
      "learning_rate": 0.000261,
      "loss": 6.4839,
      "step": 581
    },
    {
      "epoch": 0.07970419063270337,
      "grad_norm": 0.9375,
      "learning_rate": 0.00026145,
      "loss": 6.4804,
      "step": 582
    },
    {
      "epoch": 0.07984113941385922,
      "grad_norm": 1.28125,
      "learning_rate": 0.00026189999999999997,
      "loss": 6.4554,
      "step": 583
    },
    {
      "epoch": 0.07997808819501506,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00026234999999999995,
      "loss": 6.4675,
      "step": 584
    },
    {
      "epoch": 0.08011503697617091,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002628,
      "loss": 6.501,
      "step": 585
    },
    {
      "epoch": 0.08025198575732675,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00026325,
      "loss": 6.4513,
      "step": 586
    },
    {
      "epoch": 0.0803889345384826,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00026369999999999996,
      "loss": 6.4377,
      "step": 587
    },
    {
      "epoch": 0.08052588331963846,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00026415,
      "loss": 6.4819,
      "step": 588
    },
    {
      "epoch": 0.08066283210079431,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002646,
      "loss": 6.4782,
      "step": 589
    },
    {
      "epoch": 0.08079978088195015,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00026504999999999996,
      "loss": 6.477,
      "step": 590
    },
    {
      "epoch": 0.080936729663106,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002655,
      "loss": 6.4317,
      "step": 591
    },
    {
      "epoch": 0.08107367844426185,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00026595,
      "loss": 6.4745,
      "step": 592
    },
    {
      "epoch": 0.08121062722541769,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00026639999999999997,
      "loss": 6.4817,
      "step": 593
    },
    {
      "epoch": 0.08134757600657354,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00026684999999999995,
      "loss": 6.455,
      "step": 594
    },
    {
      "epoch": 0.08148452478772938,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002673,
      "loss": 6.4891,
      "step": 595
    },
    {
      "epoch": 0.08162147356888523,
      "grad_norm": 1.359375,
      "learning_rate": 0.00026775,
      "loss": 6.4493,
      "step": 596
    },
    {
      "epoch": 0.08175842235004109,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026819999999999996,
      "loss": 6.4452,
      "step": 597
    },
    {
      "epoch": 0.08189537113119694,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00026865,
      "loss": 6.4425,
      "step": 598
    },
    {
      "epoch": 0.08203231991235278,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002691,
      "loss": 6.4555,
      "step": 599
    },
    {
      "epoch": 0.08216926869350863,
      "grad_norm": 1.078125,
      "learning_rate": 0.00026954999999999997,
      "loss": 6.4244,
      "step": 600
    },
    {
      "epoch": 0.08230621747466448,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00027,
      "loss": 6.4531,
      "step": 601
    },
    {
      "epoch": 0.08244316625582032,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00027045,
      "loss": 6.4648,
      "step": 602
    },
    {
      "epoch": 0.08258011503697617,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00027089999999999997,
      "loss": 6.474,
      "step": 603
    },
    {
      "epoch": 0.08271706381813201,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00027135,
      "loss": 6.432,
      "step": 604
    },
    {
      "epoch": 0.08285401259928786,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0002718,
      "loss": 6.4232,
      "step": 605
    },
    {
      "epoch": 0.08299096138044372,
      "grad_norm": 2.46875,
      "learning_rate": 0.00027225,
      "loss": 6.4429,
      "step": 606
    },
    {
      "epoch": 0.08312791016159957,
      "grad_norm": 4.21875,
      "learning_rate": 0.00027269999999999996,
      "loss": 6.4966,
      "step": 607
    },
    {
      "epoch": 0.08326485894275541,
      "grad_norm": 2.0,
      "learning_rate": 0.00027315,
      "loss": 6.4513,
      "step": 608
    },
    {
      "epoch": 0.08340180772391126,
      "grad_norm": 1.953125,
      "learning_rate": 0.0002736,
      "loss": 6.447,
      "step": 609
    },
    {
      "epoch": 0.0835387565050671,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00027404999999999997,
      "loss": 6.4488,
      "step": 610
    },
    {
      "epoch": 0.08367570528622295,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0002745,
      "loss": 6.4878,
      "step": 611
    },
    {
      "epoch": 0.0838126540673788,
      "grad_norm": 1.671875,
      "learning_rate": 0.00027495,
      "loss": 6.4373,
      "step": 612
    },
    {
      "epoch": 0.08394960284853464,
      "grad_norm": 1.734375,
      "learning_rate": 0.00027539999999999997,
      "loss": 6.4298,
      "step": 613
    },
    {
      "epoch": 0.08408655162969049,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00027585,
      "loss": 6.481,
      "step": 614
    },
    {
      "epoch": 0.08422350041084635,
      "grad_norm": 1.375,
      "learning_rate": 0.0002763,
      "loss": 6.4455,
      "step": 615
    },
    {
      "epoch": 0.0843604491920022,
      "grad_norm": 1.84375,
      "learning_rate": 0.00027675,
      "loss": 6.4987,
      "step": 616
    },
    {
      "epoch": 0.08449739797315804,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002772,
      "loss": 6.44,
      "step": 617
    },
    {
      "epoch": 0.08463434675431389,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00027765,
      "loss": 6.4532,
      "step": 618
    },
    {
      "epoch": 0.08477129553546973,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002781,
      "loss": 6.4619,
      "step": 619
    },
    {
      "epoch": 0.08490824431662558,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00027854999999999997,
      "loss": 6.3982,
      "step": 620
    },
    {
      "epoch": 0.08504519309778143,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000279,
      "loss": 6.3724,
      "step": 621
    },
    {
      "epoch": 0.08518214187893727,
      "grad_norm": 1.34375,
      "learning_rate": 0.00027945,
      "loss": 6.4173,
      "step": 622
    },
    {
      "epoch": 0.08531909066009312,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0002799,
      "loss": 6.4149,
      "step": 623
    },
    {
      "epoch": 0.08545603944124898,
      "grad_norm": 2.5625,
      "learning_rate": 0.00028035,
      "loss": 6.4254,
      "step": 624
    },
    {
      "epoch": 0.08559298822240483,
      "grad_norm": 2.46875,
      "learning_rate": 0.0002808,
      "loss": 6.4614,
      "step": 625
    },
    {
      "epoch": 0.08572993700356067,
      "grad_norm": 1.734375,
      "learning_rate": 0.00028125,
      "loss": 6.4338,
      "step": 626
    },
    {
      "epoch": 0.08586688578471652,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002817,
      "loss": 6.383,
      "step": 627
    },
    {
      "epoch": 0.08600383456587236,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00028215,
      "loss": 6.4008,
      "step": 628
    },
    {
      "epoch": 0.08614078334702821,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002826,
      "loss": 6.4237,
      "step": 629
    },
    {
      "epoch": 0.08627773212818406,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00028304999999999997,
      "loss": 6.4413,
      "step": 630
    },
    {
      "epoch": 0.0864146809093399,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0002835,
      "loss": 6.4112,
      "step": 631
    },
    {
      "epoch": 0.08655162969049575,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00028395,
      "loss": 6.4089,
      "step": 632
    },
    {
      "epoch": 0.08668857847165161,
      "grad_norm": 2.28125,
      "learning_rate": 0.0002844,
      "loss": 6.4417,
      "step": 633
    },
    {
      "epoch": 0.08682552725280746,
      "grad_norm": 1.9921875,
      "learning_rate": 0.00028485,
      "loss": 6.417,
      "step": 634
    },
    {
      "epoch": 0.0869624760339633,
      "grad_norm": 1.84375,
      "learning_rate": 0.0002853,
      "loss": 6.4578,
      "step": 635
    },
    {
      "epoch": 0.08709942481511915,
      "grad_norm": 2.4375,
      "learning_rate": 0.00028575,
      "loss": 6.4641,
      "step": 636
    },
    {
      "epoch": 0.087236373596275,
      "grad_norm": 1.671875,
      "learning_rate": 0.0002862,
      "loss": 6.4155,
      "step": 637
    },
    {
      "epoch": 0.08737332237743084,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00028665,
      "loss": 6.4047,
      "step": 638
    },
    {
      "epoch": 0.08751027115858669,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002871,
      "loss": 6.3902,
      "step": 639
    },
    {
      "epoch": 0.08764721993974253,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00028755,
      "loss": 6.3782,
      "step": 640
    },
    {
      "epoch": 0.08778416872089838,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000288,
      "loss": 6.42,
      "step": 641
    },
    {
      "epoch": 0.08792111750205424,
      "grad_norm": 1.3125,
      "learning_rate": 0.00028845,
      "loss": 6.3393,
      "step": 642
    },
    {
      "epoch": 0.08805806628321008,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002889,
      "loss": 6.3705,
      "step": 643
    },
    {
      "epoch": 0.08819501506436593,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00028935,
      "loss": 6.3821,
      "step": 644
    },
    {
      "epoch": 0.08833196384552178,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002898,
      "loss": 6.4257,
      "step": 645
    },
    {
      "epoch": 0.08846891262667762,
      "grad_norm": 1.03125,
      "learning_rate": 0.00029025,
      "loss": 6.3634,
      "step": 646
    },
    {
      "epoch": 0.08860586140783347,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002907,
      "loss": 6.3937,
      "step": 647
    },
    {
      "epoch": 0.08874281018898932,
      "grad_norm": 1.359375,
      "learning_rate": 0.00029115,
      "loss": 6.3849,
      "step": 648
    },
    {
      "epoch": 0.08887975897014516,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002916,
      "loss": 6.3971,
      "step": 649
    },
    {
      "epoch": 0.08901670775130101,
      "grad_norm": 1.203125,
      "learning_rate": 0.00029205,
      "loss": 6.3578,
      "step": 650
    },
    {
      "epoch": 0.08915365653245687,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002925,
      "loss": 6.3987,
      "step": 651
    },
    {
      "epoch": 0.08929060531361271,
      "grad_norm": 1.078125,
      "learning_rate": 0.00029295,
      "loss": 6.3481,
      "step": 652
    },
    {
      "epoch": 0.08942755409476856,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002934,
      "loss": 6.3594,
      "step": 653
    },
    {
      "epoch": 0.0895645028759244,
      "grad_norm": 1.09375,
      "learning_rate": 0.00029385,
      "loss": 6.3709,
      "step": 654
    },
    {
      "epoch": 0.08970145165708025,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002943,
      "loss": 6.4116,
      "step": 655
    },
    {
      "epoch": 0.0898384004382361,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00029475,
      "loss": 6.3548,
      "step": 656
    },
    {
      "epoch": 0.08997534921939195,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002952,
      "loss": 6.3934,
      "step": 657
    },
    {
      "epoch": 0.09011229800054779,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00029565,
      "loss": 6.3522,
      "step": 658
    },
    {
      "epoch": 0.09024924678170364,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002961,
      "loss": 6.3697,
      "step": 659
    },
    {
      "epoch": 0.09038619556285948,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00029655,
      "loss": 6.3554,
      "step": 660
    },
    {
      "epoch": 0.09052314434401534,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000297,
      "loss": 6.4,
      "step": 661
    },
    {
      "epoch": 0.09066009312517119,
      "grad_norm": 1.09375,
      "learning_rate": 0.00029745,
      "loss": 6.3768,
      "step": 662
    },
    {
      "epoch": 0.09079704190632704,
      "grad_norm": 1.125,
      "learning_rate": 0.00029790000000000003,
      "loss": 6.3803,
      "step": 663
    },
    {
      "epoch": 0.09093399068748288,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00029835,
      "loss": 6.352,
      "step": 664
    },
    {
      "epoch": 0.09107093946863873,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002988,
      "loss": 6.3456,
      "step": 665
    },
    {
      "epoch": 0.09120788824979457,
      "grad_norm": 1.796875,
      "learning_rate": 0.00029925,
      "loss": 6.3788,
      "step": 666
    },
    {
      "epoch": 0.09134483703095042,
      "grad_norm": 1.953125,
      "learning_rate": 0.0002997,
      "loss": 6.3638,
      "step": 667
    },
    {
      "epoch": 0.09148178581210627,
      "grad_norm": 2.109375,
      "learning_rate": 0.00030015,
      "loss": 6.3935,
      "step": 668
    },
    {
      "epoch": 0.09161873459326211,
      "grad_norm": 2.46875,
      "learning_rate": 0.0003006,
      "loss": 6.4167,
      "step": 669
    },
    {
      "epoch": 0.09175568337441797,
      "grad_norm": 2.703125,
      "learning_rate": 0.00030105,
      "loss": 6.3946,
      "step": 670
    },
    {
      "epoch": 0.09189263215557382,
      "grad_norm": 2.515625,
      "learning_rate": 0.0003015,
      "loss": 6.4274,
      "step": 671
    },
    {
      "epoch": 0.09202958093672967,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00030195,
      "loss": 6.3891,
      "step": 672
    },
    {
      "epoch": 0.09216652971788551,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00030240000000000003,
      "loss": 6.3785,
      "step": 673
    },
    {
      "epoch": 0.09230347849904136,
      "grad_norm": 1.5,
      "learning_rate": 0.00030285,
      "loss": 6.4153,
      "step": 674
    },
    {
      "epoch": 0.0924404272801972,
      "grad_norm": 1.734375,
      "learning_rate": 0.0003033,
      "loss": 6.4447,
      "step": 675
    },
    {
      "epoch": 0.09257737606135305,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00030375000000000004,
      "loss": 6.3626,
      "step": 676
    },
    {
      "epoch": 0.0927143248425089,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003042,
      "loss": 6.3656,
      "step": 677
    },
    {
      "epoch": 0.09285127362366474,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00030465,
      "loss": 6.3578,
      "step": 678
    },
    {
      "epoch": 0.0929882224048206,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003051,
      "loss": 6.3827,
      "step": 679
    },
    {
      "epoch": 0.09312517118597645,
      "grad_norm": 1.296875,
      "learning_rate": 0.00030555000000000003,
      "loss": 6.3787,
      "step": 680
    },
    {
      "epoch": 0.0932621199671323,
      "grad_norm": 1.359375,
      "learning_rate": 0.000306,
      "loss": 6.3655,
      "step": 681
    },
    {
      "epoch": 0.09339906874828814,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00030645,
      "loss": 6.3346,
      "step": 682
    },
    {
      "epoch": 0.09353601752944399,
      "grad_norm": 1.390625,
      "learning_rate": 0.00030690000000000003,
      "loss": 6.38,
      "step": 683
    },
    {
      "epoch": 0.09367296631059983,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00030735,
      "loss": 6.3816,
      "step": 684
    },
    {
      "epoch": 0.09380991509175568,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003078,
      "loss": 6.404,
      "step": 685
    },
    {
      "epoch": 0.09394686387291153,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00030825000000000004,
      "loss": 6.3652,
      "step": 686
    },
    {
      "epoch": 0.09408381265406737,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003087,
      "loss": 6.4,
      "step": 687
    },
    {
      "epoch": 0.09422076143522323,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00030915,
      "loss": 6.3771,
      "step": 688
    },
    {
      "epoch": 0.09435771021637908,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003096,
      "loss": 6.2957,
      "step": 689
    },
    {
      "epoch": 0.09449465899753492,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00031005,
      "loss": 6.3508,
      "step": 690
    },
    {
      "epoch": 0.09463160777869077,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00031049999999999996,
      "loss": 6.3646,
      "step": 691
    },
    {
      "epoch": 0.09476855655984662,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00031094999999999994,
      "loss": 6.3285,
      "step": 692
    },
    {
      "epoch": 0.09490550534100246,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003114,
      "loss": 6.3631,
      "step": 693
    },
    {
      "epoch": 0.09504245412215831,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00031184999999999996,
      "loss": 6.3509,
      "step": 694
    },
    {
      "epoch": 0.09517940290331416,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00031229999999999995,
      "loss": 6.3333,
      "step": 695
    },
    {
      "epoch": 0.09531635168447,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00031275,
      "loss": 6.3191,
      "step": 696
    },
    {
      "epoch": 0.09545330046562586,
      "grad_norm": 0.78125,
      "learning_rate": 0.00031319999999999997,
      "loss": 6.3442,
      "step": 697
    },
    {
      "epoch": 0.09559024924678171,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00031364999999999995,
      "loss": 6.3896,
      "step": 698
    },
    {
      "epoch": 0.09572719802793755,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003141,
      "loss": 6.3728,
      "step": 699
    },
    {
      "epoch": 0.0958641468090934,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00031455,
      "loss": 6.311,
      "step": 700
    },
    {
      "epoch": 0.09600109559024925,
      "grad_norm": 1.140625,
      "learning_rate": 0.00031499999999999996,
      "loss": 6.3299,
      "step": 701
    },
    {
      "epoch": 0.0961380443714051,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00031545,
      "loss": 6.3362,
      "step": 702
    },
    {
      "epoch": 0.09627499315256094,
      "grad_norm": 1.90625,
      "learning_rate": 0.0003159,
      "loss": 6.3208,
      "step": 703
    },
    {
      "epoch": 0.09641194193371679,
      "grad_norm": 2.921875,
      "learning_rate": 0.00031634999999999996,
      "loss": 6.3505,
      "step": 704
    },
    {
      "epoch": 0.09654889071487263,
      "grad_norm": 3.859375,
      "learning_rate": 0.00031679999999999995,
      "loss": 6.4579,
      "step": 705
    },
    {
      "epoch": 0.09668583949602849,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00031725,
      "loss": 6.3653,
      "step": 706
    },
    {
      "epoch": 0.09682278827718434,
      "grad_norm": 2.359375,
      "learning_rate": 0.00031769999999999997,
      "loss": 6.3928,
      "step": 707
    },
    {
      "epoch": 0.09695973705834018,
      "grad_norm": 1.703125,
      "learning_rate": 0.00031814999999999995,
      "loss": 6.3454,
      "step": 708
    },
    {
      "epoch": 0.09709668583949603,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003186,
      "loss": 6.397,
      "step": 709
    },
    {
      "epoch": 0.09723363462065188,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00031905,
      "loss": 6.4112,
      "step": 710
    },
    {
      "epoch": 0.09737058340180772,
      "grad_norm": 1.34375,
      "learning_rate": 0.00031949999999999996,
      "loss": 6.3562,
      "step": 711
    },
    {
      "epoch": 0.09750753218296357,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00031995,
      "loss": 6.381,
      "step": 712
    },
    {
      "epoch": 0.09764448096411942,
      "grad_norm": 1.125,
      "learning_rate": 0.0003204,
      "loss": 6.305,
      "step": 713
    },
    {
      "epoch": 0.09778142974527526,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00032084999999999997,
      "loss": 6.3247,
      "step": 714
    },
    {
      "epoch": 0.09791837852643112,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003213,
      "loss": 6.3379,
      "step": 715
    },
    {
      "epoch": 0.09805532730758697,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00032175,
      "loss": 6.3478,
      "step": 716
    },
    {
      "epoch": 0.09819227608874281,
      "grad_norm": 0.84375,
      "learning_rate": 0.00032219999999999997,
      "loss": 6.3495,
      "step": 717
    },
    {
      "epoch": 0.09832922486989866,
      "grad_norm": 0.9375,
      "learning_rate": 0.00032264999999999996,
      "loss": 6.3266,
      "step": 718
    },
    {
      "epoch": 0.0984661736510545,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003231,
      "loss": 6.3417,
      "step": 719
    },
    {
      "epoch": 0.09860312243221035,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00032355,
      "loss": 6.3453,
      "step": 720
    },
    {
      "epoch": 0.0987400712133662,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00032399999999999996,
      "loss": 6.3144,
      "step": 721
    },
    {
      "epoch": 0.09887701999452204,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00032445,
      "loss": 6.3252,
      "step": 722
    },
    {
      "epoch": 0.09901396877567789,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003249,
      "loss": 6.281,
      "step": 723
    },
    {
      "epoch": 0.09915091755683375,
      "grad_norm": 0.953125,
      "learning_rate": 0.00032534999999999997,
      "loss": 6.3241,
      "step": 724
    },
    {
      "epoch": 0.0992878663379896,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003258,
      "loss": 6.3203,
      "step": 725
    },
    {
      "epoch": 0.09942481511914544,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00032625,
      "loss": 6.3087,
      "step": 726
    },
    {
      "epoch": 0.09956176390030129,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00032669999999999997,
      "loss": 6.3527,
      "step": 727
    },
    {
      "epoch": 0.09969871268145714,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00032714999999999996,
      "loss": 6.2924,
      "step": 728
    },
    {
      "epoch": 0.09983566146261298,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003276,
      "loss": 6.3053,
      "step": 729
    },
    {
      "epoch": 0.09997261024376883,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00032805,
      "loss": 6.3125,
      "step": 730
    },
    {
      "epoch": 0.10010955902492467,
      "grad_norm": 0.921875,
      "learning_rate": 0.00032849999999999996,
      "loss": 6.2783,
      "step": 731
    },
    {
      "epoch": 0.10024650780608052,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00032895,
      "loss": 6.2976,
      "step": 732
    },
    {
      "epoch": 0.10038345658723638,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003294,
      "loss": 6.2545,
      "step": 733
    },
    {
      "epoch": 0.10052040536839223,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00032984999999999997,
      "loss": 6.2913,
      "step": 734
    },
    {
      "epoch": 0.10065735414954807,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003303,
      "loss": 6.2552,
      "step": 735
    },
    {
      "epoch": 0.10079430293070392,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00033075,
      "loss": 6.2811,
      "step": 736
    },
    {
      "epoch": 0.10093125171185977,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003312,
      "loss": 6.2917,
      "step": 737
    },
    {
      "epoch": 0.10106820049301561,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00033165,
      "loss": 6.3325,
      "step": 738
    },
    {
      "epoch": 0.10120514927417146,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003321,
      "loss": 6.3078,
      "step": 739
    },
    {
      "epoch": 0.1013420980553273,
      "grad_norm": 1.046875,
      "learning_rate": 0.00033255,
      "loss": 6.2369,
      "step": 740
    },
    {
      "epoch": 0.10147904683648315,
      "grad_norm": 1.03125,
      "learning_rate": 0.00033299999999999996,
      "loss": 6.2556,
      "step": 741
    },
    {
      "epoch": 0.101615995617639,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00033345,
      "loss": 6.342,
      "step": 742
    },
    {
      "epoch": 0.10175294439879486,
      "grad_norm": 1.625,
      "learning_rate": 0.0003339,
      "loss": 6.327,
      "step": 743
    },
    {
      "epoch": 0.1018898931799507,
      "grad_norm": 1.921875,
      "learning_rate": 0.00033434999999999997,
      "loss": 6.2833,
      "step": 744
    },
    {
      "epoch": 0.10202684196110655,
      "grad_norm": 2.6875,
      "learning_rate": 0.0003348,
      "loss": 6.3358,
      "step": 745
    },
    {
      "epoch": 0.1021637907422624,
      "grad_norm": 3.015625,
      "learning_rate": 0.00033525,
      "loss": 6.3329,
      "step": 746
    },
    {
      "epoch": 0.10230073952341824,
      "grad_norm": 2.171875,
      "learning_rate": 0.0003357,
      "loss": 6.3772,
      "step": 747
    },
    {
      "epoch": 0.10243768830457409,
      "grad_norm": 2.390625,
      "learning_rate": 0.00033615,
      "loss": 6.3207,
      "step": 748
    },
    {
      "epoch": 0.10257463708572993,
      "grad_norm": 2.8125,
      "learning_rate": 0.0003366,
      "loss": 6.3786,
      "step": 749
    },
    {
      "epoch": 0.10271158586688578,
      "grad_norm": 3.21875,
      "learning_rate": 0.00033705,
      "loss": 6.3561,
      "step": 750
    },
    {
      "epoch": 0.10284853464804163,
      "grad_norm": 2.0625,
      "learning_rate": 0.0003375,
      "loss": 6.4183,
      "step": 751
    },
    {
      "epoch": 0.10298548342919749,
      "grad_norm": 1.8984375,
      "learning_rate": 0.00033795,
      "loss": 6.3813,
      "step": 752
    },
    {
      "epoch": 0.10312243221035333,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0003384,
      "loss": 6.3432,
      "step": 753
    },
    {
      "epoch": 0.10325938099150918,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00033884999999999997,
      "loss": 6.338,
      "step": 754
    },
    {
      "epoch": 0.10339632977266502,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003393,
      "loss": 6.3581,
      "step": 755
    },
    {
      "epoch": 0.10353327855382087,
      "grad_norm": 0.9375,
      "learning_rate": 0.00033975,
      "loss": 6.3402,
      "step": 756
    },
    {
      "epoch": 0.10367022733497672,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003402,
      "loss": 6.3378,
      "step": 757
    },
    {
      "epoch": 0.10380717611613256,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00034065,
      "loss": 6.3121,
      "step": 758
    },
    {
      "epoch": 0.10394412489728841,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003411,
      "loss": 6.3012,
      "step": 759
    },
    {
      "epoch": 0.10408107367844426,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00034155,
      "loss": 6.3391,
      "step": 760
    },
    {
      "epoch": 0.10421802245960012,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000342,
      "loss": 6.2968,
      "step": 761
    },
    {
      "epoch": 0.10435497124075596,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00034245,
      "loss": 6.3074,
      "step": 762
    },
    {
      "epoch": 0.10449192002191181,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003429,
      "loss": 6.2987,
      "step": 763
    },
    {
      "epoch": 0.10462886880306765,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00034334999999999997,
      "loss": 6.3422,
      "step": 764
    },
    {
      "epoch": 0.1047658175842235,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0003438,
      "loss": 6.3001,
      "step": 765
    },
    {
      "epoch": 0.10490276636537935,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00034425,
      "loss": 6.3322,
      "step": 766
    },
    {
      "epoch": 0.10503971514653519,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0003447,
      "loss": 6.3198,
      "step": 767
    },
    {
      "epoch": 0.10517666392769104,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00034515,
      "loss": 6.3017,
      "step": 768
    },
    {
      "epoch": 0.10531361270884688,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003456,
      "loss": 6.2685,
      "step": 769
    },
    {
      "epoch": 0.10545056149000274,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00034605,
      "loss": 6.2564,
      "step": 770
    },
    {
      "epoch": 0.10558751027115859,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0003465,
      "loss": 6.2756,
      "step": 771
    },
    {
      "epoch": 0.10572445905231444,
      "grad_norm": 0.734375,
      "learning_rate": 0.00034695,
      "loss": 6.3303,
      "step": 772
    },
    {
      "epoch": 0.10586140783347028,
      "grad_norm": 0.796875,
      "learning_rate": 0.0003474,
      "loss": 6.222,
      "step": 773
    },
    {
      "epoch": 0.10599835661462613,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00034785,
      "loss": 6.2284,
      "step": 774
    },
    {
      "epoch": 0.10613530539578198,
      "grad_norm": 0.75,
      "learning_rate": 0.0003483,
      "loss": 6.2839,
      "step": 775
    },
    {
      "epoch": 0.10627225417693782,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00034875,
      "loss": 6.2565,
      "step": 776
    },
    {
      "epoch": 0.10640920295809367,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0003492,
      "loss": 6.2665,
      "step": 777
    },
    {
      "epoch": 0.10654615173924951,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00034965,
      "loss": 6.2929,
      "step": 778
    },
    {
      "epoch": 0.10668310052040537,
      "grad_norm": 0.703125,
      "learning_rate": 0.0003501,
      "loss": 6.269,
      "step": 779
    },
    {
      "epoch": 0.10682004930156122,
      "grad_norm": 0.75,
      "learning_rate": 0.00035055,
      "loss": 6.3121,
      "step": 780
    },
    {
      "epoch": 0.10695699808271707,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000351,
      "loss": 6.3228,
      "step": 781
    },
    {
      "epoch": 0.10709394686387291,
      "grad_norm": 0.8125,
      "learning_rate": 0.00035145,
      "loss": 6.3291,
      "step": 782
    },
    {
      "epoch": 0.10723089564502876,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003519,
      "loss": 6.2414,
      "step": 783
    },
    {
      "epoch": 0.1073678444261846,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00035235,
      "loss": 6.3373,
      "step": 784
    },
    {
      "epoch": 0.10750479320734045,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003528,
      "loss": 6.291,
      "step": 785
    },
    {
      "epoch": 0.1076417419884963,
      "grad_norm": 1.015625,
      "learning_rate": 0.00035325,
      "loss": 6.2632,
      "step": 786
    },
    {
      "epoch": 0.10777869076965214,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003537,
      "loss": 6.253,
      "step": 787
    },
    {
      "epoch": 0.107915639550808,
      "grad_norm": 1.03125,
      "learning_rate": 0.00035415,
      "loss": 6.2623,
      "step": 788
    },
    {
      "epoch": 0.10805258833196385,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003546,
      "loss": 6.2701,
      "step": 789
    },
    {
      "epoch": 0.1081895371131197,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00035505,
      "loss": 6.3209,
      "step": 790
    },
    {
      "epoch": 0.10832648589427554,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003555,
      "loss": 6.2622,
      "step": 791
    },
    {
      "epoch": 0.10846343467543139,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00035595,
      "loss": 6.2955,
      "step": 792
    },
    {
      "epoch": 0.10860038345658724,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003564,
      "loss": 6.2393,
      "step": 793
    },
    {
      "epoch": 0.10873733223774308,
      "grad_norm": 1.203125,
      "learning_rate": 0.00035685000000000003,
      "loss": 6.2743,
      "step": 794
    },
    {
      "epoch": 0.10887428101889893,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003573,
      "loss": 6.2827,
      "step": 795
    },
    {
      "epoch": 0.10901122980005477,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00035775,
      "loss": 6.2693,
      "step": 796
    },
    {
      "epoch": 0.10914817858121063,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00035820000000000003,
      "loss": 6.2713,
      "step": 797
    },
    {
      "epoch": 0.10928512736236648,
      "grad_norm": 0.953125,
      "learning_rate": 0.00035865,
      "loss": 6.2867,
      "step": 798
    },
    {
      "epoch": 0.10942207614352233,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003591,
      "loss": 6.2079,
      "step": 799
    },
    {
      "epoch": 0.10955902492467817,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00035955,
      "loss": 6.2453,
      "step": 800
    },
    {
      "epoch": 0.10969597370583402,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00036,
      "loss": 6.2821,
      "step": 801
    },
    {
      "epoch": 0.10983292248698986,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00036045,
      "loss": 6.2805,
      "step": 802
    },
    {
      "epoch": 0.10996987126814571,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003609,
      "loss": 6.1967,
      "step": 803
    },
    {
      "epoch": 0.11010682004930156,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00036135000000000003,
      "loss": 6.2526,
      "step": 804
    },
    {
      "epoch": 0.1102437688304574,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003618,
      "loss": 6.2379,
      "step": 805
    },
    {
      "epoch": 0.11038071761161326,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00036225,
      "loss": 6.2416,
      "step": 806
    },
    {
      "epoch": 0.11051766639276911,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00036270000000000003,
      "loss": 6.2629,
      "step": 807
    },
    {
      "epoch": 0.11065461517392496,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00036315,
      "loss": 6.23,
      "step": 808
    },
    {
      "epoch": 0.1107915639550808,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003636,
      "loss": 6.2696,
      "step": 809
    },
    {
      "epoch": 0.11092851273623665,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00036405000000000004,
      "loss": 6.2414,
      "step": 810
    },
    {
      "epoch": 0.1110654615173925,
      "grad_norm": 0.78125,
      "learning_rate": 0.0003645,
      "loss": 6.2584,
      "step": 811
    },
    {
      "epoch": 0.11120241029854834,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00036495,
      "loss": 6.2397,
      "step": 812
    },
    {
      "epoch": 0.11133935907970419,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003654,
      "loss": 6.219,
      "step": 813
    },
    {
      "epoch": 0.11147630786086003,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00036585,
      "loss": 6.2032,
      "step": 814
    },
    {
      "epoch": 0.11161325664201589,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00036629999999999996,
      "loss": 6.2224,
      "step": 815
    },
    {
      "epoch": 0.11175020542317174,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00036674999999999994,
      "loss": 6.2373,
      "step": 816
    },
    {
      "epoch": 0.11188715420432759,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003672,
      "loss": 6.245,
      "step": 817
    },
    {
      "epoch": 0.11202410298548343,
      "grad_norm": 0.78125,
      "learning_rate": 0.00036764999999999996,
      "loss": 6.3092,
      "step": 818
    },
    {
      "epoch": 0.11216105176663928,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00036809999999999995,
      "loss": 6.2305,
      "step": 819
    },
    {
      "epoch": 0.11229800054779512,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00036855,
      "loss": 6.3361,
      "step": 820
    },
    {
      "epoch": 0.11243494932895097,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00036899999999999997,
      "loss": 6.2656,
      "step": 821
    },
    {
      "epoch": 0.11257189811010682,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00036944999999999995,
      "loss": 6.2249,
      "step": 822
    },
    {
      "epoch": 0.11270884689126266,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003699,
      "loss": 6.2289,
      "step": 823
    },
    {
      "epoch": 0.11284579567241852,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00037035,
      "loss": 6.2614,
      "step": 824
    },
    {
      "epoch": 0.11298274445357437,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00037079999999999996,
      "loss": 6.26,
      "step": 825
    },
    {
      "epoch": 0.11311969323473021,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00037124999999999994,
      "loss": 6.2376,
      "step": 826
    },
    {
      "epoch": 0.11325664201588606,
      "grad_norm": 1.53125,
      "learning_rate": 0.0003717,
      "loss": 6.2834,
      "step": 827
    },
    {
      "epoch": 0.11339359079704191,
      "grad_norm": 1.90625,
      "learning_rate": 0.00037214999999999997,
      "loss": 6.2365,
      "step": 828
    },
    {
      "epoch": 0.11353053957819775,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00037259999999999995,
      "loss": 6.1994,
      "step": 829
    },
    {
      "epoch": 0.1136674883593536,
      "grad_norm": 1.28125,
      "learning_rate": 0.00037305,
      "loss": 6.2832,
      "step": 830
    },
    {
      "epoch": 0.11380443714050945,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00037349999999999997,
      "loss": 6.249,
      "step": 831
    },
    {
      "epoch": 0.11394138592166529,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00037394999999999996,
      "loss": 6.2402,
      "step": 832
    },
    {
      "epoch": 0.11407833470282114,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003744,
      "loss": 6.2106,
      "step": 833
    },
    {
      "epoch": 0.114215283483977,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00037485,
      "loss": 6.2318,
      "step": 834
    },
    {
      "epoch": 0.11435223226513284,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00037529999999999996,
      "loss": 6.2377,
      "step": 835
    },
    {
      "epoch": 0.11448918104628869,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00037575,
      "loss": 6.236,
      "step": 836
    },
    {
      "epoch": 0.11462612982744454,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003762,
      "loss": 6.282,
      "step": 837
    },
    {
      "epoch": 0.11476307860860038,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00037664999999999997,
      "loss": 6.1411,
      "step": 838
    },
    {
      "epoch": 0.11490002738975623,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00037709999999999995,
      "loss": 6.2031,
      "step": 839
    },
    {
      "epoch": 0.11503697617091208,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00037755,
      "loss": 6.2389,
      "step": 840
    },
    {
      "epoch": 0.11517392495206792,
      "grad_norm": 1.03125,
      "learning_rate": 0.00037799999999999997,
      "loss": 6.3064,
      "step": 841
    },
    {
      "epoch": 0.11531087373322377,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00037844999999999996,
      "loss": 6.2629,
      "step": 842
    },
    {
      "epoch": 0.11544782251437963,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003789,
      "loss": 6.1758,
      "step": 843
    },
    {
      "epoch": 0.11558477129553547,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00037935,
      "loss": 6.2383,
      "step": 844
    },
    {
      "epoch": 0.11572172007669132,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00037979999999999996,
      "loss": 6.2256,
      "step": 845
    },
    {
      "epoch": 0.11585866885784717,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00038025,
      "loss": 6.1838,
      "step": 846
    },
    {
      "epoch": 0.11599561763900301,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003807,
      "loss": 6.2387,
      "step": 847
    },
    {
      "epoch": 0.11613256642015886,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00038114999999999997,
      "loss": 6.2527,
      "step": 848
    },
    {
      "epoch": 0.1162695152013147,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003816,
      "loss": 6.1777,
      "step": 849
    },
    {
      "epoch": 0.11640646398247055,
      "grad_norm": 0.875,
      "learning_rate": 0.00038205,
      "loss": 6.2105,
      "step": 850
    },
    {
      "epoch": 0.1165434127636264,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003825,
      "loss": 6.238,
      "step": 851
    },
    {
      "epoch": 0.11668036154478226,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00038294999999999996,
      "loss": 6.244,
      "step": 852
    },
    {
      "epoch": 0.1168173103259381,
      "grad_norm": 1.125,
      "learning_rate": 0.0003834,
      "loss": 6.2579,
      "step": 853
    },
    {
      "epoch": 0.11695425910709395,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00038385,
      "loss": 6.259,
      "step": 854
    },
    {
      "epoch": 0.1170912078882498,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00038429999999999996,
      "loss": 6.265,
      "step": 855
    },
    {
      "epoch": 0.11722815666940564,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00038475,
      "loss": 6.2729,
      "step": 856
    },
    {
      "epoch": 0.11736510545056149,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0003852,
      "loss": 6.2623,
      "step": 857
    },
    {
      "epoch": 0.11750205423171733,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00038564999999999997,
      "loss": 6.1655,
      "step": 858
    },
    {
      "epoch": 0.11763900301287318,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003861,
      "loss": 6.2836,
      "step": 859
    },
    {
      "epoch": 0.11777595179402903,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00038655,
      "loss": 6.234,
      "step": 860
    },
    {
      "epoch": 0.11791290057518489,
      "grad_norm": 1.21875,
      "learning_rate": 0.000387,
      "loss": 6.2131,
      "step": 861
    },
    {
      "epoch": 0.11804984935634073,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00038744999999999996,
      "loss": 6.2708,
      "step": 862
    },
    {
      "epoch": 0.11818679813749658,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003879,
      "loss": 6.2163,
      "step": 863
    },
    {
      "epoch": 0.11832374691865243,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00038835,
      "loss": 6.2154,
      "step": 864
    },
    {
      "epoch": 0.11846069569980827,
      "grad_norm": 1.171875,
      "learning_rate": 0.00038879999999999996,
      "loss": 6.2083,
      "step": 865
    },
    {
      "epoch": 0.11859764448096412,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00038925,
      "loss": 6.2295,
      "step": 866
    },
    {
      "epoch": 0.11873459326211996,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003897,
      "loss": 6.1861,
      "step": 867
    },
    {
      "epoch": 0.11887154204327581,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00039014999999999997,
      "loss": 6.2292,
      "step": 868
    },
    {
      "epoch": 0.11900849082443166,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003906,
      "loss": 6.2101,
      "step": 869
    },
    {
      "epoch": 0.11914543960558752,
      "grad_norm": 1.09375,
      "learning_rate": 0.00039105,
      "loss": 6.2544,
      "step": 870
    },
    {
      "epoch": 0.11928238838674336,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003915,
      "loss": 6.2318,
      "step": 871
    },
    {
      "epoch": 0.11941933716789921,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00039195,
      "loss": 6.2401,
      "step": 872
    },
    {
      "epoch": 0.11955628594905506,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003924,
      "loss": 6.1788,
      "step": 873
    },
    {
      "epoch": 0.1196932347302109,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00039285,
      "loss": 6.2308,
      "step": 874
    },
    {
      "epoch": 0.11983018351136675,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00039329999999999996,
      "loss": 6.1926,
      "step": 875
    },
    {
      "epoch": 0.1199671322925226,
      "grad_norm": 1.34375,
      "learning_rate": 0.00039375,
      "loss": 6.2044,
      "step": 876
    },
    {
      "epoch": 0.12010408107367844,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0003942,
      "loss": 6.2696,
      "step": 877
    },
    {
      "epoch": 0.12024102985483429,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00039464999999999997,
      "loss": 6.265,
      "step": 878
    },
    {
      "epoch": 0.12037797863599015,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003951,
      "loss": 6.1745,
      "step": 879
    },
    {
      "epoch": 0.12051492741714599,
      "grad_norm": 1.296875,
      "learning_rate": 0.00039555,
      "loss": 6.222,
      "step": 880
    },
    {
      "epoch": 0.12065187619830184,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000396,
      "loss": 6.1884,
      "step": 881
    },
    {
      "epoch": 0.12078882497945768,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00039645,
      "loss": 6.1907,
      "step": 882
    },
    {
      "epoch": 0.12092577376061353,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003969,
      "loss": 6.2228,
      "step": 883
    },
    {
      "epoch": 0.12106272254176938,
      "grad_norm": 1.046875,
      "learning_rate": 0.00039735,
      "loss": 6.1972,
      "step": 884
    },
    {
      "epoch": 0.12119967132292522,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00039779999999999997,
      "loss": 6.2326,
      "step": 885
    },
    {
      "epoch": 0.12133662010408107,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00039825,
      "loss": 6.1812,
      "step": 886
    },
    {
      "epoch": 0.12147356888523692,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003987,
      "loss": 6.2028,
      "step": 887
    },
    {
      "epoch": 0.12161051766639278,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00039914999999999997,
      "loss": 6.1861,
      "step": 888
    },
    {
      "epoch": 0.12174746644754862,
      "grad_norm": 0.8125,
      "learning_rate": 0.0003996,
      "loss": 6.1866,
      "step": 889
    },
    {
      "epoch": 0.12188441522870447,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00040005,
      "loss": 6.1869,
      "step": 890
    },
    {
      "epoch": 0.12202136400986031,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004005,
      "loss": 6.251,
      "step": 891
    },
    {
      "epoch": 0.12215831279101616,
      "grad_norm": 1.078125,
      "learning_rate": 0.00040095,
      "loss": 6.1985,
      "step": 892
    },
    {
      "epoch": 0.122295261572172,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004014,
      "loss": 6.1784,
      "step": 893
    },
    {
      "epoch": 0.12243221035332785,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00040185,
      "loss": 6.2426,
      "step": 894
    },
    {
      "epoch": 0.1225691591344837,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004023,
      "loss": 6.2103,
      "step": 895
    },
    {
      "epoch": 0.12270610791563955,
      "grad_norm": 0.9375,
      "learning_rate": 0.00040275,
      "loss": 6.2405,
      "step": 896
    },
    {
      "epoch": 0.1228430566967954,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004032,
      "loss": 6.1847,
      "step": 897
    },
    {
      "epoch": 0.12298000547795125,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00040364999999999997,
      "loss": 6.2261,
      "step": 898
    },
    {
      "epoch": 0.1231169542591071,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004041,
      "loss": 6.1568,
      "step": 899
    },
    {
      "epoch": 0.12325390304026294,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00040455,
      "loss": 6.1979,
      "step": 900
    },
    {
      "epoch": 0.12339085182141879,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000405,
      "loss": 6.1801,
      "step": 901
    },
    {
      "epoch": 0.12352780060257464,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00040545,
      "loss": 6.2061,
      "step": 902
    },
    {
      "epoch": 0.12366474938373048,
      "grad_norm": 0.765625,
      "learning_rate": 0.0004059,
      "loss": 6.1918,
      "step": 903
    },
    {
      "epoch": 0.12380169816488633,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00040635,
      "loss": 6.2526,
      "step": 904
    },
    {
      "epoch": 0.12393864694604217,
      "grad_norm": 1.0,
      "learning_rate": 0.0004068,
      "loss": 6.1522,
      "step": 905
    },
    {
      "epoch": 0.12407559572719803,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00040725,
      "loss": 6.1332,
      "step": 906
    },
    {
      "epoch": 0.12421254450835388,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004077,
      "loss": 6.1297,
      "step": 907
    },
    {
      "epoch": 0.12434949328950973,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00040815000000000003,
      "loss": 6.1986,
      "step": 908
    },
    {
      "epoch": 0.12448644207066557,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0004086,
      "loss": 6.2048,
      "step": 909
    },
    {
      "epoch": 0.12462339085182142,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00040905,
      "loss": 6.1721,
      "step": 910
    },
    {
      "epoch": 0.12476033963297727,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0004095,
      "loss": 6.2574,
      "step": 911
    },
    {
      "epoch": 0.12489728841413311,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00040995,
      "loss": 6.2164,
      "step": 912
    },
    {
      "epoch": 0.12503423719528897,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0004104,
      "loss": 6.0989,
      "step": 913
    },
    {
      "epoch": 0.1251711859764448,
      "grad_norm": 0.78125,
      "learning_rate": 0.00041085,
      "loss": 6.1947,
      "step": 914
    },
    {
      "epoch": 0.12530813475760066,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0004113,
      "loss": 6.2156,
      "step": 915
    },
    {
      "epoch": 0.1254450835387565,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00041175,
      "loss": 6.1662,
      "step": 916
    },
    {
      "epoch": 0.12558203231991236,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004122,
      "loss": 6.1833,
      "step": 917
    },
    {
      "epoch": 0.1257189811010682,
      "grad_norm": 0.9375,
      "learning_rate": 0.00041265000000000003,
      "loss": 6.2269,
      "step": 918
    },
    {
      "epoch": 0.12585592988222405,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004131,
      "loss": 6.1649,
      "step": 919
    },
    {
      "epoch": 0.1259928786633799,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00041355,
      "loss": 6.2093,
      "step": 920
    },
    {
      "epoch": 0.12612982744453574,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000414,
      "loss": 6.1663,
      "step": 921
    },
    {
      "epoch": 0.1262667762256916,
      "grad_norm": 0.8125,
      "learning_rate": 0.00041445,
      "loss": 6.2576,
      "step": 922
    },
    {
      "epoch": 0.12640372500684743,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004149,
      "loss": 6.2365,
      "step": 923
    },
    {
      "epoch": 0.1265406737880033,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00041535,
      "loss": 6.2155,
      "step": 924
    },
    {
      "epoch": 0.12667762256915913,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004158,
      "loss": 6.1753,
      "step": 925
    },
    {
      "epoch": 0.126814571350315,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00041625,
      "loss": 6.2024,
      "step": 926
    },
    {
      "epoch": 0.12695152013147082,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004167,
      "loss": 6.2112,
      "step": 927
    },
    {
      "epoch": 0.12708846891262668,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00041715000000000003,
      "loss": 6.1901,
      "step": 928
    },
    {
      "epoch": 0.12722541769378254,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004176,
      "loss": 6.1744,
      "step": 929
    },
    {
      "epoch": 0.12736236647493837,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00041805,
      "loss": 6.1897,
      "step": 930
    },
    {
      "epoch": 0.12749931525609423,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00041850000000000004,
      "loss": 6.1958,
      "step": 931
    },
    {
      "epoch": 0.12763626403725006,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00041895,
      "loss": 6.1991,
      "step": 932
    },
    {
      "epoch": 0.12777321281840592,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004194,
      "loss": 6.1771,
      "step": 933
    },
    {
      "epoch": 0.12791016159956176,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00041985,
      "loss": 6.1939,
      "step": 934
    },
    {
      "epoch": 0.12804711038071762,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004203,
      "loss": 6.2048,
      "step": 935
    },
    {
      "epoch": 0.12818405916187345,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00042075,
      "loss": 6.1734,
      "step": 936
    },
    {
      "epoch": 0.1283210079430293,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004212,
      "loss": 6.179,
      "step": 937
    },
    {
      "epoch": 0.12845795672418517,
      "grad_norm": 0.953125,
      "learning_rate": 0.00042165000000000003,
      "loss": 6.1781,
      "step": 938
    },
    {
      "epoch": 0.128594905505341,
      "grad_norm": 0.859375,
      "learning_rate": 0.00042209999999999996,
      "loss": 6.1604,
      "step": 939
    },
    {
      "epoch": 0.12873185428649686,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00042254999999999994,
      "loss": 6.0819,
      "step": 940
    },
    {
      "epoch": 0.1288688030676527,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000423,
      "loss": 6.147,
      "step": 941
    },
    {
      "epoch": 0.12900575184880855,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00042344999999999997,
      "loss": 6.1369,
      "step": 942
    },
    {
      "epoch": 0.12914270062996439,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00042389999999999995,
      "loss": 6.1731,
      "step": 943
    },
    {
      "epoch": 0.12927964941112025,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00042435,
      "loss": 6.1973,
      "step": 944
    },
    {
      "epoch": 0.12941659819227608,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00042479999999999997,
      "loss": 6.2076,
      "step": 945
    },
    {
      "epoch": 0.12955354697343194,
      "grad_norm": 1.203125,
      "learning_rate": 0.00042524999999999996,
      "loss": 6.2125,
      "step": 946
    },
    {
      "epoch": 0.1296904957545878,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004257,
      "loss": 6.1499,
      "step": 947
    },
    {
      "epoch": 0.12982744453574363,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00042615,
      "loss": 6.2246,
      "step": 948
    },
    {
      "epoch": 0.1299643933168995,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00042659999999999996,
      "loss": 6.1768,
      "step": 949
    },
    {
      "epoch": 0.13010134209805532,
      "grad_norm": 0.875,
      "learning_rate": 0.00042704999999999994,
      "loss": 6.1648,
      "step": 950
    },
    {
      "epoch": 0.13023829087921118,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004275,
      "loss": 6.1626,
      "step": 951
    },
    {
      "epoch": 0.13037523966036701,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00042794999999999997,
      "loss": 6.1201,
      "step": 952
    },
    {
      "epoch": 0.13051218844152288,
      "grad_norm": 1.3125,
      "learning_rate": 0.00042839999999999995,
      "loss": 6.1215,
      "step": 953
    },
    {
      "epoch": 0.1306491372226787,
      "grad_norm": 1.171875,
      "learning_rate": 0.00042885,
      "loss": 6.1903,
      "step": 954
    },
    {
      "epoch": 0.13078608600383457,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00042929999999999997,
      "loss": 6.1463,
      "step": 955
    },
    {
      "epoch": 0.13092303478499043,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00042974999999999996,
      "loss": 6.1582,
      "step": 956
    },
    {
      "epoch": 0.13105998356614626,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004302,
      "loss": 6.1411,
      "step": 957
    },
    {
      "epoch": 0.13119693234730212,
      "grad_norm": 0.90625,
      "learning_rate": 0.00043065,
      "loss": 6.1981,
      "step": 958
    },
    {
      "epoch": 0.13133388112845795,
      "grad_norm": 0.953125,
      "learning_rate": 0.00043109999999999996,
      "loss": 6.1956,
      "step": 959
    },
    {
      "epoch": 0.1314708299096138,
      "grad_norm": 1.28125,
      "learning_rate": 0.00043154999999999995,
      "loss": 6.1954,
      "step": 960
    },
    {
      "epoch": 0.13160777869076964,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000432,
      "loss": 6.1466,
      "step": 961
    },
    {
      "epoch": 0.1317447274719255,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00043244999999999997,
      "loss": 6.1692,
      "step": 962
    },
    {
      "epoch": 0.13188167625308134,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00043289999999999995,
      "loss": 6.2056,
      "step": 963
    },
    {
      "epoch": 0.1320186250342372,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00043335,
      "loss": 6.1253,
      "step": 964
    },
    {
      "epoch": 0.13215557381539303,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004338,
      "loss": 6.1554,
      "step": 965
    },
    {
      "epoch": 0.1322925225965489,
      "grad_norm": 0.984375,
      "learning_rate": 0.00043424999999999996,
      "loss": 6.1391,
      "step": 966
    },
    {
      "epoch": 0.13242947137770475,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004347,
      "loss": 6.1474,
      "step": 967
    },
    {
      "epoch": 0.13256642015886058,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00043515,
      "loss": 6.1563,
      "step": 968
    },
    {
      "epoch": 0.13270336894001644,
      "grad_norm": 1.109375,
      "learning_rate": 0.00043559999999999996,
      "loss": 6.1256,
      "step": 969
    },
    {
      "epoch": 0.13284031772117227,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00043605,
      "loss": 6.1541,
      "step": 970
    },
    {
      "epoch": 0.13297726650232813,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004365,
      "loss": 6.1551,
      "step": 971
    },
    {
      "epoch": 0.13311421528348397,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00043694999999999997,
      "loss": 6.1497,
      "step": 972
    },
    {
      "epoch": 0.13325116406463983,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00043739999999999995,
      "loss": 6.1548,
      "step": 973
    },
    {
      "epoch": 0.13338811284579566,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00043785,
      "loss": 6.1237,
      "step": 974
    },
    {
      "epoch": 0.13352506162695152,
      "grad_norm": 0.734375,
      "learning_rate": 0.0004383,
      "loss": 6.1287,
      "step": 975
    },
    {
      "epoch": 0.13366201040810738,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00043874999999999996,
      "loss": 6.1667,
      "step": 976
    },
    {
      "epoch": 0.1337989591892632,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004392,
      "loss": 6.143,
      "step": 977
    },
    {
      "epoch": 0.13393590797041907,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00043965,
      "loss": 6.1445,
      "step": 978
    },
    {
      "epoch": 0.1340728567515749,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00044009999999999996,
      "loss": 6.1931,
      "step": 979
    },
    {
      "epoch": 0.13420980553273076,
      "grad_norm": 0.765625,
      "learning_rate": 0.00044055,
      "loss": 6.1693,
      "step": 980
    },
    {
      "epoch": 0.1343467543138866,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000441,
      "loss": 6.1629,
      "step": 981
    },
    {
      "epoch": 0.13448370309504246,
      "grad_norm": 0.921875,
      "learning_rate": 0.00044144999999999997,
      "loss": 6.1361,
      "step": 982
    },
    {
      "epoch": 0.1346206518761983,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004419,
      "loss": 6.149,
      "step": 983
    },
    {
      "epoch": 0.13475760065735415,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00044235,
      "loss": 6.0924,
      "step": 984
    },
    {
      "epoch": 0.13489454943851,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004428,
      "loss": 6.1549,
      "step": 985
    },
    {
      "epoch": 0.13503149821966584,
      "grad_norm": 1.109375,
      "learning_rate": 0.00044324999999999996,
      "loss": 6.1392,
      "step": 986
    },
    {
      "epoch": 0.1351684470008217,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004437,
      "loss": 6.1788,
      "step": 987
    },
    {
      "epoch": 0.13530539578197753,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00044415,
      "loss": 6.1533,
      "step": 988
    },
    {
      "epoch": 0.1354423445631334,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00044459999999999996,
      "loss": 6.1633,
      "step": 989
    },
    {
      "epoch": 0.13557929334428923,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00044505,
      "loss": 6.1277,
      "step": 990
    },
    {
      "epoch": 0.13571624212544509,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004455,
      "loss": 6.1686,
      "step": 991
    },
    {
      "epoch": 0.13585319090660092,
      "grad_norm": 0.890625,
      "learning_rate": 0.00044594999999999997,
      "loss": 6.0806,
      "step": 992
    },
    {
      "epoch": 0.13599013968775678,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004464,
      "loss": 6.1344,
      "step": 993
    },
    {
      "epoch": 0.13612708846891264,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00044685,
      "loss": 6.1429,
      "step": 994
    },
    {
      "epoch": 0.13626403725006847,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0004473,
      "loss": 6.1334,
      "step": 995
    },
    {
      "epoch": 0.13640098603122433,
      "grad_norm": 0.859375,
      "learning_rate": 0.00044774999999999996,
      "loss": 6.1015,
      "step": 996
    },
    {
      "epoch": 0.13653793481238016,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004482,
      "loss": 6.2356,
      "step": 997
    },
    {
      "epoch": 0.13667488359353602,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00044865,
      "loss": 6.1747,
      "step": 998
    },
    {
      "epoch": 0.13681183237469186,
      "grad_norm": 0.8125,
      "learning_rate": 0.00044909999999999997,
      "loss": 6.1768,
      "step": 999
    },
    {
      "epoch": 0.13694878115584772,
      "grad_norm": 1.0625,
      "learning_rate": 0.00044955,
      "loss": 6.2082,
      "step": 1000
    },
    {
      "epoch": 0.13708572993700355,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00045,
      "loss": 6.1549,
      "step": 1001
    },
    {
      "epoch": 0.1372226787181594,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004504499999999999,
      "loss": 6.1509,
      "step": 1002
    },
    {
      "epoch": 0.13735962749931527,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004509,
      "loss": 6.1854,
      "step": 1003
    },
    {
      "epoch": 0.1374965762804711,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00045134999999999994,
      "loss": 6.1602,
      "step": 1004
    },
    {
      "epoch": 0.13763352506162696,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004518,
      "loss": 6.1905,
      "step": 1005
    },
    {
      "epoch": 0.1377704738427828,
      "grad_norm": 1.0,
      "learning_rate": 0.00045224999999999996,
      "loss": 6.1479,
      "step": 1006
    },
    {
      "epoch": 0.13790742262393865,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004527,
      "loss": 6.1136,
      "step": 1007
    },
    {
      "epoch": 0.13804437140509448,
      "grad_norm": 1.15625,
      "learning_rate": 0.00045314999999999993,
      "loss": 6.1399,
      "step": 1008
    },
    {
      "epoch": 0.13818132018625034,
      "grad_norm": 1.015625,
      "learning_rate": 0.00045359999999999997,
      "loss": 6.195,
      "step": 1009
    },
    {
      "epoch": 0.13831826896740618,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00045404999999999995,
      "loss": 6.2244,
      "step": 1010
    },
    {
      "epoch": 0.13845521774856204,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004545,
      "loss": 6.2181,
      "step": 1011
    },
    {
      "epoch": 0.1385921665297179,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004549499999999999,
      "loss": 6.104,
      "step": 1012
    },
    {
      "epoch": 0.13872911531087373,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004554,
      "loss": 6.1463,
      "step": 1013
    },
    {
      "epoch": 0.1388660640920296,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00045584999999999994,
      "loss": 6.1499,
      "step": 1014
    },
    {
      "epoch": 0.13900301287318542,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004563,
      "loss": 6.1231,
      "step": 1015
    },
    {
      "epoch": 0.13913996165434128,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00045674999999999996,
      "loss": 6.0886,
      "step": 1016
    },
    {
      "epoch": 0.13927691043549711,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0004572,
      "loss": 6.1662,
      "step": 1017
    },
    {
      "epoch": 0.13941385921665297,
      "grad_norm": 0.84375,
      "learning_rate": 0.00045764999999999993,
      "loss": 6.1165,
      "step": 1018
    },
    {
      "epoch": 0.1395508079978088,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00045809999999999997,
      "loss": 6.1281,
      "step": 1019
    },
    {
      "epoch": 0.13968775677896467,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00045854999999999995,
      "loss": 6.1144,
      "step": 1020
    },
    {
      "epoch": 0.13982470556012053,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000459,
      "loss": 6.1528,
      "step": 1021
    },
    {
      "epoch": 0.13996165434127636,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004594499999999999,
      "loss": 6.1457,
      "step": 1022
    },
    {
      "epoch": 0.14009860312243222,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004599,
      "loss": 6.1299,
      "step": 1023
    },
    {
      "epoch": 0.14023555190358805,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00046034999999999994,
      "loss": 6.1762,
      "step": 1024
    },
    {
      "epoch": 0.1403725006847439,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004608,
      "loss": 6.131,
      "step": 1025
    },
    {
      "epoch": 0.14050944946589974,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00046124999999999996,
      "loss": 6.1546,
      "step": 1026
    },
    {
      "epoch": 0.1406463982470556,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004617,
      "loss": 6.1564,
      "step": 1027
    },
    {
      "epoch": 0.14078334702821144,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00046214999999999993,
      "loss": 6.1332,
      "step": 1028
    },
    {
      "epoch": 0.1409202958093673,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0004626,
      "loss": 6.099,
      "step": 1029
    },
    {
      "epoch": 0.14105724459052316,
      "grad_norm": 0.96875,
      "learning_rate": 0.00046304999999999995,
      "loss": 6.0668,
      "step": 1030
    },
    {
      "epoch": 0.141194193371679,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004635,
      "loss": 6.1294,
      "step": 1031
    },
    {
      "epoch": 0.14133114215283485,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00046395,
      "loss": 6.1507,
      "step": 1032
    },
    {
      "epoch": 0.14146809093399068,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004644,
      "loss": 6.1226,
      "step": 1033
    },
    {
      "epoch": 0.14160503971514654,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00046484999999999994,
      "loss": 6.0892,
      "step": 1034
    },
    {
      "epoch": 0.14174198849630237,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004653,
      "loss": 6.1566,
      "step": 1035
    },
    {
      "epoch": 0.14187893727745823,
      "grad_norm": 0.828125,
      "learning_rate": 0.00046574999999999996,
      "loss": 6.0949,
      "step": 1036
    },
    {
      "epoch": 0.14201588605861407,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004662,
      "loss": 6.1442,
      "step": 1037
    },
    {
      "epoch": 0.14215283483976993,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00046664999999999993,
      "loss": 6.1408,
      "step": 1038
    },
    {
      "epoch": 0.14228978362092579,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0004671,
      "loss": 6.1803,
      "step": 1039
    },
    {
      "epoch": 0.14242673240208162,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00046754999999999995,
      "loss": 6.1364,
      "step": 1040
    },
    {
      "epoch": 0.14256368118323748,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000468,
      "loss": 6.1287,
      "step": 1041
    },
    {
      "epoch": 0.1427006299643933,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00046845,
      "loss": 6.1116,
      "step": 1042
    },
    {
      "epoch": 0.14283757874554917,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0004689,
      "loss": 6.0613,
      "step": 1043
    },
    {
      "epoch": 0.142974527526705,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00046934999999999994,
      "loss": 6.1393,
      "step": 1044
    },
    {
      "epoch": 0.14311147630786086,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004698,
      "loss": 6.1562,
      "step": 1045
    },
    {
      "epoch": 0.1432484250890167,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00047024999999999996,
      "loss": 6.1678,
      "step": 1046
    },
    {
      "epoch": 0.14338537387017256,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004707,
      "loss": 6.0567,
      "step": 1047
    },
    {
      "epoch": 0.14352232265132842,
      "grad_norm": 0.921875,
      "learning_rate": 0.00047114999999999993,
      "loss": 6.1044,
      "step": 1048
    },
    {
      "epoch": 0.14365927143248425,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004716,
      "loss": 6.0982,
      "step": 1049
    },
    {
      "epoch": 0.1437962202136401,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00047204999999999995,
      "loss": 6.1268,
      "step": 1050
    },
    {
      "epoch": 0.14393316899479594,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0004725,
      "loss": 6.1003,
      "step": 1051
    },
    {
      "epoch": 0.1440701177759518,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00047295,
      "loss": 6.1514,
      "step": 1052
    },
    {
      "epoch": 0.14420706655710763,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004734,
      "loss": 6.0776,
      "step": 1053
    },
    {
      "epoch": 0.1443440153382635,
      "grad_norm": 0.875,
      "learning_rate": 0.00047384999999999994,
      "loss": 6.146,
      "step": 1054
    },
    {
      "epoch": 0.14448096411941933,
      "grad_norm": 0.703125,
      "learning_rate": 0.0004743,
      "loss": 6.1446,
      "step": 1055
    },
    {
      "epoch": 0.14461791290057519,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00047474999999999997,
      "loss": 6.0671,
      "step": 1056
    },
    {
      "epoch": 0.14475486168173105,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004752,
      "loss": 6.1627,
      "step": 1057
    },
    {
      "epoch": 0.14489181046288688,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00047564999999999993,
      "loss": 6.1487,
      "step": 1058
    },
    {
      "epoch": 0.14502875924404274,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004761,
      "loss": 6.1367,
      "step": 1059
    },
    {
      "epoch": 0.14516570802519857,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00047654999999999996,
      "loss": 6.1341,
      "step": 1060
    },
    {
      "epoch": 0.14530265680635443,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000477,
      "loss": 6.1385,
      "step": 1061
    },
    {
      "epoch": 0.14543960558751026,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00047745,
      "loss": 6.142,
      "step": 1062
    },
    {
      "epoch": 0.14557655436866612,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0004779,
      "loss": 6.1591,
      "step": 1063
    },
    {
      "epoch": 0.14571350314982195,
      "grad_norm": 0.9375,
      "learning_rate": 0.00047834999999999994,
      "loss": 6.1555,
      "step": 1064
    },
    {
      "epoch": 0.14585045193097781,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00047880000000000004,
      "loss": 6.1188,
      "step": 1065
    },
    {
      "epoch": 0.14598740071213367,
      "grad_norm": 1.4375,
      "learning_rate": 0.00047924999999999997,
      "loss": 6.1246,
      "step": 1066
    },
    {
      "epoch": 0.1461243494932895,
      "grad_norm": 1.5625,
      "learning_rate": 0.0004797,
      "loss": 6.1073,
      "step": 1067
    },
    {
      "epoch": 0.14626129827444537,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00048015,
      "loss": 6.0908,
      "step": 1068
    },
    {
      "epoch": 0.1463982470556012,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004806,
      "loss": 6.1717,
      "step": 1069
    },
    {
      "epoch": 0.14653519583675706,
      "grad_norm": 1.296875,
      "learning_rate": 0.00048104999999999996,
      "loss": 6.1349,
      "step": 1070
    },
    {
      "epoch": 0.1466721446179129,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004815,
      "loss": 6.207,
      "step": 1071
    },
    {
      "epoch": 0.14680909339906875,
      "grad_norm": 0.8125,
      "learning_rate": 0.00048195,
      "loss": 6.1578,
      "step": 1072
    },
    {
      "epoch": 0.14694604218022458,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004824,
      "loss": 6.1493,
      "step": 1073
    },
    {
      "epoch": 0.14708299096138044,
      "grad_norm": 1.0625,
      "learning_rate": 0.00048284999999999995,
      "loss": 6.1176,
      "step": 1074
    },
    {
      "epoch": 0.1472199397425363,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00048330000000000004,
      "loss": 6.1284,
      "step": 1075
    },
    {
      "epoch": 0.14735688852369214,
      "grad_norm": 0.84375,
      "learning_rate": 0.00048374999999999997,
      "loss": 6.1058,
      "step": 1076
    },
    {
      "epoch": 0.147493837304848,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0004842,
      "loss": 6.124,
      "step": 1077
    },
    {
      "epoch": 0.14763078608600383,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00048465,
      "loss": 6.125,
      "step": 1078
    },
    {
      "epoch": 0.1477677348671597,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00048510000000000003,
      "loss": 6.1522,
      "step": 1079
    },
    {
      "epoch": 0.14790468364831552,
      "grad_norm": 0.828125,
      "learning_rate": 0.00048554999999999996,
      "loss": 6.1065,
      "step": 1080
    },
    {
      "epoch": 0.14804163242947138,
      "grad_norm": 0.7421875,
      "learning_rate": 0.000486,
      "loss": 6.1129,
      "step": 1081
    },
    {
      "epoch": 0.1481785812106272,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00048645,
      "loss": 6.0943,
      "step": 1082
    },
    {
      "epoch": 0.14831552999178307,
      "grad_norm": 0.734375,
      "learning_rate": 0.0004869,
      "loss": 6.1638,
      "step": 1083
    },
    {
      "epoch": 0.14845247877293893,
      "grad_norm": 0.640625,
      "learning_rate": 0.00048734999999999995,
      "loss": 6.1732,
      "step": 1084
    },
    {
      "epoch": 0.14858942755409477,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00048780000000000004,
      "loss": 6.0756,
      "step": 1085
    },
    {
      "epoch": 0.14872637633525063,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00048824999999999997,
      "loss": 6.1347,
      "step": 1086
    },
    {
      "epoch": 0.14886332511640646,
      "grad_norm": 0.6875,
      "learning_rate": 0.0004887,
      "loss": 6.1237,
      "step": 1087
    },
    {
      "epoch": 0.14900027389756232,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0004891499999999999,
      "loss": 6.103,
      "step": 1088
    },
    {
      "epoch": 0.14913722267871815,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0004896,
      "loss": 6.0781,
      "step": 1089
    },
    {
      "epoch": 0.149274171459874,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00049005,
      "loss": 6.1087,
      "step": 1090
    },
    {
      "epoch": 0.14941112024102984,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004905,
      "loss": 6.0974,
      "step": 1091
    },
    {
      "epoch": 0.1495480690221857,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00049095,
      "loss": 6.159,
      "step": 1092
    },
    {
      "epoch": 0.14968501780334156,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004914,
      "loss": 6.1074,
      "step": 1093
    },
    {
      "epoch": 0.1498219665844974,
      "grad_norm": 0.890625,
      "learning_rate": 0.00049185,
      "loss": 6.1542,
      "step": 1094
    },
    {
      "epoch": 0.14995891536565326,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004923,
      "loss": 6.1347,
      "step": 1095
    },
    {
      "epoch": 0.1500958641468091,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00049275,
      "loss": 6.1139,
      "step": 1096
    },
    {
      "epoch": 0.15023281292796495,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004932000000000001,
      "loss": 6.0904,
      "step": 1097
    },
    {
      "epoch": 0.15036976170912078,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00049365,
      "loss": 6.1167,
      "step": 1098
    },
    {
      "epoch": 0.15050671049027664,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004941,
      "loss": 6.1101,
      "step": 1099
    },
    {
      "epoch": 0.15064365927143247,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00049455,
      "loss": 6.105,
      "step": 1100
    },
    {
      "epoch": 0.15078060805258833,
      "grad_norm": 0.859375,
      "learning_rate": 0.000495,
      "loss": 6.0973,
      "step": 1101
    },
    {
      "epoch": 0.1509175568337442,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004954499999999999,
      "loss": 6.079,
      "step": 1102
    },
    {
      "epoch": 0.15105450561490003,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004959000000000001,
      "loss": 6.1566,
      "step": 1103
    },
    {
      "epoch": 0.15119145439605589,
      "grad_norm": 0.765625,
      "learning_rate": 0.00049635,
      "loss": 6.1063,
      "step": 1104
    },
    {
      "epoch": 0.15132840317721172,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004968,
      "loss": 6.0883,
      "step": 1105
    },
    {
      "epoch": 0.15146535195836758,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00049725,
      "loss": 6.0799,
      "step": 1106
    },
    {
      "epoch": 0.1516023007395234,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004977,
      "loss": 6.1087,
      "step": 1107
    },
    {
      "epoch": 0.15173924952067927,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004981499999999999,
      "loss": 6.0972,
      "step": 1108
    },
    {
      "epoch": 0.1518761983018351,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004986,
      "loss": 6.0765,
      "step": 1109
    },
    {
      "epoch": 0.15201314708299096,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00049905,
      "loss": 6.0923,
      "step": 1110
    },
    {
      "epoch": 0.15215009586414682,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004995,
      "loss": 6.1236,
      "step": 1111
    },
    {
      "epoch": 0.15228704464530265,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00049995,
      "loss": 6.0974,
      "step": 1112
    },
    {
      "epoch": 0.15242399342645852,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005004,
      "loss": 6.084,
      "step": 1113
    },
    {
      "epoch": 0.15256094220761435,
      "grad_norm": 1.03125,
      "learning_rate": 0.00050085,
      "loss": 6.1047,
      "step": 1114
    },
    {
      "epoch": 0.1526978909887702,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005013,
      "loss": 6.0826,
      "step": 1115
    },
    {
      "epoch": 0.15283483976992604,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00050175,
      "loss": 6.0887,
      "step": 1116
    },
    {
      "epoch": 0.1529717885510819,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0005022000000000001,
      "loss": 6.0895,
      "step": 1117
    },
    {
      "epoch": 0.15310873733223773,
      "grad_norm": 0.78125,
      "learning_rate": 0.00050265,
      "loss": 6.1103,
      "step": 1118
    },
    {
      "epoch": 0.1532456861133936,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005031,
      "loss": 6.123,
      "step": 1119
    },
    {
      "epoch": 0.15338263489454945,
      "grad_norm": 0.984375,
      "learning_rate": 0.00050355,
      "loss": 6.0268,
      "step": 1120
    },
    {
      "epoch": 0.15351958367570528,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000504,
      "loss": 6.0359,
      "step": 1121
    },
    {
      "epoch": 0.15365653245686114,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00050445,
      "loss": 6.073,
      "step": 1122
    },
    {
      "epoch": 0.15379348123801698,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005049000000000001,
      "loss": 6.0649,
      "step": 1123
    },
    {
      "epoch": 0.15393043001917284,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00050535,
      "loss": 6.0466,
      "step": 1124
    },
    {
      "epoch": 0.15406737880032867,
      "grad_norm": 0.875,
      "learning_rate": 0.0005058,
      "loss": 6.1229,
      "step": 1125
    },
    {
      "epoch": 0.15420432758148453,
      "grad_norm": 1.078125,
      "learning_rate": 0.00050625,
      "loss": 6.0895,
      "step": 1126
    },
    {
      "epoch": 0.15434127636264036,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005066999999999999,
      "loss": 6.0028,
      "step": 1127
    },
    {
      "epoch": 0.15447822514379622,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005071499999999999,
      "loss": 6.1149,
      "step": 1128
    },
    {
      "epoch": 0.15461517392495208,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005076,
      "loss": 6.1115,
      "step": 1129
    },
    {
      "epoch": 0.15475212270610791,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00050805,
      "loss": 6.0724,
      "step": 1130
    },
    {
      "epoch": 0.15488907148726377,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005085,
      "loss": 6.0997,
      "step": 1131
    },
    {
      "epoch": 0.1550260202684196,
      "grad_norm": 1.078125,
      "learning_rate": 0.00050895,
      "loss": 6.0796,
      "step": 1132
    },
    {
      "epoch": 0.15516296904957547,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005093999999999999,
      "loss": 6.0715,
      "step": 1133
    },
    {
      "epoch": 0.1552999178307313,
      "grad_norm": 1.140625,
      "learning_rate": 0.00050985,
      "loss": 6.0653,
      "step": 1134
    },
    {
      "epoch": 0.15543686661188716,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005103,
      "loss": 6.134,
      "step": 1135
    },
    {
      "epoch": 0.155573815393043,
      "grad_norm": 0.96875,
      "learning_rate": 0.00051075,
      "loss": 6.1028,
      "step": 1136
    },
    {
      "epoch": 0.15571076417419885,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005112,
      "loss": 6.1112,
      "step": 1137
    },
    {
      "epoch": 0.15584771295535468,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00051165,
      "loss": 6.1013,
      "step": 1138
    },
    {
      "epoch": 0.15598466173651054,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005120999999999999,
      "loss": 6.0894,
      "step": 1139
    },
    {
      "epoch": 0.1561216105176664,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00051255,
      "loss": 6.1344,
      "step": 1140
    },
    {
      "epoch": 0.15625855929882224,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005129999999999999,
      "loss": 6.0806,
      "step": 1141
    },
    {
      "epoch": 0.1563955080799781,
      "grad_norm": 1.046875,
      "learning_rate": 0.00051345,
      "loss": 6.1356,
      "step": 1142
    },
    {
      "epoch": 0.15653245686113393,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005139,
      "loss": 6.1359,
      "step": 1143
    },
    {
      "epoch": 0.1566694056422898,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00051435,
      "loss": 6.0965,
      "step": 1144
    },
    {
      "epoch": 0.15680635442344562,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005147999999999999,
      "loss": 6.0737,
      "step": 1145
    },
    {
      "epoch": 0.15694330320460148,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00051525,
      "loss": 6.0778,
      "step": 1146
    },
    {
      "epoch": 0.1570802519857573,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005156999999999999,
      "loss": 6.0804,
      "step": 1147
    },
    {
      "epoch": 0.15721720076691317,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005161499999999999,
      "loss": 6.0157,
      "step": 1148
    },
    {
      "epoch": 0.15735414954806903,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005166,
      "loss": 6.1231,
      "step": 1149
    },
    {
      "epoch": 0.15749109832922487,
      "grad_norm": 1.03125,
      "learning_rate": 0.00051705,
      "loss": 6.0983,
      "step": 1150
    },
    {
      "epoch": 0.15762804711038073,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005175,
      "loss": 6.062,
      "step": 1151
    },
    {
      "epoch": 0.15776499589153656,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00051795,
      "loss": 6.0715,
      "step": 1152
    },
    {
      "epoch": 0.15790194467269242,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005183999999999999,
      "loss": 6.135,
      "step": 1153
    },
    {
      "epoch": 0.15803889345384825,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00051885,
      "loss": 6.1322,
      "step": 1154
    },
    {
      "epoch": 0.1581758422350041,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005193,
      "loss": 6.0599,
      "step": 1155
    },
    {
      "epoch": 0.15831279101615994,
      "grad_norm": 0.828125,
      "learning_rate": 0.00051975,
      "loss": 6.0751,
      "step": 1156
    },
    {
      "epoch": 0.1584497397973158,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005202,
      "loss": 6.0287,
      "step": 1157
    },
    {
      "epoch": 0.15858668857847166,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00052065,
      "loss": 6.1027,
      "step": 1158
    },
    {
      "epoch": 0.1587236373596275,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005210999999999999,
      "loss": 6.0389,
      "step": 1159
    },
    {
      "epoch": 0.15886058614078336,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00052155,
      "loss": 6.0863,
      "step": 1160
    },
    {
      "epoch": 0.1589975349219392,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000522,
      "loss": 6.0925,
      "step": 1161
    },
    {
      "epoch": 0.15913448370309505,
      "grad_norm": 0.875,
      "learning_rate": 0.00052245,
      "loss": 5.9879,
      "step": 1162
    },
    {
      "epoch": 0.15927143248425088,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005229,
      "loss": 6.0075,
      "step": 1163
    },
    {
      "epoch": 0.15940838126540674,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00052335,
      "loss": 6.1175,
      "step": 1164
    },
    {
      "epoch": 0.15954533004656257,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005237999999999999,
      "loss": 5.9755,
      "step": 1165
    },
    {
      "epoch": 0.15968227882771843,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00052425,
      "loss": 6.0959,
      "step": 1166
    },
    {
      "epoch": 0.1598192276088743,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005246999999999999,
      "loss": 6.0419,
      "step": 1167
    },
    {
      "epoch": 0.15995617639003012,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00052515,
      "loss": 6.0739,
      "step": 1168
    },
    {
      "epoch": 0.16009312517118598,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005256,
      "loss": 6.1051,
      "step": 1169
    },
    {
      "epoch": 0.16023007395234182,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00052605,
      "loss": 6.0175,
      "step": 1170
    },
    {
      "epoch": 0.16036702273349768,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005265,
      "loss": 6.0282,
      "step": 1171
    },
    {
      "epoch": 0.1605039715146535,
      "grad_norm": 1.09375,
      "learning_rate": 0.00052695,
      "loss": 6.1528,
      "step": 1172
    },
    {
      "epoch": 0.16064092029580937,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005273999999999999,
      "loss": 6.029,
      "step": 1173
    },
    {
      "epoch": 0.1607778690769652,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00052785,
      "loss": 6.045,
      "step": 1174
    },
    {
      "epoch": 0.16091481785812106,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005283,
      "loss": 6.0479,
      "step": 1175
    },
    {
      "epoch": 0.16105176663927692,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00052875,
      "loss": 6.064,
      "step": 1176
    },
    {
      "epoch": 0.16118871542043275,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005292,
      "loss": 6.048,
      "step": 1177
    },
    {
      "epoch": 0.16132566420158861,
      "grad_norm": 1.3125,
      "learning_rate": 0.00052965,
      "loss": 6.0686,
      "step": 1178
    },
    {
      "epoch": 0.16146261298274445,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005300999999999999,
      "loss": 6.1258,
      "step": 1179
    },
    {
      "epoch": 0.1615995617639003,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00053055,
      "loss": 6.1173,
      "step": 1180
    },
    {
      "epoch": 0.16173651054505614,
      "grad_norm": 1.265625,
      "learning_rate": 0.000531,
      "loss": 6.0828,
      "step": 1181
    },
    {
      "epoch": 0.161873459326212,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00053145,
      "loss": 6.0403,
      "step": 1182
    },
    {
      "epoch": 0.16201040810736783,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005319,
      "loss": 5.9873,
      "step": 1183
    },
    {
      "epoch": 0.1621473568885237,
      "grad_norm": 1.25,
      "learning_rate": 0.00053235,
      "loss": 6.0915,
      "step": 1184
    },
    {
      "epoch": 0.16228430566967955,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005327999999999999,
      "loss": 6.059,
      "step": 1185
    },
    {
      "epoch": 0.16242125445083538,
      "grad_norm": 0.90625,
      "learning_rate": 0.00053325,
      "loss": 6.1228,
      "step": 1186
    },
    {
      "epoch": 0.16255820323199124,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005336999999999999,
      "loss": 6.0493,
      "step": 1187
    },
    {
      "epoch": 0.16269515201314708,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005341500000000001,
      "loss": 6.0072,
      "step": 1188
    },
    {
      "epoch": 0.16283210079430294,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005346,
      "loss": 6.1017,
      "step": 1189
    },
    {
      "epoch": 0.16296904957545877,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00053505,
      "loss": 6.0239,
      "step": 1190
    },
    {
      "epoch": 0.16310599835661463,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005355,
      "loss": 6.0483,
      "step": 1191
    },
    {
      "epoch": 0.16324294713777046,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00053595,
      "loss": 6.0948,
      "step": 1192
    },
    {
      "epoch": 0.16337989591892632,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005363999999999999,
      "loss": 6.0965,
      "step": 1193
    },
    {
      "epoch": 0.16351684470008218,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005368500000000001,
      "loss": 6.037,
      "step": 1194
    },
    {
      "epoch": 0.163653793481238,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005373,
      "loss": 6.0741,
      "step": 1195
    },
    {
      "epoch": 0.16379074226239387,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00053775,
      "loss": 6.0487,
      "step": 1196
    },
    {
      "epoch": 0.1639276910435497,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0005382,
      "loss": 6.0423,
      "step": 1197
    },
    {
      "epoch": 0.16406463982470557,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00053865,
      "loss": 6.0873,
      "step": 1198
    },
    {
      "epoch": 0.1642015886058614,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005390999999999999,
      "loss": 6.029,
      "step": 1199
    },
    {
      "epoch": 0.16433853738701726,
      "grad_norm": 0.796875,
      "learning_rate": 0.00053955,
      "loss": 6.0321,
      "step": 1200
    },
    {
      "epoch": 0.1644754861681731,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00054,
      "loss": 5.9961,
      "step": 1201
    },
    {
      "epoch": 0.16461243494932895,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00054045,
      "loss": 6.0688,
      "step": 1202
    },
    {
      "epoch": 0.1647493837304848,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005409,
      "loss": 6.0338,
      "step": 1203
    },
    {
      "epoch": 0.16488633251164064,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00054135,
      "loss": 6.0495,
      "step": 1204
    },
    {
      "epoch": 0.1650232812927965,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005417999999999999,
      "loss": 5.9952,
      "step": 1205
    },
    {
      "epoch": 0.16516023007395234,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00054225,
      "loss": 5.9873,
      "step": 1206
    },
    {
      "epoch": 0.1652971788551082,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005427,
      "loss": 5.9873,
      "step": 1207
    },
    {
      "epoch": 0.16543412763626403,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005431500000000001,
      "loss": 6.0355,
      "step": 1208
    },
    {
      "epoch": 0.1655710764174199,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005436,
      "loss": 6.0134,
      "step": 1209
    },
    {
      "epoch": 0.16570802519857572,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00054405,
      "loss": 6.057,
      "step": 1210
    },
    {
      "epoch": 0.16584497397973158,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005445,
      "loss": 6.0213,
      "step": 1211
    },
    {
      "epoch": 0.16598192276088744,
      "grad_norm": 1.09375,
      "learning_rate": 0.00054495,
      "loss": 5.994,
      "step": 1212
    },
    {
      "epoch": 0.16611887154204327,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005453999999999999,
      "loss": 5.9994,
      "step": 1213
    },
    {
      "epoch": 0.16625582032319913,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005458500000000001,
      "loss": 6.0165,
      "step": 1214
    },
    {
      "epoch": 0.16639276910435497,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005463,
      "loss": 6.0247,
      "step": 1215
    },
    {
      "epoch": 0.16652971788551083,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00054675,
      "loss": 6.0861,
      "step": 1216
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005472,
      "loss": 6.0438,
      "step": 1217
    },
    {
      "epoch": 0.16680361544782252,
      "grad_norm": 0.75,
      "learning_rate": 0.00054765,
      "loss": 6.0039,
      "step": 1218
    },
    {
      "epoch": 0.16694056422897835,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005480999999999999,
      "loss": 6.0095,
      "step": 1219
    },
    {
      "epoch": 0.1670775130101342,
      "grad_norm": 0.859375,
      "learning_rate": 0.00054855,
      "loss": 6.0199,
      "step": 1220
    },
    {
      "epoch": 0.16721446179129007,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000549,
      "loss": 6.0312,
      "step": 1221
    },
    {
      "epoch": 0.1673514105724459,
      "grad_norm": 0.984375,
      "learning_rate": 0.00054945,
      "loss": 5.9382,
      "step": 1222
    },
    {
      "epoch": 0.16748835935360176,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005499,
      "loss": 6.0047,
      "step": 1223
    },
    {
      "epoch": 0.1676253081347576,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00055035,
      "loss": 6.0436,
      "step": 1224
    },
    {
      "epoch": 0.16776225691591345,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005507999999999999,
      "loss": 6.0343,
      "step": 1225
    },
    {
      "epoch": 0.1678992056970693,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00055125,
      "loss": 6.0003,
      "step": 1226
    },
    {
      "epoch": 0.16803615447822515,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005517,
      "loss": 6.0285,
      "step": 1227
    },
    {
      "epoch": 0.16817310325938098,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005521500000000001,
      "loss": 5.9753,
      "step": 1228
    },
    {
      "epoch": 0.16831005204053684,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005526,
      "loss": 6.0497,
      "step": 1229
    },
    {
      "epoch": 0.1684470008216927,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00055305,
      "loss": 6.014,
      "step": 1230
    },
    {
      "epoch": 0.16858394960284853,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005535,
      "loss": 6.0294,
      "step": 1231
    },
    {
      "epoch": 0.1687208983840044,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00055395,
      "loss": 6.0153,
      "step": 1232
    },
    {
      "epoch": 0.16885784716516022,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0005544,
      "loss": 5.9678,
      "step": 1233
    },
    {
      "epoch": 0.16899479594631608,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005548500000000001,
      "loss": 5.9757,
      "step": 1234
    },
    {
      "epoch": 0.16913174472747192,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005553,
      "loss": 5.9634,
      "step": 1235
    },
    {
      "epoch": 0.16926869350862778,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00055575,
      "loss": 5.9669,
      "step": 1236
    },
    {
      "epoch": 0.1694056422897836,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0005562,
      "loss": 5.9713,
      "step": 1237
    },
    {
      "epoch": 0.16954259107093947,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00055665,
      "loss": 5.9687,
      "step": 1238
    },
    {
      "epoch": 0.16967953985209533,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005570999999999999,
      "loss": 6.0171,
      "step": 1239
    },
    {
      "epoch": 0.16981648863325116,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005575500000000001,
      "loss": 6.0472,
      "step": 1240
    },
    {
      "epoch": 0.16995343741440702,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000558,
      "loss": 5.9517,
      "step": 1241
    },
    {
      "epoch": 0.17009038619556285,
      "grad_norm": 0.890625,
      "learning_rate": 0.00055845,
      "loss": 6.0084,
      "step": 1242
    },
    {
      "epoch": 0.1702273349767187,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005589,
      "loss": 6.0163,
      "step": 1243
    },
    {
      "epoch": 0.17036428375787455,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00055935,
      "loss": 5.8878,
      "step": 1244
    },
    {
      "epoch": 0.1705012325390304,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005598,
      "loss": 6.0191,
      "step": 1245
    },
    {
      "epoch": 0.17063818132018624,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00056025,
      "loss": 5.9253,
      "step": 1246
    },
    {
      "epoch": 0.1707751301013421,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005607,
      "loss": 5.9969,
      "step": 1247
    },
    {
      "epoch": 0.17091207888249796,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005611500000000001,
      "loss": 5.9739,
      "step": 1248
    },
    {
      "epoch": 0.1710490276636538,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005616,
      "loss": 5.9901,
      "step": 1249
    },
    {
      "epoch": 0.17118597644480965,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00056205,
      "loss": 5.978,
      "step": 1250
    },
    {
      "epoch": 0.17132292522596548,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005625,
      "loss": 5.9966,
      "step": 1251
    },
    {
      "epoch": 0.17145987400712134,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005629499999999999,
      "loss": 5.9872,
      "step": 1252
    },
    {
      "epoch": 0.17159682278827718,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005634,
      "loss": 6.005,
      "step": 1253
    },
    {
      "epoch": 0.17173377156943304,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00056385,
      "loss": 5.999,
      "step": 1254
    },
    {
      "epoch": 0.17187072035058887,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005643,
      "loss": 6.0295,
      "step": 1255
    },
    {
      "epoch": 0.17200766913174473,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005647499999999999,
      "loss": 6.0312,
      "step": 1256
    },
    {
      "epoch": 0.1721446179129006,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005652,
      "loss": 6.0025,
      "step": 1257
    },
    {
      "epoch": 0.17228156669405642,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005656499999999999,
      "loss": 6.0452,
      "step": 1258
    },
    {
      "epoch": 0.17241851547521228,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005660999999999999,
      "loss": 6.0112,
      "step": 1259
    },
    {
      "epoch": 0.1725554642563681,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00056655,
      "loss": 5.9983,
      "step": 1260
    },
    {
      "epoch": 0.17269241303752397,
      "grad_norm": 0.984375,
      "learning_rate": 0.000567,
      "loss": 6.0583,
      "step": 1261
    },
    {
      "epoch": 0.1728293618186798,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005674499999999999,
      "loss": 5.9882,
      "step": 1262
    },
    {
      "epoch": 0.17296631059983567,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005679,
      "loss": 6.0243,
      "step": 1263
    },
    {
      "epoch": 0.1731032593809915,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005683499999999999,
      "loss": 5.9865,
      "step": 1264
    },
    {
      "epoch": 0.17324020816214736,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005688,
      "loss": 5.975,
      "step": 1265
    },
    {
      "epoch": 0.17337715694330322,
      "grad_norm": 1.359375,
      "learning_rate": 0.00056925,
      "loss": 6.0364,
      "step": 1266
    },
    {
      "epoch": 0.17351410572445905,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005697,
      "loss": 5.9803,
      "step": 1267
    },
    {
      "epoch": 0.1736510545056149,
      "grad_norm": 0.734375,
      "learning_rate": 0.00057015,
      "loss": 6.0264,
      "step": 1268
    },
    {
      "epoch": 0.17378800328677074,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005706,
      "loss": 5.9849,
      "step": 1269
    },
    {
      "epoch": 0.1739249520679266,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0005710499999999999,
      "loss": 5.924,
      "step": 1270
    },
    {
      "epoch": 0.17406190084908243,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005715,
      "loss": 5.9658,
      "step": 1271
    },
    {
      "epoch": 0.1741988496302383,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005719499999999999,
      "loss": 5.9743,
      "step": 1272
    },
    {
      "epoch": 0.17433579841139413,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005724,
      "loss": 5.9973,
      "step": 1273
    },
    {
      "epoch": 0.17447274719255,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00057285,
      "loss": 5.9996,
      "step": 1274
    },
    {
      "epoch": 0.17460969597370585,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005733,
      "loss": 5.9761,
      "step": 1275
    },
    {
      "epoch": 0.17474664475486168,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005737499999999999,
      "loss": 5.99,
      "step": 1276
    },
    {
      "epoch": 0.17488359353601754,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005742,
      "loss": 5.9406,
      "step": 1277
    },
    {
      "epoch": 0.17502054231717337,
      "grad_norm": 0.65625,
      "learning_rate": 0.0005746499999999999,
      "loss": 5.9606,
      "step": 1278
    },
    {
      "epoch": 0.17515749109832923,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005751,
      "loss": 5.9087,
      "step": 1279
    },
    {
      "epoch": 0.17529443987948506,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00057555,
      "loss": 5.947,
      "step": 1280
    },
    {
      "epoch": 0.17543138866064092,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000576,
      "loss": 5.9651,
      "step": 1281
    },
    {
      "epoch": 0.17556833744179676,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005764499999999999,
      "loss": 5.9552,
      "step": 1282
    },
    {
      "epoch": 0.17570528622295262,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005769,
      "loss": 5.9834,
      "step": 1283
    },
    {
      "epoch": 0.17584223500410848,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005773499999999999,
      "loss": 5.9616,
      "step": 1284
    },
    {
      "epoch": 0.1759791837852643,
      "grad_norm": 1.0,
      "learning_rate": 0.0005778,
      "loss": 5.9672,
      "step": 1285
    },
    {
      "epoch": 0.17611613256642017,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00057825,
      "loss": 5.9003,
      "step": 1286
    },
    {
      "epoch": 0.176253081347576,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005787,
      "loss": 6.0206,
      "step": 1287
    },
    {
      "epoch": 0.17639003012873186,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00057915,
      "loss": 5.9511,
      "step": 1288
    },
    {
      "epoch": 0.1765269789098877,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005796,
      "loss": 5.985,
      "step": 1289
    },
    {
      "epoch": 0.17666392769104355,
      "grad_norm": 0.75,
      "learning_rate": 0.0005800499999999999,
      "loss": 5.8748,
      "step": 1290
    },
    {
      "epoch": 0.1768008764721994,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005805,
      "loss": 5.8984,
      "step": 1291
    },
    {
      "epoch": 0.17693782525335525,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00058095,
      "loss": 5.9826,
      "step": 1292
    },
    {
      "epoch": 0.1770747740345111,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005814,
      "loss": 5.9718,
      "step": 1293
    },
    {
      "epoch": 0.17721172281566694,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00058185,
      "loss": 5.964,
      "step": 1294
    },
    {
      "epoch": 0.1773486715968228,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005823,
      "loss": 5.9382,
      "step": 1295
    },
    {
      "epoch": 0.17748562037797863,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005827499999999999,
      "loss": 5.9143,
      "step": 1296
    },
    {
      "epoch": 0.1776225691591345,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005832,
      "loss": 5.9841,
      "step": 1297
    },
    {
      "epoch": 0.17775951794029032,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005836499999999999,
      "loss": 5.9313,
      "step": 1298
    },
    {
      "epoch": 0.17789646672144618,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005841,
      "loss": 5.9836,
      "step": 1299
    },
    {
      "epoch": 0.17803341550260202,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00058455,
      "loss": 5.9492,
      "step": 1300
    },
    {
      "epoch": 0.17817036428375788,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000585,
      "loss": 5.9477,
      "step": 1301
    },
    {
      "epoch": 0.17830731306491374,
      "grad_norm": 0.953125,
      "learning_rate": 0.00058545,
      "loss": 5.9554,
      "step": 1302
    },
    {
      "epoch": 0.17844426184606957,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005859,
      "loss": 5.934,
      "step": 1303
    },
    {
      "epoch": 0.17858121062722543,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005863499999999999,
      "loss": 5.9699,
      "step": 1304
    },
    {
      "epoch": 0.17871815940838126,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005868,
      "loss": 5.9027,
      "step": 1305
    },
    {
      "epoch": 0.17885510818953712,
      "grad_norm": 0.96875,
      "learning_rate": 0.00058725,
      "loss": 5.8777,
      "step": 1306
    },
    {
      "epoch": 0.17899205697069295,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005877,
      "loss": 5.9285,
      "step": 1307
    },
    {
      "epoch": 0.1791290057518488,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00058815,
      "loss": 5.9769,
      "step": 1308
    },
    {
      "epoch": 0.17926595453300465,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005886,
      "loss": 5.8952,
      "step": 1309
    },
    {
      "epoch": 0.1794029033141605,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005890499999999999,
      "loss": 5.9579,
      "step": 1310
    },
    {
      "epoch": 0.17953985209531634,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005895,
      "loss": 5.9022,
      "step": 1311
    },
    {
      "epoch": 0.1796768008764722,
      "grad_norm": 0.828125,
      "learning_rate": 0.00058995,
      "loss": 5.9099,
      "step": 1312
    },
    {
      "epoch": 0.17981374965762806,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005904,
      "loss": 5.8932,
      "step": 1313
    },
    {
      "epoch": 0.1799506984387839,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00059085,
      "loss": 5.9212,
      "step": 1314
    },
    {
      "epoch": 0.18008764721993975,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005913,
      "loss": 5.9626,
      "step": 1315
    },
    {
      "epoch": 0.18022459600109558,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005917499999999999,
      "loss": 5.9256,
      "step": 1316
    },
    {
      "epoch": 0.18036154478225144,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005922,
      "loss": 5.9157,
      "step": 1317
    },
    {
      "epoch": 0.18049849356340728,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00059265,
      "loss": 5.953,
      "step": 1318
    },
    {
      "epoch": 0.18063544234456314,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005931,
      "loss": 5.8785,
      "step": 1319
    },
    {
      "epoch": 0.18077239112571897,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00059355,
      "loss": 5.8751,
      "step": 1320
    },
    {
      "epoch": 0.18090933990687483,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000594,
      "loss": 5.9181,
      "step": 1321
    },
    {
      "epoch": 0.1810462886880307,
      "grad_norm": 1.125,
      "learning_rate": 0.00059445,
      "loss": 5.9266,
      "step": 1322
    },
    {
      "epoch": 0.18118323746918652,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005949,
      "loss": 5.8838,
      "step": 1323
    },
    {
      "epoch": 0.18132018625034238,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005953499999999999,
      "loss": 5.9269,
      "step": 1324
    },
    {
      "epoch": 0.1814571350314982,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005958000000000001,
      "loss": 5.954,
      "step": 1325
    },
    {
      "epoch": 0.18159408381265407,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00059625,
      "loss": 5.7993,
      "step": 1326
    },
    {
      "epoch": 0.1817310325938099,
      "grad_norm": 0.71875,
      "learning_rate": 0.0005967,
      "loss": 5.925,
      "step": 1327
    },
    {
      "epoch": 0.18186798137496576,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00059715,
      "loss": 5.8582,
      "step": 1328
    },
    {
      "epoch": 0.1820049301561216,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005976,
      "loss": 5.8871,
      "step": 1329
    },
    {
      "epoch": 0.18214187893727746,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005980499999999999,
      "loss": 5.8807,
      "step": 1330
    },
    {
      "epoch": 0.18227882771843332,
      "grad_norm": 0.75,
      "learning_rate": 0.0005985,
      "loss": 5.8974,
      "step": 1331
    },
    {
      "epoch": 0.18241577649958915,
      "grad_norm": 0.875,
      "learning_rate": 0.00059895,
      "loss": 5.8244,
      "step": 1332
    },
    {
      "epoch": 0.182552725280745,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005994,
      "loss": 5.9109,
      "step": 1333
    },
    {
      "epoch": 0.18268967406190084,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00059985,
      "loss": 5.8962,
      "step": 1334
    },
    {
      "epoch": 0.1828266228430567,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006003,
      "loss": 5.8731,
      "step": 1335
    },
    {
      "epoch": 0.18296357162421253,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006007499999999999,
      "loss": 5.8485,
      "step": 1336
    },
    {
      "epoch": 0.1831005204053684,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0006012,
      "loss": 5.9041,
      "step": 1337
    },
    {
      "epoch": 0.18323746918652423,
      "grad_norm": 0.78125,
      "learning_rate": 0.00060165,
      "loss": 5.8581,
      "step": 1338
    },
    {
      "epoch": 0.1833744179676801,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006021,
      "loss": 5.8464,
      "step": 1339
    },
    {
      "epoch": 0.18351136674883595,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00060255,
      "loss": 5.8867,
      "step": 1340
    },
    {
      "epoch": 0.18364831552999178,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000603,
      "loss": 5.9444,
      "step": 1341
    },
    {
      "epoch": 0.18378526431114764,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00060345,
      "loss": 5.8571,
      "step": 1342
    },
    {
      "epoch": 0.18392221309230347,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006039,
      "loss": 5.8641,
      "step": 1343
    },
    {
      "epoch": 0.18405916187345933,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006043499999999999,
      "loss": 5.8329,
      "step": 1344
    },
    {
      "epoch": 0.18419611065461516,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006048000000000001,
      "loss": 5.9109,
      "step": 1345
    },
    {
      "epoch": 0.18433305943577102,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00060525,
      "loss": 5.8913,
      "step": 1346
    },
    {
      "epoch": 0.18447000821692686,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006057,
      "loss": 5.8934,
      "step": 1347
    },
    {
      "epoch": 0.18460695699808272,
      "grad_norm": 0.953125,
      "learning_rate": 0.00060615,
      "loss": 5.8805,
      "step": 1348
    },
    {
      "epoch": 0.18474390577923858,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006066,
      "loss": 5.9202,
      "step": 1349
    },
    {
      "epoch": 0.1848808545603944,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006070499999999999,
      "loss": 5.8951,
      "step": 1350
    },
    {
      "epoch": 0.18501780334155027,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006075000000000001,
      "loss": 5.8636,
      "step": 1351
    },
    {
      "epoch": 0.1851547521227061,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00060795,
      "loss": 5.8512,
      "step": 1352
    },
    {
      "epoch": 0.18529170090386196,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006084,
      "loss": 5.869,
      "step": 1353
    },
    {
      "epoch": 0.1854286496850178,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00060885,
      "loss": 5.8687,
      "step": 1354
    },
    {
      "epoch": 0.18556559846617365,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006093,
      "loss": 5.8319,
      "step": 1355
    },
    {
      "epoch": 0.18570254724732949,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006097499999999999,
      "loss": 5.9076,
      "step": 1356
    },
    {
      "epoch": 0.18583949602848535,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006102,
      "loss": 5.9028,
      "step": 1357
    },
    {
      "epoch": 0.1859764448096412,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00061065,
      "loss": 5.8749,
      "step": 1358
    },
    {
      "epoch": 0.18611339359079704,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006111000000000001,
      "loss": 5.8542,
      "step": 1359
    },
    {
      "epoch": 0.1862503423719529,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00061155,
      "loss": 5.8261,
      "step": 1360
    },
    {
      "epoch": 0.18638729115310873,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000612,
      "loss": 5.8911,
      "step": 1361
    },
    {
      "epoch": 0.1865242399342646,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00061245,
      "loss": 5.8899,
      "step": 1362
    },
    {
      "epoch": 0.18666118871542042,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006129,
      "loss": 5.8667,
      "step": 1363
    },
    {
      "epoch": 0.18679813749657628,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00061335,
      "loss": 5.9193,
      "step": 1364
    },
    {
      "epoch": 0.18693508627773212,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006138000000000001,
      "loss": 5.8513,
      "step": 1365
    },
    {
      "epoch": 0.18707203505888798,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00061425,
      "loss": 5.8655,
      "step": 1366
    },
    {
      "epoch": 0.18720898384004384,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006147,
      "loss": 5.8783,
      "step": 1367
    },
    {
      "epoch": 0.18734593262119967,
      "grad_norm": 0.953125,
      "learning_rate": 0.00061515,
      "loss": 5.9234,
      "step": 1368
    },
    {
      "epoch": 0.18748288140235553,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006156,
      "loss": 5.8908,
      "step": 1369
    },
    {
      "epoch": 0.18761983018351136,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006160499999999999,
      "loss": 5.9222,
      "step": 1370
    },
    {
      "epoch": 0.18775677896466722,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006165000000000001,
      "loss": 5.8953,
      "step": 1371
    },
    {
      "epoch": 0.18789372774582305,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00061695,
      "loss": 5.8902,
      "step": 1372
    },
    {
      "epoch": 0.1880306765269789,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006174,
      "loss": 5.9009,
      "step": 1373
    },
    {
      "epoch": 0.18816762530813474,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00061785,
      "loss": 5.8021,
      "step": 1374
    },
    {
      "epoch": 0.1883045740892906,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006183,
      "loss": 5.8567,
      "step": 1375
    },
    {
      "epoch": 0.18844152287044647,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006187499999999999,
      "loss": 5.8812,
      "step": 1376
    },
    {
      "epoch": 0.1885784716516023,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006192,
      "loss": 5.9065,
      "step": 1377
    },
    {
      "epoch": 0.18871542043275816,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00061965,
      "loss": 5.9411,
      "step": 1378
    },
    {
      "epoch": 0.188852369213914,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006201,
      "loss": 5.8847,
      "step": 1379
    },
    {
      "epoch": 0.18898931799506985,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00062055,
      "loss": 5.8657,
      "step": 1380
    },
    {
      "epoch": 0.18912626677622568,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006209999999999999,
      "loss": 5.9208,
      "step": 1381
    },
    {
      "epoch": 0.18926321555738154,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00062145,
      "loss": 5.9003,
      "step": 1382
    },
    {
      "epoch": 0.18940016433853737,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006218999999999999,
      "loss": 5.8442,
      "step": 1383
    },
    {
      "epoch": 0.18953711311969323,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00062235,
      "loss": 5.8742,
      "step": 1384
    },
    {
      "epoch": 0.1896740619008491,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006228,
      "loss": 5.8445,
      "step": 1385
    },
    {
      "epoch": 0.18981101068200493,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00062325,
      "loss": 5.8113,
      "step": 1386
    },
    {
      "epoch": 0.1899479594631608,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006236999999999999,
      "loss": 5.8059,
      "step": 1387
    },
    {
      "epoch": 0.19008490824431662,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00062415,
      "loss": 5.8478,
      "step": 1388
    },
    {
      "epoch": 0.19022185702547248,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006245999999999999,
      "loss": 5.8388,
      "step": 1389
    },
    {
      "epoch": 0.1903588058066283,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00062505,
      "loss": 5.7698,
      "step": 1390
    },
    {
      "epoch": 0.19049575458778417,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006255,
      "loss": 5.8729,
      "step": 1391
    },
    {
      "epoch": 0.19063270336894,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00062595,
      "loss": 5.8144,
      "step": 1392
    },
    {
      "epoch": 0.19076965215009586,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0006263999999999999,
      "loss": 5.8307,
      "step": 1393
    },
    {
      "epoch": 0.19090660093125172,
      "grad_norm": 0.75,
      "learning_rate": 0.00062685,
      "loss": 5.8869,
      "step": 1394
    },
    {
      "epoch": 0.19104354971240756,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006272999999999999,
      "loss": 5.8875,
      "step": 1395
    },
    {
      "epoch": 0.19118049849356342,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006277499999999999,
      "loss": 5.8344,
      "step": 1396
    },
    {
      "epoch": 0.19131744727471925,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006282,
      "loss": 5.8011,
      "step": 1397
    },
    {
      "epoch": 0.1914543960558751,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00062865,
      "loss": 5.8637,
      "step": 1398
    },
    {
      "epoch": 0.19159134483703094,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006291,
      "loss": 5.8312,
      "step": 1399
    },
    {
      "epoch": 0.1917282936181868,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00062955,
      "loss": 5.8677,
      "step": 1400
    },
    {
      "epoch": 0.19186524239934263,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006299999999999999,
      "loss": 5.852,
      "step": 1401
    },
    {
      "epoch": 0.1920021911804985,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00063045,
      "loss": 5.8318,
      "step": 1402
    },
    {
      "epoch": 0.19213913996165435,
      "grad_norm": 1.578125,
      "learning_rate": 0.0006309,
      "loss": 5.7608,
      "step": 1403
    },
    {
      "epoch": 0.1922760887428102,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00063135,
      "loss": 5.8808,
      "step": 1404
    },
    {
      "epoch": 0.19241303752396605,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006318,
      "loss": 5.7963,
      "step": 1405
    },
    {
      "epoch": 0.19254998630512188,
      "grad_norm": 1.203125,
      "learning_rate": 0.00063225,
      "loss": 5.8205,
      "step": 1406
    },
    {
      "epoch": 0.19268693508627774,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006326999999999999,
      "loss": 5.8991,
      "step": 1407
    },
    {
      "epoch": 0.19282388386743357,
      "grad_norm": 1.421875,
      "learning_rate": 0.00063315,
      "loss": 5.8193,
      "step": 1408
    },
    {
      "epoch": 0.19296083264858943,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006335999999999999,
      "loss": 5.8313,
      "step": 1409
    },
    {
      "epoch": 0.19309778142974526,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00063405,
      "loss": 5.8173,
      "step": 1410
    },
    {
      "epoch": 0.19323473021090112,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006345,
      "loss": 5.8907,
      "step": 1411
    },
    {
      "epoch": 0.19337167899205698,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00063495,
      "loss": 5.8963,
      "step": 1412
    },
    {
      "epoch": 0.19350862777321282,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006353999999999999,
      "loss": 5.8769,
      "step": 1413
    },
    {
      "epoch": 0.19364557655436868,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00063585,
      "loss": 5.896,
      "step": 1414
    },
    {
      "epoch": 0.1937825253355245,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006362999999999999,
      "loss": 5.8934,
      "step": 1415
    },
    {
      "epoch": 0.19391947411668037,
      "grad_norm": 1.125,
      "learning_rate": 0.00063675,
      "loss": 5.8178,
      "step": 1416
    },
    {
      "epoch": 0.1940564228978362,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006372,
      "loss": 5.862,
      "step": 1417
    },
    {
      "epoch": 0.19419337167899206,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00063765,
      "loss": 5.8449,
      "step": 1418
    },
    {
      "epoch": 0.1943303204601479,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0006381,
      "loss": 5.7763,
      "step": 1419
    },
    {
      "epoch": 0.19446726924130375,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00063855,
      "loss": 5.8284,
      "step": 1420
    },
    {
      "epoch": 0.1946042180224596,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006389999999999999,
      "loss": 5.8525,
      "step": 1421
    },
    {
      "epoch": 0.19474116680361545,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00063945,
      "loss": 5.8541,
      "step": 1422
    },
    {
      "epoch": 0.1948781155847713,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006399,
      "loss": 5.8152,
      "step": 1423
    },
    {
      "epoch": 0.19501506436592714,
      "grad_norm": 0.796875,
      "learning_rate": 0.00064035,
      "loss": 5.8134,
      "step": 1424
    },
    {
      "epoch": 0.195152013147083,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006408,
      "loss": 5.8765,
      "step": 1425
    },
    {
      "epoch": 0.19528896192823883,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00064125,
      "loss": 5.8632,
      "step": 1426
    },
    {
      "epoch": 0.1954259107093947,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006416999999999999,
      "loss": 5.8751,
      "step": 1427
    },
    {
      "epoch": 0.19556285949055052,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00064215,
      "loss": 5.8302,
      "step": 1428
    },
    {
      "epoch": 0.19569980827170638,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0006426,
      "loss": 5.7995,
      "step": 1429
    },
    {
      "epoch": 0.19583675705286224,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00064305,
      "loss": 5.8056,
      "step": 1430
    },
    {
      "epoch": 0.19597370583401807,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006435,
      "loss": 5.8145,
      "step": 1431
    },
    {
      "epoch": 0.19611065461517393,
      "grad_norm": 0.703125,
      "learning_rate": 0.00064395,
      "loss": 5.7601,
      "step": 1432
    },
    {
      "epoch": 0.19624760339632977,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006443999999999999,
      "loss": 5.8215,
      "step": 1433
    },
    {
      "epoch": 0.19638455217748563,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00064485,
      "loss": 5.7444,
      "step": 1434
    },
    {
      "epoch": 0.19652150095864146,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006452999999999999,
      "loss": 5.7691,
      "step": 1435
    },
    {
      "epoch": 0.19665844973979732,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006457500000000001,
      "loss": 5.8017,
      "step": 1436
    },
    {
      "epoch": 0.19679539852095315,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006462,
      "loss": 5.8421,
      "step": 1437
    },
    {
      "epoch": 0.196932347302109,
      "grad_norm": 0.609375,
      "learning_rate": 0.00064665,
      "loss": 5.7697,
      "step": 1438
    },
    {
      "epoch": 0.19706929608326487,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0006471,
      "loss": 5.7823,
      "step": 1439
    },
    {
      "epoch": 0.1972062448644207,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00064755,
      "loss": 5.7902,
      "step": 1440
    },
    {
      "epoch": 0.19734319364557656,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006479999999999999,
      "loss": 5.9156,
      "step": 1441
    },
    {
      "epoch": 0.1974801424267324,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00064845,
      "loss": 5.8252,
      "step": 1442
    },
    {
      "epoch": 0.19761709120788826,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0006489,
      "loss": 5.8301,
      "step": 1443
    },
    {
      "epoch": 0.1977540399890441,
      "grad_norm": 0.625,
      "learning_rate": 0.00064935,
      "loss": 5.7742,
      "step": 1444
    },
    {
      "epoch": 0.19789098877019995,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006498,
      "loss": 5.717,
      "step": 1445
    },
    {
      "epoch": 0.19802793755135578,
      "grad_norm": 0.625,
      "learning_rate": 0.00065025,
      "loss": 5.8223,
      "step": 1446
    },
    {
      "epoch": 0.19816488633251164,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006506999999999999,
      "loss": 5.789,
      "step": 1447
    },
    {
      "epoch": 0.1983018351136675,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00065115,
      "loss": 5.7842,
      "step": 1448
    },
    {
      "epoch": 0.19843878389482333,
      "grad_norm": 0.75,
      "learning_rate": 0.0006516,
      "loss": 5.7952,
      "step": 1449
    },
    {
      "epoch": 0.1985757326759792,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00065205,
      "loss": 5.7865,
      "step": 1450
    },
    {
      "epoch": 0.19871268145713503,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006525,
      "loss": 5.8187,
      "step": 1451
    },
    {
      "epoch": 0.1988496302382909,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00065295,
      "loss": 5.7894,
      "step": 1452
    },
    {
      "epoch": 0.19898657901944672,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006533999999999999,
      "loss": 5.7491,
      "step": 1453
    },
    {
      "epoch": 0.19912352780060258,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00065385,
      "loss": 5.7822,
      "step": 1454
    },
    {
      "epoch": 0.1992604765817584,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006542999999999999,
      "loss": 5.802,
      "step": 1455
    },
    {
      "epoch": 0.19939742536291427,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006547500000000001,
      "loss": 5.7911,
      "step": 1456
    },
    {
      "epoch": 0.19953437414407013,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006552,
      "loss": 5.8551,
      "step": 1457
    },
    {
      "epoch": 0.19967132292522596,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00065565,
      "loss": 5.8519,
      "step": 1458
    },
    {
      "epoch": 0.19980827170638182,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006561,
      "loss": 5.7962,
      "step": 1459
    },
    {
      "epoch": 0.19994522048753766,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00065655,
      "loss": 5.7866,
      "step": 1460
    },
    {
      "epoch": 0.20008216926869352,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006569999999999999,
      "loss": 5.7977,
      "step": 1461
    },
    {
      "epoch": 0.20021911804984935,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006574500000000001,
      "loss": 5.8325,
      "step": 1462
    },
    {
      "epoch": 0.2003560668310052,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006579,
      "loss": 5.8052,
      "step": 1463
    },
    {
      "epoch": 0.20049301561216104,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00065835,
      "loss": 5.7224,
      "step": 1464
    },
    {
      "epoch": 0.2006299643933169,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0006588,
      "loss": 5.8522,
      "step": 1465
    },
    {
      "epoch": 0.20076691317447276,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00065925,
      "loss": 5.7928,
      "step": 1466
    },
    {
      "epoch": 0.2009038619556286,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0006596999999999999,
      "loss": 5.7702,
      "step": 1467
    },
    {
      "epoch": 0.20104081073678445,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00066015,
      "loss": 5.7679,
      "step": 1468
    },
    {
      "epoch": 0.20117775951794029,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006606,
      "loss": 5.7907,
      "step": 1469
    },
    {
      "epoch": 0.20131470829909615,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00066105,
      "loss": 5.8216,
      "step": 1470
    },
    {
      "epoch": 0.20145165708025198,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006615,
      "loss": 5.7319,
      "step": 1471
    },
    {
      "epoch": 0.20158860586140784,
      "grad_norm": 0.75,
      "learning_rate": 0.00066195,
      "loss": 5.8125,
      "step": 1472
    },
    {
      "epoch": 0.20172555464256367,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006624,
      "loss": 5.8081,
      "step": 1473
    },
    {
      "epoch": 0.20186250342371953,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00066285,
      "loss": 5.7967,
      "step": 1474
    },
    {
      "epoch": 0.2019994522048754,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006633,
      "loss": 5.7898,
      "step": 1475
    },
    {
      "epoch": 0.20213640098603122,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0006637500000000001,
      "loss": 5.7544,
      "step": 1476
    },
    {
      "epoch": 0.20227334976718708,
      "grad_norm": 0.765625,
      "learning_rate": 0.0006642,
      "loss": 5.7488,
      "step": 1477
    },
    {
      "epoch": 0.20241029854834292,
      "grad_norm": 0.734375,
      "learning_rate": 0.00066465,
      "loss": 5.7685,
      "step": 1478
    },
    {
      "epoch": 0.20254724732949878,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006651,
      "loss": 5.7954,
      "step": 1479
    },
    {
      "epoch": 0.2026841961106546,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00066555,
      "loss": 5.7669,
      "step": 1480
    },
    {
      "epoch": 0.20282114489181047,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006659999999999999,
      "loss": 5.8169,
      "step": 1481
    },
    {
      "epoch": 0.2029580936729663,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006664500000000001,
      "loss": 5.774,
      "step": 1482
    },
    {
      "epoch": 0.20309504245412216,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006669,
      "loss": 5.816,
      "step": 1483
    },
    {
      "epoch": 0.203231991235278,
      "grad_norm": 0.65625,
      "learning_rate": 0.00066735,
      "loss": 5.7578,
      "step": 1484
    },
    {
      "epoch": 0.20336894001643385,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0006678,
      "loss": 5.7089,
      "step": 1485
    },
    {
      "epoch": 0.2035058887975897,
      "grad_norm": 0.609375,
      "learning_rate": 0.00066825,
      "loss": 5.8113,
      "step": 1486
    },
    {
      "epoch": 0.20364283757874554,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006686999999999999,
      "loss": 5.7677,
      "step": 1487
    },
    {
      "epoch": 0.2037797863599014,
      "grad_norm": 0.703125,
      "learning_rate": 0.00066915,
      "loss": 5.7855,
      "step": 1488
    },
    {
      "epoch": 0.20391673514105724,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0006696,
      "loss": 5.8229,
      "step": 1489
    },
    {
      "epoch": 0.2040536839222131,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00067005,
      "loss": 5.6973,
      "step": 1490
    },
    {
      "epoch": 0.20419063270336893,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006705,
      "loss": 5.8143,
      "step": 1491
    },
    {
      "epoch": 0.2043275814845248,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00067095,
      "loss": 5.8015,
      "step": 1492
    },
    {
      "epoch": 0.20446453026568062,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006714,
      "loss": 5.806,
      "step": 1493
    },
    {
      "epoch": 0.20460147904683648,
      "grad_norm": 0.75,
      "learning_rate": 0.00067185,
      "loss": 5.737,
      "step": 1494
    },
    {
      "epoch": 0.20473842782799234,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006723,
      "loss": 5.773,
      "step": 1495
    },
    {
      "epoch": 0.20487537660914817,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006727500000000001,
      "loss": 5.7287,
      "step": 1496
    },
    {
      "epoch": 0.20501232539030403,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006732,
      "loss": 5.7773,
      "step": 1497
    },
    {
      "epoch": 0.20514927417145987,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00067365,
      "loss": 5.7676,
      "step": 1498
    },
    {
      "epoch": 0.20528622295261573,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006741,
      "loss": 5.7483,
      "step": 1499
    },
    {
      "epoch": 0.20542317173377156,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00067455,
      "loss": 5.7639,
      "step": 1500
    },
    {
      "epoch": 0.20556012051492742,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000675,
      "loss": 5.743,
      "step": 1501
    },
    {
      "epoch": 0.20569706929608325,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00067545,
      "loss": 5.7515,
      "step": 1502
    },
    {
      "epoch": 0.2058340180772391,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006759,
      "loss": 5.7305,
      "step": 1503
    },
    {
      "epoch": 0.20597096685839497,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006763499999999999,
      "loss": 5.7697,
      "step": 1504
    },
    {
      "epoch": 0.2061079156395508,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006768,
      "loss": 5.781,
      "step": 1505
    },
    {
      "epoch": 0.20624486442070666,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006772499999999999,
      "loss": 5.696,
      "step": 1506
    },
    {
      "epoch": 0.2063818132018625,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006776999999999999,
      "loss": 5.7619,
      "step": 1507
    },
    {
      "epoch": 0.20651876198301836,
      "grad_norm": 0.609375,
      "learning_rate": 0.00067815,
      "loss": 5.7948,
      "step": 1508
    },
    {
      "epoch": 0.2066557107641742,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006786,
      "loss": 5.8173,
      "step": 1509
    },
    {
      "epoch": 0.20679265954533005,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006790499999999999,
      "loss": 5.7242,
      "step": 1510
    },
    {
      "epoch": 0.20692960832648588,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006795,
      "loss": 5.782,
      "step": 1511
    },
    {
      "epoch": 0.20706655710764174,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006799499999999999,
      "loss": 5.7715,
      "step": 1512
    },
    {
      "epoch": 0.2072035058887976,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006804,
      "loss": 5.7472,
      "step": 1513
    },
    {
      "epoch": 0.20734045466995343,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00068085,
      "loss": 5.7338,
      "step": 1514
    },
    {
      "epoch": 0.2074774034511093,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006813,
      "loss": 5.7421,
      "step": 1515
    },
    {
      "epoch": 0.20761435223226513,
      "grad_norm": 0.75,
      "learning_rate": 0.00068175,
      "loss": 5.6874,
      "step": 1516
    },
    {
      "epoch": 0.20775130101342099,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006822,
      "loss": 5.7151,
      "step": 1517
    },
    {
      "epoch": 0.20788824979457682,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0006826499999999999,
      "loss": 5.73,
      "step": 1518
    },
    {
      "epoch": 0.20802519857573268,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006831,
      "loss": 5.7996,
      "step": 1519
    },
    {
      "epoch": 0.2081621473568885,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006835499999999999,
      "loss": 5.7651,
      "step": 1520
    },
    {
      "epoch": 0.20829909613804437,
      "grad_norm": 0.62109375,
      "learning_rate": 0.000684,
      "loss": 5.8195,
      "step": 1521
    },
    {
      "epoch": 0.20843604491920023,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00068445,
      "loss": 5.7514,
      "step": 1522
    },
    {
      "epoch": 0.20857299370035606,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006849,
      "loss": 5.7515,
      "step": 1523
    },
    {
      "epoch": 0.20870994248151192,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006853499999999999,
      "loss": 5.8068,
      "step": 1524
    },
    {
      "epoch": 0.20884689126266776,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006858,
      "loss": 5.735,
      "step": 1525
    },
    {
      "epoch": 0.20898384004382362,
      "grad_norm": 0.65625,
      "learning_rate": 0.0006862499999999999,
      "loss": 5.7796,
      "step": 1526
    },
    {
      "epoch": 0.20912078882497945,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006866999999999999,
      "loss": 5.7786,
      "step": 1527
    },
    {
      "epoch": 0.2092577376061353,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00068715,
      "loss": 5.7185,
      "step": 1528
    },
    {
      "epoch": 0.20939468638729114,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006876,
      "loss": 5.7057,
      "step": 1529
    },
    {
      "epoch": 0.209531635168447,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00068805,
      "loss": 5.7549,
      "step": 1530
    },
    {
      "epoch": 0.20966858394960286,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006885,
      "loss": 5.743,
      "step": 1531
    },
    {
      "epoch": 0.2098055327307587,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006889499999999999,
      "loss": 5.7591,
      "step": 1532
    },
    {
      "epoch": 0.20994248151191455,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006894,
      "loss": 5.7741,
      "step": 1533
    },
    {
      "epoch": 0.21007943029307038,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00068985,
      "loss": 5.6764,
      "step": 1534
    },
    {
      "epoch": 0.21021637907422625,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006903,
      "loss": 5.7513,
      "step": 1535
    },
    {
      "epoch": 0.21035332785538208,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00069075,
      "loss": 5.7447,
      "step": 1536
    },
    {
      "epoch": 0.21049027663653794,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0006912,
      "loss": 5.6961,
      "step": 1537
    },
    {
      "epoch": 0.21062722541769377,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006916499999999999,
      "loss": 5.7024,
      "step": 1538
    },
    {
      "epoch": 0.21076417419884963,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006921,
      "loss": 5.7633,
      "step": 1539
    },
    {
      "epoch": 0.2109011229800055,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006925499999999999,
      "loss": 5.7267,
      "step": 1540
    },
    {
      "epoch": 0.21103807176116132,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000693,
      "loss": 5.7358,
      "step": 1541
    },
    {
      "epoch": 0.21117502054231718,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00069345,
      "loss": 5.7645,
      "step": 1542
    },
    {
      "epoch": 0.21131196932347301,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0006939,
      "loss": 5.6653,
      "step": 1543
    },
    {
      "epoch": 0.21144891810462887,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0006943499999999999,
      "loss": 5.6267,
      "step": 1544
    },
    {
      "epoch": 0.2115858668857847,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006948,
      "loss": 5.7594,
      "step": 1545
    },
    {
      "epoch": 0.21172281566694057,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006952499999999999,
      "loss": 5.6908,
      "step": 1546
    },
    {
      "epoch": 0.2118597644480964,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006957,
      "loss": 5.7637,
      "step": 1547
    },
    {
      "epoch": 0.21199671322925226,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00069615,
      "loss": 5.7592,
      "step": 1548
    },
    {
      "epoch": 0.21213366201040812,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006966,
      "loss": 5.738,
      "step": 1549
    },
    {
      "epoch": 0.21227061079156395,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00069705,
      "loss": 5.707,
      "step": 1550
    },
    {
      "epoch": 0.2124075595727198,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006975,
      "loss": 5.7737,
      "step": 1551
    },
    {
      "epoch": 0.21254450835387564,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0006979499999999999,
      "loss": 5.7899,
      "step": 1552
    },
    {
      "epoch": 0.2126814571350315,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006984,
      "loss": 5.7364,
      "step": 1553
    },
    {
      "epoch": 0.21281840591618734,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00069885,
      "loss": 5.7413,
      "step": 1554
    },
    {
      "epoch": 0.2129553546973432,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006993,
      "loss": 5.7028,
      "step": 1555
    },
    {
      "epoch": 0.21309230347849903,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00069975,
      "loss": 5.7019,
      "step": 1556
    },
    {
      "epoch": 0.2132292522596549,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0007002,
      "loss": 5.743,
      "step": 1557
    },
    {
      "epoch": 0.21336620104081075,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0007006499999999999,
      "loss": 5.7088,
      "step": 1558
    },
    {
      "epoch": 0.21350314982196658,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007011,
      "loss": 5.7406,
      "step": 1559
    },
    {
      "epoch": 0.21364009860312244,
      "grad_norm": 0.828125,
      "learning_rate": 0.00070155,
      "loss": 5.7589,
      "step": 1560
    },
    {
      "epoch": 0.21377704738427827,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000702,
      "loss": 5.7726,
      "step": 1561
    },
    {
      "epoch": 0.21391399616543413,
      "grad_norm": 0.828125,
      "learning_rate": 0.00070245,
      "loss": 5.68,
      "step": 1562
    },
    {
      "epoch": 0.21405094494658997,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0007029,
      "loss": 5.6829,
      "step": 1563
    },
    {
      "epoch": 0.21418789372774583,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007033499999999999,
      "loss": 5.7518,
      "step": 1564
    },
    {
      "epoch": 0.21432484250890166,
      "grad_norm": 0.875,
      "learning_rate": 0.0007038,
      "loss": 5.7657,
      "step": 1565
    },
    {
      "epoch": 0.21446179129005752,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007042499999999999,
      "loss": 5.7141,
      "step": 1566
    },
    {
      "epoch": 0.21459874007121338,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0007047,
      "loss": 5.6677,
      "step": 1567
    },
    {
      "epoch": 0.2147356888523692,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00070515,
      "loss": 5.7456,
      "step": 1568
    },
    {
      "epoch": 0.21487263763352507,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007056,
      "loss": 5.6845,
      "step": 1569
    },
    {
      "epoch": 0.2150095864146809,
      "grad_norm": 0.78125,
      "learning_rate": 0.00070605,
      "loss": 5.7088,
      "step": 1570
    },
    {
      "epoch": 0.21514653519583676,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0007065,
      "loss": 5.6438,
      "step": 1571
    },
    {
      "epoch": 0.2152834839769926,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0007069499999999999,
      "loss": 5.6754,
      "step": 1572
    },
    {
      "epoch": 0.21542043275814846,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0007074,
      "loss": 5.7274,
      "step": 1573
    },
    {
      "epoch": 0.2155573815393043,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00070785,
      "loss": 5.76,
      "step": 1574
    },
    {
      "epoch": 0.21569433032046015,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0007083,
      "loss": 5.7411,
      "step": 1575
    },
    {
      "epoch": 0.215831279101616,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00070875,
      "loss": 5.755,
      "step": 1576
    },
    {
      "epoch": 0.21596822788277184,
      "grad_norm": 0.71875,
      "learning_rate": 0.0007092,
      "loss": 5.7069,
      "step": 1577
    },
    {
      "epoch": 0.2161051766639277,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0007096499999999999,
      "loss": 5.7636,
      "step": 1578
    },
    {
      "epoch": 0.21624212544508353,
      "grad_norm": 0.6875,
      "learning_rate": 0.0007101,
      "loss": 5.7034,
      "step": 1579
    },
    {
      "epoch": 0.2163790742262394,
      "grad_norm": 0.640625,
      "learning_rate": 0.00071055,
      "loss": 5.7449,
      "step": 1580
    },
    {
      "epoch": 0.21651602300739523,
      "grad_norm": 0.765625,
      "learning_rate": 0.000711,
      "loss": 5.6591,
      "step": 1581
    },
    {
      "epoch": 0.21665297178855109,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00071145,
      "loss": 5.7234,
      "step": 1582
    },
    {
      "epoch": 0.21678992056970692,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007119,
      "loss": 5.813,
      "step": 1583
    },
    {
      "epoch": 0.21692686935086278,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007123499999999999,
      "loss": 5.6991,
      "step": 1584
    },
    {
      "epoch": 0.21706381813201864,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0007128,
      "loss": 5.6271,
      "step": 1585
    },
    {
      "epoch": 0.21720076691317447,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00071325,
      "loss": 5.6658,
      "step": 1586
    },
    {
      "epoch": 0.21733771569433033,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007137000000000001,
      "loss": 5.6977,
      "step": 1587
    },
    {
      "epoch": 0.21747466447548616,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00071415,
      "loss": 5.7259,
      "step": 1588
    },
    {
      "epoch": 0.21761161325664202,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007146,
      "loss": 5.6995,
      "step": 1589
    },
    {
      "epoch": 0.21774856203779785,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00071505,
      "loss": 5.7169,
      "step": 1590
    },
    {
      "epoch": 0.21788551081895371,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0007155,
      "loss": 5.6862,
      "step": 1591
    },
    {
      "epoch": 0.21802245960010955,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007159499999999999,
      "loss": 5.6809,
      "step": 1592
    },
    {
      "epoch": 0.2181594083812654,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0007164000000000001,
      "loss": 5.7443,
      "step": 1593
    },
    {
      "epoch": 0.21829635716242127,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00071685,
      "loss": 5.712,
      "step": 1594
    },
    {
      "epoch": 0.2184333059435771,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0007173,
      "loss": 5.7228,
      "step": 1595
    },
    {
      "epoch": 0.21857025472473296,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00071775,
      "loss": 5.6891,
      "step": 1596
    },
    {
      "epoch": 0.2187072035058888,
      "grad_norm": 0.625,
      "learning_rate": 0.0007182,
      "loss": 5.7003,
      "step": 1597
    },
    {
      "epoch": 0.21884415228704465,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0007186499999999999,
      "loss": 5.7061,
      "step": 1598
    },
    {
      "epoch": 0.21898110106820048,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007191,
      "loss": 5.7066,
      "step": 1599
    },
    {
      "epoch": 0.21911804984935634,
      "grad_norm": 1.0625,
      "learning_rate": 0.00071955,
      "loss": 5.6961,
      "step": 1600
    },
    {
      "epoch": 0.21925499863051218,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00072,
      "loss": 5.6985,
      "step": 1601
    },
    {
      "epoch": 0.21939194741166804,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00072045,
      "loss": 5.6378,
      "step": 1602
    },
    {
      "epoch": 0.2195288961928239,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007209,
      "loss": 5.6674,
      "step": 1603
    },
    {
      "epoch": 0.21966584497397973,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007213499999999999,
      "loss": 5.7317,
      "step": 1604
    },
    {
      "epoch": 0.2198027937551356,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0007218,
      "loss": 5.7248,
      "step": 1605
    },
    {
      "epoch": 0.21993974253629142,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00072225,
      "loss": 5.7211,
      "step": 1606
    },
    {
      "epoch": 0.22007669131744728,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007227000000000001,
      "loss": 5.7592,
      "step": 1607
    },
    {
      "epoch": 0.22021364009860311,
      "grad_norm": 0.734375,
      "learning_rate": 0.00072315,
      "loss": 5.7307,
      "step": 1608
    },
    {
      "epoch": 0.22035058887975897,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007236,
      "loss": 5.6853,
      "step": 1609
    },
    {
      "epoch": 0.2204875376609148,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00072405,
      "loss": 5.7115,
      "step": 1610
    },
    {
      "epoch": 0.22062448644207067,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007245,
      "loss": 5.661,
      "step": 1611
    },
    {
      "epoch": 0.22076143522322653,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007249499999999999,
      "loss": 5.6952,
      "step": 1612
    },
    {
      "epoch": 0.22089838400438236,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007254000000000001,
      "loss": 5.6546,
      "step": 1613
    },
    {
      "epoch": 0.22103533278553822,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00072585,
      "loss": 5.6989,
      "step": 1614
    },
    {
      "epoch": 0.22117228156669405,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007263,
      "loss": 5.6547,
      "step": 1615
    },
    {
      "epoch": 0.2213092303478499,
      "grad_norm": 0.875,
      "learning_rate": 0.00072675,
      "loss": 5.6425,
      "step": 1616
    },
    {
      "epoch": 0.22144617912900574,
      "grad_norm": 0.734375,
      "learning_rate": 0.0007272,
      "loss": 5.7053,
      "step": 1617
    },
    {
      "epoch": 0.2215831279101616,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0007276499999999999,
      "loss": 5.6525,
      "step": 1618
    },
    {
      "epoch": 0.22172007669131744,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007281000000000001,
      "loss": 5.7139,
      "step": 1619
    },
    {
      "epoch": 0.2218570254724733,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00072855,
      "loss": 5.6691,
      "step": 1620
    },
    {
      "epoch": 0.22199397425362916,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000729,
      "loss": 5.7323,
      "step": 1621
    },
    {
      "epoch": 0.222130923034785,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00072945,
      "loss": 5.6972,
      "step": 1622
    },
    {
      "epoch": 0.22226787181594085,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0007299,
      "loss": 5.6793,
      "step": 1623
    },
    {
      "epoch": 0.22240482059709668,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0007303499999999999,
      "loss": 5.6696,
      "step": 1624
    },
    {
      "epoch": 0.22254176937825254,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007308,
      "loss": 5.6398,
      "step": 1625
    },
    {
      "epoch": 0.22267871815940837,
      "grad_norm": 0.5625,
      "learning_rate": 0.00073125,
      "loss": 5.6701,
      "step": 1626
    },
    {
      "epoch": 0.22281566694056423,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007317,
      "loss": 5.6941,
      "step": 1627
    },
    {
      "epoch": 0.22295261572172007,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00073215,
      "loss": 5.736,
      "step": 1628
    },
    {
      "epoch": 0.22308956450287593,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007325999999999999,
      "loss": 5.6677,
      "step": 1629
    },
    {
      "epoch": 0.22322651328403179,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00073305,
      "loss": 5.6904,
      "step": 1630
    },
    {
      "epoch": 0.22336346206518762,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007334999999999999,
      "loss": 5.6335,
      "step": 1631
    },
    {
      "epoch": 0.22350041084634348,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00073395,
      "loss": 5.7056,
      "step": 1632
    },
    {
      "epoch": 0.2236373596274993,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0007344,
      "loss": 5.6598,
      "step": 1633
    },
    {
      "epoch": 0.22377430840865517,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00073485,
      "loss": 5.6576,
      "step": 1634
    },
    {
      "epoch": 0.223911257189811,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007352999999999999,
      "loss": 5.7009,
      "step": 1635
    },
    {
      "epoch": 0.22404820597096686,
      "grad_norm": 0.890625,
      "learning_rate": 0.00073575,
      "loss": 5.7265,
      "step": 1636
    },
    {
      "epoch": 0.2241851547521227,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007361999999999999,
      "loss": 5.658,
      "step": 1637
    },
    {
      "epoch": 0.22432210353327856,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0007366499999999999,
      "loss": 5.6609,
      "step": 1638
    },
    {
      "epoch": 0.22445905231443442,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007371,
      "loss": 5.6803,
      "step": 1639
    },
    {
      "epoch": 0.22459600109559025,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00073755,
      "loss": 5.6509,
      "step": 1640
    },
    {
      "epoch": 0.2247329498767461,
      "grad_norm": 0.625,
      "learning_rate": 0.0007379999999999999,
      "loss": 5.6787,
      "step": 1641
    },
    {
      "epoch": 0.22486989865790194,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00073845,
      "loss": 5.7296,
      "step": 1642
    },
    {
      "epoch": 0.2250068474390578,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007388999999999999,
      "loss": 5.6866,
      "step": 1643
    },
    {
      "epoch": 0.22514379622021363,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00073935,
      "loss": 5.6005,
      "step": 1644
    },
    {
      "epoch": 0.2252807450013695,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007398,
      "loss": 5.6595,
      "step": 1645
    },
    {
      "epoch": 0.22541769378252532,
      "grad_norm": 0.71875,
      "learning_rate": 0.00074025,
      "loss": 5.675,
      "step": 1646
    },
    {
      "epoch": 0.22555464256368118,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0007407,
      "loss": 5.6734,
      "step": 1647
    },
    {
      "epoch": 0.22569159134483704,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00074115,
      "loss": 5.6573,
      "step": 1648
    },
    {
      "epoch": 0.22582854012599288,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007415999999999999,
      "loss": 5.6877,
      "step": 1649
    },
    {
      "epoch": 0.22596548890714874,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00074205,
      "loss": 5.6322,
      "step": 1650
    },
    {
      "epoch": 0.22610243768830457,
      "grad_norm": 0.59375,
      "learning_rate": 0.0007424999999999999,
      "loss": 5.6997,
      "step": 1651
    },
    {
      "epoch": 0.22623938646946043,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00074295,
      "loss": 5.665,
      "step": 1652
    },
    {
      "epoch": 0.22637633525061626,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0007434,
      "loss": 5.6847,
      "step": 1653
    },
    {
      "epoch": 0.22651328403177212,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00074385,
      "loss": 5.6257,
      "step": 1654
    },
    {
      "epoch": 0.22665023281292795,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0007442999999999999,
      "loss": 5.6103,
      "step": 1655
    },
    {
      "epoch": 0.22678718159408381,
      "grad_norm": 1.453125,
      "learning_rate": 0.00074475,
      "loss": 5.6568,
      "step": 1656
    },
    {
      "epoch": 0.22692413037523965,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007451999999999999,
      "loss": 5.7143,
      "step": 1657
    },
    {
      "epoch": 0.2270610791563955,
      "grad_norm": 1.1875,
      "learning_rate": 0.00074565,
      "loss": 5.6768,
      "step": 1658
    },
    {
      "epoch": 0.22719802793755137,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007461,
      "loss": 5.713,
      "step": 1659
    },
    {
      "epoch": 0.2273349767187072,
      "grad_norm": 1.15625,
      "learning_rate": 0.00074655,
      "loss": 5.694,
      "step": 1660
    },
    {
      "epoch": 0.22747192549986306,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007469999999999999,
      "loss": 5.6918,
      "step": 1661
    },
    {
      "epoch": 0.2276088742810189,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00074745,
      "loss": 5.6231,
      "step": 1662
    },
    {
      "epoch": 0.22774582306217475,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007478999999999999,
      "loss": 5.6414,
      "step": 1663
    },
    {
      "epoch": 0.22788277184333058,
      "grad_norm": 1.234375,
      "learning_rate": 0.00074835,
      "loss": 5.7066,
      "step": 1664
    },
    {
      "epoch": 0.22801972062448644,
      "grad_norm": 1.125,
      "learning_rate": 0.0007488,
      "loss": 5.7562,
      "step": 1665
    },
    {
      "epoch": 0.22815666940564228,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00074925,
      "loss": 5.7141,
      "step": 1666
    },
    {
      "epoch": 0.22829361818679814,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007497,
      "loss": 5.6603,
      "step": 1667
    },
    {
      "epoch": 0.228430566967954,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00075015,
      "loss": 5.7741,
      "step": 1668
    },
    {
      "epoch": 0.22856751574910983,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007505999999999999,
      "loss": 5.7081,
      "step": 1669
    },
    {
      "epoch": 0.2287044645302657,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00075105,
      "loss": 5.7014,
      "step": 1670
    },
    {
      "epoch": 0.22884141331142152,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007515,
      "loss": 5.6745,
      "step": 1671
    },
    {
      "epoch": 0.22897836209257738,
      "grad_norm": 0.875,
      "learning_rate": 0.00075195,
      "loss": 5.6807,
      "step": 1672
    },
    {
      "epoch": 0.2291153108737332,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007524,
      "loss": 5.6313,
      "step": 1673
    },
    {
      "epoch": 0.22925225965488907,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00075285,
      "loss": 5.635,
      "step": 1674
    },
    {
      "epoch": 0.2293892084360449,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007532999999999999,
      "loss": 5.705,
      "step": 1675
    },
    {
      "epoch": 0.22952615721720077,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00075375,
      "loss": 5.6938,
      "step": 1676
    },
    {
      "epoch": 0.22966310599835663,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007541999999999999,
      "loss": 5.6961,
      "step": 1677
    },
    {
      "epoch": 0.22980005477951246,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00075465,
      "loss": 5.6077,
      "step": 1678
    },
    {
      "epoch": 0.22993700356066832,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0007551,
      "loss": 5.6499,
      "step": 1679
    },
    {
      "epoch": 0.23007395234182415,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00075555,
      "loss": 5.6354,
      "step": 1680
    },
    {
      "epoch": 0.23021090112298,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0007559999999999999,
      "loss": 5.6048,
      "step": 1681
    },
    {
      "epoch": 0.23034784990413584,
      "grad_norm": 0.71875,
      "learning_rate": 0.00075645,
      "loss": 5.6328,
      "step": 1682
    },
    {
      "epoch": 0.2304847986852917,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007568999999999999,
      "loss": 5.7192,
      "step": 1683
    },
    {
      "epoch": 0.23062174746644754,
      "grad_norm": 0.6875,
      "learning_rate": 0.00075735,
      "loss": 5.6172,
      "step": 1684
    },
    {
      "epoch": 0.2307586962476034,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007578,
      "loss": 5.6239,
      "step": 1685
    },
    {
      "epoch": 0.23089564502875926,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00075825,
      "loss": 5.6543,
      "step": 1686
    },
    {
      "epoch": 0.2310325938099151,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007587,
      "loss": 5.7049,
      "step": 1687
    },
    {
      "epoch": 0.23116954259107095,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00075915,
      "loss": 5.6655,
      "step": 1688
    },
    {
      "epoch": 0.23130649137222678,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007595999999999999,
      "loss": 5.719,
      "step": 1689
    },
    {
      "epoch": 0.23144344015338264,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00076005,
      "loss": 5.7137,
      "step": 1690
    },
    {
      "epoch": 0.23158038893453847,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007605,
      "loss": 5.6186,
      "step": 1691
    },
    {
      "epoch": 0.23171733771569433,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00076095,
      "loss": 5.6875,
      "step": 1692
    },
    {
      "epoch": 0.23185428649685016,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0007614,
      "loss": 5.6002,
      "step": 1693
    },
    {
      "epoch": 0.23199123527800602,
      "grad_norm": 1.109375,
      "learning_rate": 0.00076185,
      "loss": 5.6485,
      "step": 1694
    },
    {
      "epoch": 0.23212818405916189,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0007622999999999999,
      "loss": 5.6761,
      "step": 1695
    },
    {
      "epoch": 0.23226513284031772,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00076275,
      "loss": 5.6422,
      "step": 1696
    },
    {
      "epoch": 0.23240208162147358,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007632,
      "loss": 5.6877,
      "step": 1697
    },
    {
      "epoch": 0.2325390304026294,
      "grad_norm": 0.578125,
      "learning_rate": 0.00076365,
      "loss": 5.7244,
      "step": 1698
    },
    {
      "epoch": 0.23267597918378527,
      "grad_norm": 0.625,
      "learning_rate": 0.0007641,
      "loss": 5.6835,
      "step": 1699
    },
    {
      "epoch": 0.2328129279649411,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00076455,
      "loss": 5.656,
      "step": 1700
    },
    {
      "epoch": 0.23294987674609696,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000765,
      "loss": 5.5815,
      "step": 1701
    },
    {
      "epoch": 0.2330868255272528,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00076545,
      "loss": 5.6567,
      "step": 1702
    },
    {
      "epoch": 0.23322377430840865,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007658999999999999,
      "loss": 5.6711,
      "step": 1703
    },
    {
      "epoch": 0.23336072308956451,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0007663500000000001,
      "loss": 5.7212,
      "step": 1704
    },
    {
      "epoch": 0.23349767187072035,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007668,
      "loss": 5.6685,
      "step": 1705
    },
    {
      "epoch": 0.2336346206518762,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00076725,
      "loss": 5.7001,
      "step": 1706
    },
    {
      "epoch": 0.23377156943303204,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0007677,
      "loss": 5.6798,
      "step": 1707
    },
    {
      "epoch": 0.2339085182141879,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00076815,
      "loss": 5.6515,
      "step": 1708
    },
    {
      "epoch": 0.23404546699534373,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007685999999999999,
      "loss": 5.6027,
      "step": 1709
    },
    {
      "epoch": 0.2341824157764996,
      "grad_norm": 0.984375,
      "learning_rate": 0.00076905,
      "loss": 5.6595,
      "step": 1710
    },
    {
      "epoch": 0.23431936455765542,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007695,
      "loss": 5.6236,
      "step": 1711
    },
    {
      "epoch": 0.23445631333881128,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00076995,
      "loss": 5.6401,
      "step": 1712
    },
    {
      "epoch": 0.23459326211996714,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007704,
      "loss": 5.6502,
      "step": 1713
    },
    {
      "epoch": 0.23473021090112298,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00077085,
      "loss": 5.6089,
      "step": 1714
    },
    {
      "epoch": 0.23486715968227884,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0007712999999999999,
      "loss": 5.6619,
      "step": 1715
    },
    {
      "epoch": 0.23500410846343467,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00077175,
      "loss": 5.5992,
      "step": 1716
    },
    {
      "epoch": 0.23514105724459053,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0007722,
      "loss": 5.6644,
      "step": 1717
    },
    {
      "epoch": 0.23527800602574636,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00077265,
      "loss": 5.607,
      "step": 1718
    },
    {
      "epoch": 0.23541495480690222,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007731,
      "loss": 5.6565,
      "step": 1719
    },
    {
      "epoch": 0.23555190358805805,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00077355,
      "loss": 5.7416,
      "step": 1720
    },
    {
      "epoch": 0.2356888523692139,
      "grad_norm": 0.609375,
      "learning_rate": 0.000774,
      "loss": 5.6354,
      "step": 1721
    },
    {
      "epoch": 0.23582580115036977,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00077445,
      "loss": 5.6891,
      "step": 1722
    },
    {
      "epoch": 0.2359627499315256,
      "grad_norm": 0.71875,
      "learning_rate": 0.0007748999999999999,
      "loss": 5.6312,
      "step": 1723
    },
    {
      "epoch": 0.23609969871268147,
      "grad_norm": 0.59375,
      "learning_rate": 0.0007753500000000001,
      "loss": 5.6917,
      "step": 1724
    },
    {
      "epoch": 0.2362366474938373,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0007758,
      "loss": 5.7261,
      "step": 1725
    },
    {
      "epoch": 0.23637359627499316,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00077625,
      "loss": 5.5953,
      "step": 1726
    },
    {
      "epoch": 0.236510545056149,
      "grad_norm": 0.5625,
      "learning_rate": 0.0007767,
      "loss": 5.6636,
      "step": 1727
    },
    {
      "epoch": 0.23664749383730485,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00077715,
      "loss": 5.6085,
      "step": 1728
    },
    {
      "epoch": 0.23678444261846068,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007775999999999999,
      "loss": 5.7104,
      "step": 1729
    },
    {
      "epoch": 0.23692139139961654,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007780500000000001,
      "loss": 5.602,
      "step": 1730
    },
    {
      "epoch": 0.2370583401807724,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007785,
      "loss": 5.6481,
      "step": 1731
    },
    {
      "epoch": 0.23719528896192824,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00077895,
      "loss": 5.6301,
      "step": 1732
    },
    {
      "epoch": 0.2373322377430841,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0007794,
      "loss": 5.6542,
      "step": 1733
    },
    {
      "epoch": 0.23746918652423993,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00077985,
      "loss": 5.662,
      "step": 1734
    },
    {
      "epoch": 0.2376061353053958,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007802999999999999,
      "loss": 5.6116,
      "step": 1735
    },
    {
      "epoch": 0.23774308408655162,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00078075,
      "loss": 5.6634,
      "step": 1736
    },
    {
      "epoch": 0.23788003286770748,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007812,
      "loss": 5.6737,
      "step": 1737
    },
    {
      "epoch": 0.2380169816488633,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00078165,
      "loss": 5.6176,
      "step": 1738
    },
    {
      "epoch": 0.23815393043001917,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0007821,
      "loss": 5.6296,
      "step": 1739
    },
    {
      "epoch": 0.23829087921117503,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00078255,
      "loss": 5.5284,
      "step": 1740
    },
    {
      "epoch": 0.23842782799233087,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000783,
      "loss": 5.5547,
      "step": 1741
    },
    {
      "epoch": 0.23856477677348673,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00078345,
      "loss": 5.6473,
      "step": 1742
    },
    {
      "epoch": 0.23870172555464256,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007839,
      "loss": 5.5595,
      "step": 1743
    },
    {
      "epoch": 0.23883867433579842,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0007843500000000001,
      "loss": 5.6616,
      "step": 1744
    },
    {
      "epoch": 0.23897562311695425,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007848,
      "loss": 5.6141,
      "step": 1745
    },
    {
      "epoch": 0.2391125718981101,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00078525,
      "loss": 5.656,
      "step": 1746
    },
    {
      "epoch": 0.23924952067926594,
      "grad_norm": 0.578125,
      "learning_rate": 0.0007857,
      "loss": 5.6265,
      "step": 1747
    },
    {
      "epoch": 0.2393864694604218,
      "grad_norm": 0.578125,
      "learning_rate": 0.00078615,
      "loss": 5.5927,
      "step": 1748
    },
    {
      "epoch": 0.23952341824157766,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0007865999999999999,
      "loss": 5.5663,
      "step": 1749
    },
    {
      "epoch": 0.2396603670227335,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0007870500000000001,
      "loss": 5.676,
      "step": 1750
    },
    {
      "epoch": 0.23979731580388935,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0007875,
      "loss": 5.6506,
      "step": 1751
    },
    {
      "epoch": 0.2399342645850452,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0007879499999999999,
      "loss": 5.5719,
      "step": 1752
    },
    {
      "epoch": 0.24007121336620105,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007884,
      "loss": 5.6717,
      "step": 1753
    },
    {
      "epoch": 0.24020816214735688,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0007888499999999999,
      "loss": 5.6584,
      "step": 1754
    },
    {
      "epoch": 0.24034511092851274,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0007892999999999999,
      "loss": 5.5624,
      "step": 1755
    },
    {
      "epoch": 0.24048205970966857,
      "grad_norm": 0.640625,
      "learning_rate": 0.00078975,
      "loss": 5.668,
      "step": 1756
    },
    {
      "epoch": 0.24061900849082443,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0007902,
      "loss": 5.5621,
      "step": 1757
    },
    {
      "epoch": 0.2407559572719803,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00079065,
      "loss": 5.5978,
      "step": 1758
    },
    {
      "epoch": 0.24089290605313612,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007911,
      "loss": 5.6327,
      "step": 1759
    },
    {
      "epoch": 0.24102985483429198,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007915499999999999,
      "loss": 5.6657,
      "step": 1760
    },
    {
      "epoch": 0.24116680361544782,
      "grad_norm": 0.78125,
      "learning_rate": 0.000792,
      "loss": 5.6193,
      "step": 1761
    },
    {
      "epoch": 0.24130375239660368,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007924499999999999,
      "loss": 5.66,
      "step": 1762
    },
    {
      "epoch": 0.2414407011777595,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007929,
      "loss": 5.6791,
      "step": 1763
    },
    {
      "epoch": 0.24157764995891537,
      "grad_norm": 1.109375,
      "learning_rate": 0.00079335,
      "loss": 5.6397,
      "step": 1764
    },
    {
      "epoch": 0.2417145987400712,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007938,
      "loss": 5.6591,
      "step": 1765
    },
    {
      "epoch": 0.24185154752122706,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0007942499999999999,
      "loss": 5.5695,
      "step": 1766
    },
    {
      "epoch": 0.24198849630238292,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0007947,
      "loss": 5.6387,
      "step": 1767
    },
    {
      "epoch": 0.24212544508353875,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007951499999999999,
      "loss": 5.5817,
      "step": 1768
    },
    {
      "epoch": 0.24226239386469461,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007955999999999999,
      "loss": 5.6539,
      "step": 1769
    },
    {
      "epoch": 0.24239934264585045,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00079605,
      "loss": 5.6707,
      "step": 1770
    },
    {
      "epoch": 0.2425362914270063,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007965,
      "loss": 5.6545,
      "step": 1771
    },
    {
      "epoch": 0.24267324020816214,
      "grad_norm": 0.625,
      "learning_rate": 0.0007969499999999999,
      "loss": 5.6079,
      "step": 1772
    },
    {
      "epoch": 0.242810188989318,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007974,
      "loss": 5.6214,
      "step": 1773
    },
    {
      "epoch": 0.24294713777047383,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0007978499999999999,
      "loss": 5.6428,
      "step": 1774
    },
    {
      "epoch": 0.2430840865516297,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007982999999999999,
      "loss": 5.629,
      "step": 1775
    },
    {
      "epoch": 0.24322103533278555,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00079875,
      "loss": 5.6707,
      "step": 1776
    },
    {
      "epoch": 0.24335798411394138,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007992,
      "loss": 5.5817,
      "step": 1777
    },
    {
      "epoch": 0.24349493289509724,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00079965,
      "loss": 5.6067,
      "step": 1778
    },
    {
      "epoch": 0.24363188167625308,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008001,
      "loss": 5.5785,
      "step": 1779
    },
    {
      "epoch": 0.24376883045740894,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008005499999999999,
      "loss": 5.6371,
      "step": 1780
    },
    {
      "epoch": 0.24390577923856477,
      "grad_norm": 0.703125,
      "learning_rate": 0.000801,
      "loss": 5.6769,
      "step": 1781
    },
    {
      "epoch": 0.24404272801972063,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00080145,
      "loss": 5.6314,
      "step": 1782
    },
    {
      "epoch": 0.24417967680087646,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008019,
      "loss": 5.5631,
      "step": 1783
    },
    {
      "epoch": 0.24431662558203232,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00080235,
      "loss": 5.7018,
      "step": 1784
    },
    {
      "epoch": 0.24445357436318818,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008028,
      "loss": 5.6441,
      "step": 1785
    },
    {
      "epoch": 0.244590523144344,
      "grad_norm": 0.75,
      "learning_rate": 0.0008032499999999999,
      "loss": 5.5831,
      "step": 1786
    },
    {
      "epoch": 0.24472747192549987,
      "grad_norm": 0.8125,
      "learning_rate": 0.0008037,
      "loss": 5.6854,
      "step": 1787
    },
    {
      "epoch": 0.2448644207066557,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008041499999999999,
      "loss": 5.6209,
      "step": 1788
    },
    {
      "epoch": 0.24500136948781157,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008046,
      "loss": 5.6263,
      "step": 1789
    },
    {
      "epoch": 0.2451383182689674,
      "grad_norm": 0.5625,
      "learning_rate": 0.00080505,
      "loss": 5.6318,
      "step": 1790
    },
    {
      "epoch": 0.24527526705012326,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008055,
      "loss": 5.6511,
      "step": 1791
    },
    {
      "epoch": 0.2454122158312791,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008059499999999999,
      "loss": 5.6726,
      "step": 1792
    },
    {
      "epoch": 0.24554916461243495,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008064,
      "loss": 5.6164,
      "step": 1793
    },
    {
      "epoch": 0.2456861133935908,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008068499999999999,
      "loss": 5.5634,
      "step": 1794
    },
    {
      "epoch": 0.24582306217474664,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008072999999999999,
      "loss": 5.5941,
      "step": 1795
    },
    {
      "epoch": 0.2459600109559025,
      "grad_norm": 1.25,
      "learning_rate": 0.00080775,
      "loss": 5.6254,
      "step": 1796
    },
    {
      "epoch": 0.24609695973705834,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008082,
      "loss": 5.6084,
      "step": 1797
    },
    {
      "epoch": 0.2462339085182142,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00080865,
      "loss": 5.6054,
      "step": 1798
    },
    {
      "epoch": 0.24637085729937003,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0008091,
      "loss": 5.5905,
      "step": 1799
    },
    {
      "epoch": 0.2465078060805259,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008095499999999999,
      "loss": 5.6886,
      "step": 1800
    },
    {
      "epoch": 0.24664475486168172,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00081,
      "loss": 5.6628,
      "step": 1801
    },
    {
      "epoch": 0.24678170364283758,
      "grad_norm": 0.625,
      "learning_rate": 0.00081045,
      "loss": 5.6123,
      "step": 1802
    },
    {
      "epoch": 0.24691865242399344,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008109,
      "loss": 5.6241,
      "step": 1803
    },
    {
      "epoch": 0.24705560120514927,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00081135,
      "loss": 5.6542,
      "step": 1804
    },
    {
      "epoch": 0.24719254998630513,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008118,
      "loss": 5.5808,
      "step": 1805
    },
    {
      "epoch": 0.24732949876746096,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008122499999999999,
      "loss": 5.5877,
      "step": 1806
    },
    {
      "epoch": 0.24746644754861682,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008127,
      "loss": 5.6694,
      "step": 1807
    },
    {
      "epoch": 0.24760339632977266,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008131499999999999,
      "loss": 5.6282,
      "step": 1808
    },
    {
      "epoch": 0.24774034511092852,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008136,
      "loss": 5.5376,
      "step": 1809
    },
    {
      "epoch": 0.24787729389208435,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00081405,
      "loss": 5.6759,
      "step": 1810
    },
    {
      "epoch": 0.2480142426732402,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008145,
      "loss": 5.6181,
      "step": 1811
    },
    {
      "epoch": 0.24815119145439607,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008149499999999999,
      "loss": 5.6004,
      "step": 1812
    },
    {
      "epoch": 0.2482881402355519,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008154,
      "loss": 5.5866,
      "step": 1813
    },
    {
      "epoch": 0.24842508901670776,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008158499999999999,
      "loss": 5.6194,
      "step": 1814
    },
    {
      "epoch": 0.2485620377978636,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008163000000000001,
      "loss": 5.6492,
      "step": 1815
    },
    {
      "epoch": 0.24869898657901945,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00081675,
      "loss": 5.5969,
      "step": 1816
    },
    {
      "epoch": 0.2488359353601753,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008172,
      "loss": 5.5798,
      "step": 1817
    },
    {
      "epoch": 0.24897288414133115,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00081765,
      "loss": 5.575,
      "step": 1818
    },
    {
      "epoch": 0.24910983292248698,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008181,
      "loss": 5.6014,
      "step": 1819
    },
    {
      "epoch": 0.24924678170364284,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008185499999999999,
      "loss": 5.6135,
      "step": 1820
    },
    {
      "epoch": 0.2493837304847987,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000819,
      "loss": 5.6436,
      "step": 1821
    },
    {
      "epoch": 0.24952067926595453,
      "grad_norm": 0.609375,
      "learning_rate": 0.00081945,
      "loss": 5.6268,
      "step": 1822
    },
    {
      "epoch": 0.2496576280471104,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008199,
      "loss": 5.5542,
      "step": 1823
    },
    {
      "epoch": 0.24979457682826622,
      "grad_norm": 0.609375,
      "learning_rate": 0.00082035,
      "loss": 5.5717,
      "step": 1824
    },
    {
      "epoch": 0.24993152560942208,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008208,
      "loss": 5.6046,
      "step": 1825
    },
    {
      "epoch": 0.25006847439057794,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008212499999999999,
      "loss": 5.6158,
      "step": 1826
    },
    {
      "epoch": 0.2502054231717338,
      "grad_norm": 0.75,
      "learning_rate": 0.0008217,
      "loss": 5.6222,
      "step": 1827
    },
    {
      "epoch": 0.2503423719528896,
      "grad_norm": 0.8125,
      "learning_rate": 0.00082215,
      "loss": 5.6384,
      "step": 1828
    },
    {
      "epoch": 0.25047932073404544,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008226,
      "loss": 5.6829,
      "step": 1829
    },
    {
      "epoch": 0.25061626951520133,
      "grad_norm": 0.640625,
      "learning_rate": 0.00082305,
      "loss": 5.6499,
      "step": 1830
    },
    {
      "epoch": 0.25075321829635716,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008235,
      "loss": 5.5449,
      "step": 1831
    },
    {
      "epoch": 0.250890167077513,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008239499999999999,
      "loss": 5.6055,
      "step": 1832
    },
    {
      "epoch": 0.2510271158586689,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008244,
      "loss": 5.5793,
      "step": 1833
    },
    {
      "epoch": 0.2511640646398247,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008248499999999999,
      "loss": 5.6298,
      "step": 1834
    },
    {
      "epoch": 0.25130101342098055,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008253000000000001,
      "loss": 5.6578,
      "step": 1835
    },
    {
      "epoch": 0.2514379622021364,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00082575,
      "loss": 5.6147,
      "step": 1836
    },
    {
      "epoch": 0.25157491098329227,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008262,
      "loss": 5.5755,
      "step": 1837
    },
    {
      "epoch": 0.2517118597644481,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00082665,
      "loss": 5.5559,
      "step": 1838
    },
    {
      "epoch": 0.25184880854560393,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008271,
      "loss": 5.598,
      "step": 1839
    },
    {
      "epoch": 0.2519857573267598,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008275499999999999,
      "loss": 5.5929,
      "step": 1840
    },
    {
      "epoch": 0.25212270610791565,
      "grad_norm": 0.71875,
      "learning_rate": 0.000828,
      "loss": 5.5704,
      "step": 1841
    },
    {
      "epoch": 0.2522596548890715,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00082845,
      "loss": 5.5984,
      "step": 1842
    },
    {
      "epoch": 0.2523966036702273,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008289,
      "loss": 5.5881,
      "step": 1843
    },
    {
      "epoch": 0.2525335524513832,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00082935,
      "loss": 5.6243,
      "step": 1844
    },
    {
      "epoch": 0.25267050123253904,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008298,
      "loss": 5.6185,
      "step": 1845
    },
    {
      "epoch": 0.25280745001369487,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008302499999999999,
      "loss": 5.5954,
      "step": 1846
    },
    {
      "epoch": 0.2529443987948507,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008307,
      "loss": 5.5305,
      "step": 1847
    },
    {
      "epoch": 0.2530813475760066,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00083115,
      "loss": 5.6085,
      "step": 1848
    },
    {
      "epoch": 0.2532182963571624,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008316,
      "loss": 5.5721,
      "step": 1849
    },
    {
      "epoch": 0.25335524513831825,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00083205,
      "loss": 5.579,
      "step": 1850
    },
    {
      "epoch": 0.25349219391947414,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008325,
      "loss": 5.5839,
      "step": 1851
    },
    {
      "epoch": 0.25362914270063,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008329499999999999,
      "loss": 5.5795,
      "step": 1852
    },
    {
      "epoch": 0.2537660914817858,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008334,
      "loss": 5.6557,
      "step": 1853
    },
    {
      "epoch": 0.25390304026294164,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00083385,
      "loss": 5.626,
      "step": 1854
    },
    {
      "epoch": 0.2540399890440975,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008343000000000001,
      "loss": 5.5584,
      "step": 1855
    },
    {
      "epoch": 0.25417693782525336,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00083475,
      "loss": 5.578,
      "step": 1856
    },
    {
      "epoch": 0.2543138866064092,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008352,
      "loss": 5.6298,
      "step": 1857
    },
    {
      "epoch": 0.2544508353875651,
      "grad_norm": 0.84375,
      "learning_rate": 0.00083565,
      "loss": 5.6566,
      "step": 1858
    },
    {
      "epoch": 0.2545877841687209,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008361,
      "loss": 5.5961,
      "step": 1859
    },
    {
      "epoch": 0.25472473294987674,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008365499999999999,
      "loss": 5.6033,
      "step": 1860
    },
    {
      "epoch": 0.2548616817310326,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008370000000000001,
      "loss": 5.6306,
      "step": 1861
    },
    {
      "epoch": 0.25499863051218846,
      "grad_norm": 0.625,
      "learning_rate": 0.00083745,
      "loss": 5.5725,
      "step": 1862
    },
    {
      "epoch": 0.2551355792933443,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008379,
      "loss": 5.5727,
      "step": 1863
    },
    {
      "epoch": 0.2552725280745001,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00083835,
      "loss": 5.5719,
      "step": 1864
    },
    {
      "epoch": 0.25540947685565596,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008388,
      "loss": 5.5377,
      "step": 1865
    },
    {
      "epoch": 0.25554642563681185,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008392499999999999,
      "loss": 5.6022,
      "step": 1866
    },
    {
      "epoch": 0.2556833744179677,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008397,
      "loss": 5.6066,
      "step": 1867
    },
    {
      "epoch": 0.2558203231991235,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00084015,
      "loss": 5.6335,
      "step": 1868
    },
    {
      "epoch": 0.2559572719802794,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008406,
      "loss": 5.6199,
      "step": 1869
    },
    {
      "epoch": 0.25609422076143523,
      "grad_norm": 0.609375,
      "learning_rate": 0.00084105,
      "loss": 5.6021,
      "step": 1870
    },
    {
      "epoch": 0.25623116954259106,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008415,
      "loss": 5.5959,
      "step": 1871
    },
    {
      "epoch": 0.2563681183237469,
      "grad_norm": 0.609375,
      "learning_rate": 0.00084195,
      "loss": 5.5352,
      "step": 1872
    },
    {
      "epoch": 0.2565050671049028,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008424,
      "loss": 5.5674,
      "step": 1873
    },
    {
      "epoch": 0.2566420158860586,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00084285,
      "loss": 5.6306,
      "step": 1874
    },
    {
      "epoch": 0.25677896466721445,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008433000000000001,
      "loss": 5.5416,
      "step": 1875
    },
    {
      "epoch": 0.25691591344837034,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00084375,
      "loss": 5.583,
      "step": 1876
    },
    {
      "epoch": 0.25705286222952617,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008441999999999999,
      "loss": 5.5666,
      "step": 1877
    },
    {
      "epoch": 0.257189811010682,
      "grad_norm": 0.625,
      "learning_rate": 0.00084465,
      "loss": 5.5658,
      "step": 1878
    },
    {
      "epoch": 0.25732675979183783,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008450999999999999,
      "loss": 5.5293,
      "step": 1879
    },
    {
      "epoch": 0.2574637085729937,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008455499999999999,
      "loss": 5.5689,
      "step": 1880
    },
    {
      "epoch": 0.25760065735414955,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000846,
      "loss": 5.6384,
      "step": 1881
    },
    {
      "epoch": 0.2577376061353054,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00084645,
      "loss": 5.6113,
      "step": 1882
    },
    {
      "epoch": 0.2578745549164612,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008468999999999999,
      "loss": 5.6119,
      "step": 1883
    },
    {
      "epoch": 0.2580115036976171,
      "grad_norm": 0.671875,
      "learning_rate": 0.00084735,
      "loss": 5.6074,
      "step": 1884
    },
    {
      "epoch": 0.25814845247877294,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008477999999999999,
      "loss": 5.632,
      "step": 1885
    },
    {
      "epoch": 0.25828540125992877,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008482499999999999,
      "loss": 5.5559,
      "step": 1886
    },
    {
      "epoch": 0.25842235004108466,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008487,
      "loss": 5.5717,
      "step": 1887
    },
    {
      "epoch": 0.2585592988222405,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00084915,
      "loss": 5.4796,
      "step": 1888
    },
    {
      "epoch": 0.2586962476033963,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008495999999999999,
      "loss": 5.5896,
      "step": 1889
    },
    {
      "epoch": 0.25883319638455216,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00085005,
      "loss": 5.5476,
      "step": 1890
    },
    {
      "epoch": 0.25897014516570804,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008504999999999999,
      "loss": 5.5868,
      "step": 1891
    },
    {
      "epoch": 0.2591070939468639,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00085095,
      "loss": 5.5862,
      "step": 1892
    },
    {
      "epoch": 0.2592440427280197,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008514,
      "loss": 5.587,
      "step": 1893
    },
    {
      "epoch": 0.2593809915091756,
      "grad_norm": 0.6875,
      "learning_rate": 0.00085185,
      "loss": 5.5301,
      "step": 1894
    },
    {
      "epoch": 0.25951794029033143,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008523,
      "loss": 5.6275,
      "step": 1895
    },
    {
      "epoch": 0.25965488907148726,
      "grad_norm": 0.671875,
      "learning_rate": 0.00085275,
      "loss": 5.5536,
      "step": 1896
    },
    {
      "epoch": 0.2597918378526431,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008531999999999999,
      "loss": 5.6139,
      "step": 1897
    },
    {
      "epoch": 0.259928786633799,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00085365,
      "loss": 5.6015,
      "step": 1898
    },
    {
      "epoch": 0.2600657354149548,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008540999999999999,
      "loss": 5.514,
      "step": 1899
    },
    {
      "epoch": 0.26020268419611065,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00085455,
      "loss": 5.6079,
      "step": 1900
    },
    {
      "epoch": 0.2603396329772665,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000855,
      "loss": 5.5618,
      "step": 1901
    },
    {
      "epoch": 0.26047658175842237,
      "grad_norm": 0.84375,
      "learning_rate": 0.00085545,
      "loss": 5.5266,
      "step": 1902
    },
    {
      "epoch": 0.2606135305395782,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008558999999999999,
      "loss": 5.6237,
      "step": 1903
    },
    {
      "epoch": 0.26075047932073403,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00085635,
      "loss": 5.6213,
      "step": 1904
    },
    {
      "epoch": 0.2608874281018899,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008567999999999999,
      "loss": 5.5183,
      "step": 1905
    },
    {
      "epoch": 0.26102437688304575,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008572499999999999,
      "loss": 5.5364,
      "step": 1906
    },
    {
      "epoch": 0.2611613256642016,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008577,
      "loss": 5.6165,
      "step": 1907
    },
    {
      "epoch": 0.2612982744453574,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00085815,
      "loss": 5.5694,
      "step": 1908
    },
    {
      "epoch": 0.2614352232265133,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008585999999999999,
      "loss": 5.5694,
      "step": 1909
    },
    {
      "epoch": 0.26157217200766913,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00085905,
      "loss": 5.5571,
      "step": 1910
    },
    {
      "epoch": 0.26170912078882497,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008594999999999999,
      "loss": 5.5581,
      "step": 1911
    },
    {
      "epoch": 0.26184606956998085,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00085995,
      "loss": 5.5922,
      "step": 1912
    },
    {
      "epoch": 0.2619830183511367,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008604,
      "loss": 5.5737,
      "step": 1913
    },
    {
      "epoch": 0.2621199671322925,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00086085,
      "loss": 5.5675,
      "step": 1914
    },
    {
      "epoch": 0.26225691591344835,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008613,
      "loss": 5.5194,
      "step": 1915
    },
    {
      "epoch": 0.26239386469460424,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00086175,
      "loss": 5.6152,
      "step": 1916
    },
    {
      "epoch": 0.26253081347576007,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008621999999999999,
      "loss": 5.5152,
      "step": 1917
    },
    {
      "epoch": 0.2626677622569159,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00086265,
      "loss": 5.5256,
      "step": 1918
    },
    {
      "epoch": 0.26280471103807174,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008630999999999999,
      "loss": 5.6171,
      "step": 1919
    },
    {
      "epoch": 0.2629416598192276,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00086355,
      "loss": 5.6272,
      "step": 1920
    },
    {
      "epoch": 0.26307860860038346,
      "grad_norm": 0.515625,
      "learning_rate": 0.000864,
      "loss": 5.5942,
      "step": 1921
    },
    {
      "epoch": 0.2632155573815393,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00086445,
      "loss": 5.5791,
      "step": 1922
    },
    {
      "epoch": 0.2633525061626952,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008648999999999999,
      "loss": 5.5942,
      "step": 1923
    },
    {
      "epoch": 0.263489454943851,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00086535,
      "loss": 5.5739,
      "step": 1924
    },
    {
      "epoch": 0.26362640372500684,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008657999999999999,
      "loss": 5.5383,
      "step": 1925
    },
    {
      "epoch": 0.2637633525061627,
      "grad_norm": 0.53125,
      "learning_rate": 0.00086625,
      "loss": 5.5317,
      "step": 1926
    },
    {
      "epoch": 0.26390030128731856,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008667,
      "loss": 5.5346,
      "step": 1927
    },
    {
      "epoch": 0.2640372500684744,
      "grad_norm": 0.6875,
      "learning_rate": 0.00086715,
      "loss": 5.5512,
      "step": 1928
    },
    {
      "epoch": 0.2641741988496302,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008676,
      "loss": 5.4822,
      "step": 1929
    },
    {
      "epoch": 0.26431114763078606,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00086805,
      "loss": 5.5993,
      "step": 1930
    },
    {
      "epoch": 0.26444809641194195,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008684999999999999,
      "loss": 5.6348,
      "step": 1931
    },
    {
      "epoch": 0.2645850451930978,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00086895,
      "loss": 5.6172,
      "step": 1932
    },
    {
      "epoch": 0.2647219939742536,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008694,
      "loss": 5.6104,
      "step": 1933
    },
    {
      "epoch": 0.2648589427554095,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00086985,
      "loss": 5.5128,
      "step": 1934
    },
    {
      "epoch": 0.26499589153656533,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008703,
      "loss": 5.6216,
      "step": 1935
    },
    {
      "epoch": 0.26513284031772116,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00087075,
      "loss": 5.5225,
      "step": 1936
    },
    {
      "epoch": 0.265269789098877,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008711999999999999,
      "loss": 5.5628,
      "step": 1937
    },
    {
      "epoch": 0.2654067378800329,
      "grad_norm": 0.6875,
      "learning_rate": 0.00087165,
      "loss": 5.5784,
      "step": 1938
    },
    {
      "epoch": 0.2655436866611887,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008721,
      "loss": 5.5005,
      "step": 1939
    },
    {
      "epoch": 0.26568063544234455,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00087255,
      "loss": 5.5244,
      "step": 1940
    },
    {
      "epoch": 0.26581758422350044,
      "grad_norm": 0.796875,
      "learning_rate": 0.000873,
      "loss": 5.5723,
      "step": 1941
    },
    {
      "epoch": 0.26595453300465627,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00087345,
      "loss": 5.5581,
      "step": 1942
    },
    {
      "epoch": 0.2660914817858121,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008738999999999999,
      "loss": 5.5457,
      "step": 1943
    },
    {
      "epoch": 0.26622843056696793,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00087435,
      "loss": 5.5777,
      "step": 1944
    },
    {
      "epoch": 0.2663653793481238,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008747999999999999,
      "loss": 5.5399,
      "step": 1945
    },
    {
      "epoch": 0.26650232812927965,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00087525,
      "loss": 5.5242,
      "step": 1946
    },
    {
      "epoch": 0.2666392769104355,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008757,
      "loss": 5.5831,
      "step": 1947
    },
    {
      "epoch": 0.2667762256915913,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00087615,
      "loss": 5.5985,
      "step": 1948
    },
    {
      "epoch": 0.2669131744727472,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008766,
      "loss": 5.5175,
      "step": 1949
    },
    {
      "epoch": 0.26705012325390304,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00087705,
      "loss": 5.513,
      "step": 1950
    },
    {
      "epoch": 0.26718707203505887,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008774999999999999,
      "loss": 5.5616,
      "step": 1951
    },
    {
      "epoch": 0.26732402081621476,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00087795,
      "loss": 5.5546,
      "step": 1952
    },
    {
      "epoch": 0.2674609695973706,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008784,
      "loss": 5.5092,
      "step": 1953
    },
    {
      "epoch": 0.2675979183785264,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00087885,
      "loss": 5.5489,
      "step": 1954
    },
    {
      "epoch": 0.26773486715968225,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008793,
      "loss": 5.565,
      "step": 1955
    },
    {
      "epoch": 0.26787181594083814,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00087975,
      "loss": 5.5631,
      "step": 1956
    },
    {
      "epoch": 0.268008764721994,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008801999999999999,
      "loss": 5.5901,
      "step": 1957
    },
    {
      "epoch": 0.2681457135031498,
      "grad_norm": 0.515625,
      "learning_rate": 0.00088065,
      "loss": 5.5527,
      "step": 1958
    },
    {
      "epoch": 0.2682826622843057,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008811,
      "loss": 5.5438,
      "step": 1959
    },
    {
      "epoch": 0.2684196110654615,
      "grad_norm": 0.5625,
      "learning_rate": 0.00088155,
      "loss": 5.5933,
      "step": 1960
    },
    {
      "epoch": 0.26855655984661736,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000882,
      "loss": 5.5674,
      "step": 1961
    },
    {
      "epoch": 0.2686935086277732,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00088245,
      "loss": 5.4829,
      "step": 1962
    },
    {
      "epoch": 0.2688304574089291,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008828999999999999,
      "loss": 5.5843,
      "step": 1963
    },
    {
      "epoch": 0.2689674061900849,
      "grad_norm": 0.609375,
      "learning_rate": 0.00088335,
      "loss": 5.5943,
      "step": 1964
    },
    {
      "epoch": 0.26910435497124074,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008838,
      "loss": 5.5243,
      "step": 1965
    },
    {
      "epoch": 0.2692413037523966,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00088425,
      "loss": 5.5143,
      "step": 1966
    },
    {
      "epoch": 0.26937825253355246,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008847,
      "loss": 5.5048,
      "step": 1967
    },
    {
      "epoch": 0.2695152013147083,
      "grad_norm": 0.921875,
      "learning_rate": 0.00088515,
      "loss": 5.5107,
      "step": 1968
    },
    {
      "epoch": 0.26965215009586413,
      "grad_norm": 1.0,
      "learning_rate": 0.0008856,
      "loss": 5.582,
      "step": 1969
    },
    {
      "epoch": 0.26978909887702,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00088605,
      "loss": 5.529,
      "step": 1970
    },
    {
      "epoch": 0.26992604765817585,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008864999999999999,
      "loss": 5.5181,
      "step": 1971
    },
    {
      "epoch": 0.2700629964393317,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008869500000000001,
      "loss": 5.5835,
      "step": 1972
    },
    {
      "epoch": 0.2701999452204875,
      "grad_norm": 0.8125,
      "learning_rate": 0.0008874,
      "loss": 5.5731,
      "step": 1973
    },
    {
      "epoch": 0.2703368940016434,
      "grad_norm": 0.703125,
      "learning_rate": 0.00088785,
      "loss": 5.5657,
      "step": 1974
    },
    {
      "epoch": 0.27047384278279923,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008883,
      "loss": 5.5269,
      "step": 1975
    },
    {
      "epoch": 0.27061079156395507,
      "grad_norm": 0.6875,
      "learning_rate": 0.00088875,
      "loss": 5.6001,
      "step": 1976
    },
    {
      "epoch": 0.27074774034511095,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008891999999999999,
      "loss": 5.5771,
      "step": 1977
    },
    {
      "epoch": 0.2708846891262668,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00088965,
      "loss": 5.5518,
      "step": 1978
    },
    {
      "epoch": 0.2710216379074226,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008901,
      "loss": 5.5893,
      "step": 1979
    },
    {
      "epoch": 0.27115858668857845,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00089055,
      "loss": 5.5682,
      "step": 1980
    },
    {
      "epoch": 0.27129553546973434,
      "grad_norm": 0.71875,
      "learning_rate": 0.000891,
      "loss": 5.4399,
      "step": 1981
    },
    {
      "epoch": 0.27143248425089017,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00089145,
      "loss": 5.5034,
      "step": 1982
    },
    {
      "epoch": 0.271569433032046,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008918999999999999,
      "loss": 5.4821,
      "step": 1983
    },
    {
      "epoch": 0.27170638181320184,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00089235,
      "loss": 5.5544,
      "step": 1984
    },
    {
      "epoch": 0.2718433305943577,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008928,
      "loss": 5.5416,
      "step": 1985
    },
    {
      "epoch": 0.27198027937551356,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008932500000000001,
      "loss": 5.4881,
      "step": 1986
    },
    {
      "epoch": 0.2721172281566694,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008937,
      "loss": 5.5343,
      "step": 1987
    },
    {
      "epoch": 0.2722541769378253,
      "grad_norm": 0.6875,
      "learning_rate": 0.00089415,
      "loss": 5.619,
      "step": 1988
    },
    {
      "epoch": 0.2723911257189811,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008946,
      "loss": 5.5286,
      "step": 1989
    },
    {
      "epoch": 0.27252807450013694,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00089505,
      "loss": 5.5587,
      "step": 1990
    },
    {
      "epoch": 0.2726650232812928,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008954999999999999,
      "loss": 5.5458,
      "step": 1991
    },
    {
      "epoch": 0.27280197206244866,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008959500000000001,
      "loss": 5.5222,
      "step": 1992
    },
    {
      "epoch": 0.2729389208436045,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008964,
      "loss": 5.5112,
      "step": 1993
    },
    {
      "epoch": 0.2730758696247603,
      "grad_norm": 0.578125,
      "learning_rate": 0.00089685,
      "loss": 5.5508,
      "step": 1994
    },
    {
      "epoch": 0.2732128184059162,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008973,
      "loss": 5.54,
      "step": 1995
    },
    {
      "epoch": 0.27334976718707205,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00089775,
      "loss": 5.5335,
      "step": 1996
    },
    {
      "epoch": 0.2734867159682279,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008981999999999999,
      "loss": 5.5011,
      "step": 1997
    },
    {
      "epoch": 0.2736236647493837,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008986500000000001,
      "loss": 5.5409,
      "step": 1998
    },
    {
      "epoch": 0.2737606135305396,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008991,
      "loss": 5.5267,
      "step": 1999
    },
    {
      "epoch": 0.27389756231169543,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00089955,
      "loss": 5.4655,
      "step": 2000
    },
    {
      "epoch": 0.27403451109285126,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0009,
      "loss": 5.5548,
      "step": 2001
    },
    {
      "epoch": 0.2741714598740071,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008999999981353713,
      "loss": 5.5387,
      "step": 2002
    },
    {
      "epoch": 0.274308408655163,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008999999925414856,
      "loss": 5.59,
      "step": 2003
    },
    {
      "epoch": 0.2744453574363188,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0008999999832183425,
      "loss": 5.5438,
      "step": 2004
    },
    {
      "epoch": 0.27458230621747465,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008999999701659424,
      "loss": 5.522,
      "step": 2005
    },
    {
      "epoch": 0.27471925499863054,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008999999533842854,
      "loss": 5.496,
      "step": 2006
    },
    {
      "epoch": 0.27485620377978637,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008999999328733715,
      "loss": 5.5858,
      "step": 2007
    },
    {
      "epoch": 0.2749931525609422,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008999999086332009,
      "loss": 5.5544,
      "step": 2008
    },
    {
      "epoch": 0.27513010134209803,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008999998806637739,
      "loss": 5.4703,
      "step": 2009
    },
    {
      "epoch": 0.2752670501232539,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008999998489650906,
      "loss": 5.5136,
      "step": 2010
    },
    {
      "epoch": 0.27540399890440975,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008999998135371512,
      "loss": 5.5486,
      "step": 2011
    },
    {
      "epoch": 0.2755409476855656,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008999997743799563,
      "loss": 5.5492,
      "step": 2012
    },
    {
      "epoch": 0.2756778964667215,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008999997314935059,
      "loss": 5.4869,
      "step": 2013
    },
    {
      "epoch": 0.2758148452478773,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008999996848778006,
      "loss": 5.5514,
      "step": 2014
    },
    {
      "epoch": 0.27595179402903314,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008999996345328407,
      "loss": 5.5156,
      "step": 2015
    },
    {
      "epoch": 0.27608874281018897,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008999995804586265,
      "loss": 5.5426,
      "step": 2016
    },
    {
      "epoch": 0.27622569159134486,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008999995226551587,
      "loss": 5.4967,
      "step": 2017
    },
    {
      "epoch": 0.2763626403725007,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008999994611224376,
      "loss": 5.5449,
      "step": 2018
    },
    {
      "epoch": 0.2764995891536565,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008999993958604636,
      "loss": 5.5804,
      "step": 2019
    },
    {
      "epoch": 0.27663653793481235,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008999993268692375,
      "loss": 5.4789,
      "step": 2020
    },
    {
      "epoch": 0.27677348671596824,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008999992541487596,
      "loss": 5.5138,
      "step": 2021
    },
    {
      "epoch": 0.2769104354971241,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008999991776990309,
      "loss": 5.4879,
      "step": 2022
    },
    {
      "epoch": 0.2770473842782799,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008999990975200516,
      "loss": 5.4969,
      "step": 2023
    },
    {
      "epoch": 0.2771843330594358,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008999990136118226,
      "loss": 5.4894,
      "step": 2024
    },
    {
      "epoch": 0.2773212818405916,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008999989259743445,
      "loss": 5.4859,
      "step": 2025
    },
    {
      "epoch": 0.27745823062174746,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008999988346076181,
      "loss": 5.4341,
      "step": 2026
    },
    {
      "epoch": 0.2775951794029033,
      "grad_norm": 0.90625,
      "learning_rate": 0.000899998739511644,
      "loss": 5.5987,
      "step": 2027
    },
    {
      "epoch": 0.2777321281840592,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008999986406864233,
      "loss": 5.5205,
      "step": 2028
    },
    {
      "epoch": 0.277869076965215,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008999985381319567,
      "loss": 5.5507,
      "step": 2029
    },
    {
      "epoch": 0.27800602574637084,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008999984318482448,
      "loss": 5.4764,
      "step": 2030
    },
    {
      "epoch": 0.27814297452752673,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008999983218352888,
      "loss": 5.4908,
      "step": 2031
    },
    {
      "epoch": 0.27827992330868256,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008999982080930894,
      "loss": 5.5235,
      "step": 2032
    },
    {
      "epoch": 0.2784168720898384,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008999980906216475,
      "loss": 5.5304,
      "step": 2033
    },
    {
      "epoch": 0.27855382087099423,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008999979694209644,
      "loss": 5.5142,
      "step": 2034
    },
    {
      "epoch": 0.2786907696521501,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008999978444910408,
      "loss": 5.5108,
      "step": 2035
    },
    {
      "epoch": 0.27882771843330595,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000899997715831878,
      "loss": 5.4806,
      "step": 2036
    },
    {
      "epoch": 0.2789646672144618,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008999975834434767,
      "loss": 5.513,
      "step": 2037
    },
    {
      "epoch": 0.2791016159956176,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008999974473258382,
      "loss": 5.5551,
      "step": 2038
    },
    {
      "epoch": 0.2792385647767735,
      "grad_norm": 0.625,
      "learning_rate": 0.0008999973074789638,
      "loss": 5.5797,
      "step": 2039
    },
    {
      "epoch": 0.27937551355792933,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008999971639028542,
      "loss": 5.4992,
      "step": 2040
    },
    {
      "epoch": 0.27951246233908517,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008999970165975112,
      "loss": 5.5208,
      "step": 2041
    },
    {
      "epoch": 0.27964941112024105,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008999968655629355,
      "loss": 5.4881,
      "step": 2042
    },
    {
      "epoch": 0.2797863599013969,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008999967107991285,
      "loss": 5.4958,
      "step": 2043
    },
    {
      "epoch": 0.2799233086825527,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008999965523060917,
      "loss": 5.5973,
      "step": 2044
    },
    {
      "epoch": 0.28006025746370855,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008999963900838262,
      "loss": 5.4611,
      "step": 2045
    },
    {
      "epoch": 0.28019720624486444,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008999962241323333,
      "loss": 5.516,
      "step": 2046
    },
    {
      "epoch": 0.28033415502602027,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008999960544516146,
      "loss": 5.4768,
      "step": 2047
    },
    {
      "epoch": 0.2804711038071761,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008999958810416712,
      "loss": 5.4634,
      "step": 2048
    },
    {
      "epoch": 0.280608052588332,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008999957039025049,
      "loss": 5.4741,
      "step": 2049
    },
    {
      "epoch": 0.2807450013694878,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008999955230341168,
      "loss": 5.4778,
      "step": 2050
    },
    {
      "epoch": 0.28088195015064366,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008999953384365085,
      "loss": 5.4775,
      "step": 2051
    },
    {
      "epoch": 0.2810188989317995,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008999951501096819,
      "loss": 5.511,
      "step": 2052
    },
    {
      "epoch": 0.2811558477129554,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008999949580536379,
      "loss": 5.4636,
      "step": 2053
    },
    {
      "epoch": 0.2812927964941112,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008999947622683787,
      "loss": 5.5389,
      "step": 2054
    },
    {
      "epoch": 0.28142974527526704,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008999945627539056,
      "loss": 5.5384,
      "step": 2055
    },
    {
      "epoch": 0.2815666940564229,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008999943595102203,
      "loss": 5.4866,
      "step": 2056
    },
    {
      "epoch": 0.28170364283757876,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008999941525373247,
      "loss": 5.5672,
      "step": 2057
    },
    {
      "epoch": 0.2818405916187346,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008999939418352201,
      "loss": 5.4992,
      "step": 2058
    },
    {
      "epoch": 0.2819775403998904,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008999937274039087,
      "loss": 5.4585,
      "step": 2059
    },
    {
      "epoch": 0.2821144891810463,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008999935092433919,
      "loss": 5.468,
      "step": 2060
    },
    {
      "epoch": 0.28225143796220215,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008999932873536716,
      "loss": 5.4894,
      "step": 2061
    },
    {
      "epoch": 0.282388386743358,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008999930617347499,
      "loss": 5.5384,
      "step": 2062
    },
    {
      "epoch": 0.2825253355245138,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008999928323866283,
      "loss": 5.5056,
      "step": 2063
    },
    {
      "epoch": 0.2826622843056697,
      "grad_norm": 0.78125,
      "learning_rate": 0.000899992599309309,
      "loss": 5.5407,
      "step": 2064
    },
    {
      "epoch": 0.28279923308682553,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008999923625027937,
      "loss": 5.5479,
      "step": 2065
    },
    {
      "epoch": 0.28293618186798136,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008999921219670845,
      "loss": 5.4836,
      "step": 2066
    },
    {
      "epoch": 0.28307313064913725,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008999918777021833,
      "loss": 5.5212,
      "step": 2067
    },
    {
      "epoch": 0.2832100794302931,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008999916297080923,
      "loss": 5.5245,
      "step": 2068
    },
    {
      "epoch": 0.2833470282114489,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008999913779848133,
      "loss": 5.4895,
      "step": 2069
    },
    {
      "epoch": 0.28348397699260475,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008999911225323486,
      "loss": 5.5616,
      "step": 2070
    },
    {
      "epoch": 0.28362092577376063,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008999908633507002,
      "loss": 5.4854,
      "step": 2071
    },
    {
      "epoch": 0.28375787455491647,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008999906004398704,
      "loss": 5.4884,
      "step": 2072
    },
    {
      "epoch": 0.2838948233360723,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008999903337998613,
      "loss": 5.5427,
      "step": 2073
    },
    {
      "epoch": 0.28403177211722813,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008999900634306749,
      "loss": 5.5016,
      "step": 2074
    },
    {
      "epoch": 0.284168720898384,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008999897893323137,
      "loss": 5.5156,
      "step": 2075
    },
    {
      "epoch": 0.28430566967953985,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008999895115047798,
      "loss": 5.5081,
      "step": 2076
    },
    {
      "epoch": 0.2844426184606957,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008999892299480757,
      "loss": 5.4199,
      "step": 2077
    },
    {
      "epoch": 0.28457956724185157,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008999889446622036,
      "loss": 5.4388,
      "step": 2078
    },
    {
      "epoch": 0.2847165160230074,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008999886556471659,
      "loss": 5.5067,
      "step": 2079
    },
    {
      "epoch": 0.28485346480416324,
      "grad_norm": 0.625,
      "learning_rate": 0.000899988362902965,
      "loss": 5.5141,
      "step": 2080
    },
    {
      "epoch": 0.28499041358531907,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008999880664296034,
      "loss": 5.4241,
      "step": 2081
    },
    {
      "epoch": 0.28512736236647496,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008999877662270833,
      "loss": 5.4971,
      "step": 2082
    },
    {
      "epoch": 0.2852643111476308,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008999874622954073,
      "loss": 5.5589,
      "step": 2083
    },
    {
      "epoch": 0.2854012599287866,
      "grad_norm": 0.640625,
      "learning_rate": 0.000899987154634578,
      "loss": 5.4732,
      "step": 2084
    },
    {
      "epoch": 0.2855382087099425,
      "grad_norm": 0.5390625,
      "learning_rate": 0.000899986843244598,
      "loss": 5.4666,
      "step": 2085
    },
    {
      "epoch": 0.28567515749109834,
      "grad_norm": 0.625,
      "learning_rate": 0.0008999865281254699,
      "loss": 5.4854,
      "step": 2086
    },
    {
      "epoch": 0.2858121062722542,
      "grad_norm": 0.640625,
      "learning_rate": 0.000899986209277196,
      "loss": 5.4537,
      "step": 2087
    },
    {
      "epoch": 0.28594905505341,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008999858866997794,
      "loss": 5.4625,
      "step": 2088
    },
    {
      "epoch": 0.2860860038345659,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008999855603932225,
      "loss": 5.4645,
      "step": 2089
    },
    {
      "epoch": 0.2862229526157217,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008999852303575278,
      "loss": 5.4637,
      "step": 2090
    },
    {
      "epoch": 0.28635990139687756,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008999848965926985,
      "loss": 5.5101,
      "step": 2091
    },
    {
      "epoch": 0.2864968501780334,
      "grad_norm": 0.5,
      "learning_rate": 0.000899984559098737,
      "loss": 5.468,
      "step": 2092
    },
    {
      "epoch": 0.2866337989591893,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008999842178756463,
      "loss": 5.3984,
      "step": 2093
    },
    {
      "epoch": 0.2867707477403451,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008999838729234292,
      "loss": 5.4961,
      "step": 2094
    },
    {
      "epoch": 0.28690769652150094,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008999835242420884,
      "loss": 5.5536,
      "step": 2095
    },
    {
      "epoch": 0.28704464530265683,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000899983171831627,
      "loss": 5.489,
      "step": 2096
    },
    {
      "epoch": 0.28718159408381266,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008999828156920478,
      "loss": 5.4851,
      "step": 2097
    },
    {
      "epoch": 0.2873185428649685,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008999824558233537,
      "loss": 5.4999,
      "step": 2098
    },
    {
      "epoch": 0.28745549164612433,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008999820922255479,
      "loss": 5.5177,
      "step": 2099
    },
    {
      "epoch": 0.2875924404272802,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008999817248986332,
      "loss": 5.4708,
      "step": 2100
    },
    {
      "epoch": 0.28772938920843605,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008999813538426128,
      "loss": 5.4658,
      "step": 2101
    },
    {
      "epoch": 0.2878663379895919,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008999809790574896,
      "loss": 5.4971,
      "step": 2102
    },
    {
      "epoch": 0.2880032867707477,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008999806005432669,
      "loss": 5.5042,
      "step": 2103
    },
    {
      "epoch": 0.2881402355519036,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008999802182999477,
      "loss": 5.4941,
      "step": 2104
    },
    {
      "epoch": 0.28827718433305943,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008999798323275351,
      "loss": 5.5236,
      "step": 2105
    },
    {
      "epoch": 0.28841413311421527,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008999794426260325,
      "loss": 5.5181,
      "step": 2106
    },
    {
      "epoch": 0.28855108189537115,
      "grad_norm": 0.5625,
      "learning_rate": 0.000899979049195443,
      "loss": 5.5413,
      "step": 2107
    },
    {
      "epoch": 0.288688030676527,
      "grad_norm": 0.8125,
      "learning_rate": 0.0008999786520357701,
      "loss": 5.472,
      "step": 2108
    },
    {
      "epoch": 0.2888249794576828,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008999782511470167,
      "loss": 5.539,
      "step": 2109
    },
    {
      "epoch": 0.28896192823883865,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008999778465291862,
      "loss": 5.5095,
      "step": 2110
    },
    {
      "epoch": 0.28909887701999454,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008999774381822822,
      "loss": 5.4502,
      "step": 2111
    },
    {
      "epoch": 0.28923582580115037,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008999770261063079,
      "loss": 5.4822,
      "step": 2112
    },
    {
      "epoch": 0.2893727745823062,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008999766103012667,
      "loss": 5.4909,
      "step": 2113
    },
    {
      "epoch": 0.2895097233634621,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008999761907671623,
      "loss": 5.4305,
      "step": 2114
    },
    {
      "epoch": 0.2896466721446179,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008999757675039979,
      "loss": 5.4786,
      "step": 2115
    },
    {
      "epoch": 0.28978362092577375,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000899975340511777,
      "loss": 5.5196,
      "step": 2116
    },
    {
      "epoch": 0.2899205697069296,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008999749097905034,
      "loss": 5.4063,
      "step": 2117
    },
    {
      "epoch": 0.2900575184880855,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008999744753401804,
      "loss": 5.4747,
      "step": 2118
    },
    {
      "epoch": 0.2901944672692413,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008999740371608116,
      "loss": 5.4967,
      "step": 2119
    },
    {
      "epoch": 0.29033141605039714,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008999735952524008,
      "loss": 5.4989,
      "step": 2120
    },
    {
      "epoch": 0.29046836483155297,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008999731496149517,
      "loss": 5.4745,
      "step": 2121
    },
    {
      "epoch": 0.29060531361270886,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008999727002484678,
      "loss": 5.5113,
      "step": 2122
    },
    {
      "epoch": 0.2907422623938647,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008999722471529529,
      "loss": 5.5213,
      "step": 2123
    },
    {
      "epoch": 0.2908792111750205,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008999717903284108,
      "loss": 5.5187,
      "step": 2124
    },
    {
      "epoch": 0.2910161599561764,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008999713297748452,
      "loss": 5.4712,
      "step": 2125
    },
    {
      "epoch": 0.29115310873733224,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00089997086549226,
      "loss": 5.4265,
      "step": 2126
    },
    {
      "epoch": 0.2912900575184881,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008999703974806591,
      "loss": 5.4623,
      "step": 2127
    },
    {
      "epoch": 0.2914270062996439,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008999699257400462,
      "loss": 5.4086,
      "step": 2128
    },
    {
      "epoch": 0.2915639550807998,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008999694502704254,
      "loss": 5.476,
      "step": 2129
    },
    {
      "epoch": 0.29170090386195563,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008999689710718003,
      "loss": 5.4662,
      "step": 2130
    },
    {
      "epoch": 0.29183785264311146,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008999684881441754,
      "loss": 5.5297,
      "step": 2131
    },
    {
      "epoch": 0.29197480142426735,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008999680014875542,
      "loss": 5.5063,
      "step": 2132
    },
    {
      "epoch": 0.2921117502054232,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008999675111019411,
      "loss": 5.4885,
      "step": 2133
    },
    {
      "epoch": 0.292248698986579,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008999670169873399,
      "loss": 5.509,
      "step": 2134
    },
    {
      "epoch": 0.29238564776773485,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000899966519143755,
      "loss": 5.4001,
      "step": 2135
    },
    {
      "epoch": 0.29252259654889073,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00089996601757119,
      "loss": 5.4821,
      "step": 2136
    },
    {
      "epoch": 0.29265954533004657,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008999655122696496,
      "loss": 5.4891,
      "step": 2137
    },
    {
      "epoch": 0.2927964941112024,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008999650032391379,
      "loss": 5.5149,
      "step": 2138
    },
    {
      "epoch": 0.29293344289235823,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008999644904796588,
      "loss": 5.4824,
      "step": 2139
    },
    {
      "epoch": 0.2930703916735141,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008999639739912168,
      "loss": 5.5309,
      "step": 2140
    },
    {
      "epoch": 0.29320734045466995,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008999634537738163,
      "loss": 5.5217,
      "step": 2141
    },
    {
      "epoch": 0.2933442892358258,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008999629298274611,
      "loss": 5.4947,
      "step": 2142
    },
    {
      "epoch": 0.29348123801698167,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008999624021521561,
      "loss": 5.4914,
      "step": 2143
    },
    {
      "epoch": 0.2936181867981375,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008999618707479055,
      "loss": 5.4548,
      "step": 2144
    },
    {
      "epoch": 0.29375513557929334,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008999613356147135,
      "loss": 5.4795,
      "step": 2145
    },
    {
      "epoch": 0.29389208436044917,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008999607967525848,
      "loss": 5.4338,
      "step": 2146
    },
    {
      "epoch": 0.29402903314160506,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008999602541615237,
      "loss": 5.4754,
      "step": 2147
    },
    {
      "epoch": 0.2941659819227609,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008999597078415348,
      "loss": 5.4838,
      "step": 2148
    },
    {
      "epoch": 0.2943029307039167,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008999591577926225,
      "loss": 5.4527,
      "step": 2149
    },
    {
      "epoch": 0.2944398794850726,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008999586040147915,
      "loss": 5.488,
      "step": 2150
    },
    {
      "epoch": 0.29457682826622844,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008999580465080463,
      "loss": 5.4944,
      "step": 2151
    },
    {
      "epoch": 0.2947137770473843,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008999574852723916,
      "loss": 5.4841,
      "step": 2152
    },
    {
      "epoch": 0.2948507258285401,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000899956920307832,
      "loss": 5.4437,
      "step": 2153
    },
    {
      "epoch": 0.294987674609696,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008999563516143721,
      "loss": 5.4214,
      "step": 2154
    },
    {
      "epoch": 0.2951246233908518,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008999557791920168,
      "loss": 5.4721,
      "step": 2155
    },
    {
      "epoch": 0.29526157217200766,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008999552030407706,
      "loss": 5.4619,
      "step": 2156
    },
    {
      "epoch": 0.2953985209531635,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008999546231606386,
      "loss": 5.5573,
      "step": 2157
    },
    {
      "epoch": 0.2955354697343194,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008999540395516253,
      "loss": 5.5063,
      "step": 2158
    },
    {
      "epoch": 0.2956724185154752,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008999534522137357,
      "loss": 5.4287,
      "step": 2159
    },
    {
      "epoch": 0.29580936729663104,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008999528611469746,
      "loss": 5.4698,
      "step": 2160
    },
    {
      "epoch": 0.29594631607778693,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000899952266351347,
      "loss": 5.5127,
      "step": 2161
    },
    {
      "epoch": 0.29608326485894276,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008999516678268577,
      "loss": 5.3837,
      "step": 2162
    },
    {
      "epoch": 0.2962202136400986,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008999510655735116,
      "loss": 5.4761,
      "step": 2163
    },
    {
      "epoch": 0.2963571624212544,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008999504595913138,
      "loss": 5.4884,
      "step": 2164
    },
    {
      "epoch": 0.2964941112024103,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008999498498802696,
      "loss": 5.5111,
      "step": 2165
    },
    {
      "epoch": 0.29663105998356615,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008999492364403835,
      "loss": 5.412,
      "step": 2166
    },
    {
      "epoch": 0.296768008764722,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000899948619271661,
      "loss": 5.4368,
      "step": 2167
    },
    {
      "epoch": 0.29690495754587787,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000899947998374107,
      "loss": 5.4365,
      "step": 2168
    },
    {
      "epoch": 0.2970419063270337,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008999473737477268,
      "loss": 5.3911,
      "step": 2169
    },
    {
      "epoch": 0.29717885510818953,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008999467453925254,
      "loss": 5.5144,
      "step": 2170
    },
    {
      "epoch": 0.29731580388934536,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008999461133085082,
      "loss": 5.5349,
      "step": 2171
    },
    {
      "epoch": 0.29745275267050125,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008999454774956802,
      "loss": 5.4189,
      "step": 2172
    },
    {
      "epoch": 0.2975897014516571,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008999448379540469,
      "loss": 5.5497,
      "step": 2173
    },
    {
      "epoch": 0.2977266502328129,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008999441946836136,
      "loss": 5.4945,
      "step": 2174
    },
    {
      "epoch": 0.29786359901396875,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008999435476843854,
      "loss": 5.4658,
      "step": 2175
    },
    {
      "epoch": 0.29800054779512464,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008999428969563678,
      "loss": 5.4234,
      "step": 2176
    },
    {
      "epoch": 0.29813749657628047,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008999422424995664,
      "loss": 5.5108,
      "step": 2177
    },
    {
      "epoch": 0.2982744453574363,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008999415843139862,
      "loss": 5.4435,
      "step": 2178
    },
    {
      "epoch": 0.2984113941385922,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008999409223996329,
      "loss": 5.5124,
      "step": 2179
    },
    {
      "epoch": 0.298548342919748,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000899940256756512,
      "loss": 5.4668,
      "step": 2180
    },
    {
      "epoch": 0.29868529170090385,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008999395873846291,
      "loss": 5.4025,
      "step": 2181
    },
    {
      "epoch": 0.2988222404820597,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008999389142839895,
      "loss": 5.4293,
      "step": 2182
    },
    {
      "epoch": 0.2989591892632156,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008999382374545989,
      "loss": 5.3769,
      "step": 2183
    },
    {
      "epoch": 0.2990961380443714,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008999375568964629,
      "loss": 5.437,
      "step": 2184
    },
    {
      "epoch": 0.29923308682552724,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0008999368726095873,
      "loss": 5.5136,
      "step": 2185
    },
    {
      "epoch": 0.2993700356066831,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008999361845939774,
      "loss": 5.4695,
      "step": 2186
    },
    {
      "epoch": 0.29950698438783896,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008999354928496393,
      "loss": 5.4782,
      "step": 2187
    },
    {
      "epoch": 0.2996439331689948,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008999347973765785,
      "loss": 5.4389,
      "step": 2188
    },
    {
      "epoch": 0.2997808819501506,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008999340981748009,
      "loss": 5.4669,
      "step": 2189
    },
    {
      "epoch": 0.2999178307313065,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000899933395244312,
      "loss": 5.4488,
      "step": 2190
    },
    {
      "epoch": 0.30005477951246234,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008999326885851181,
      "loss": 5.4551,
      "step": 2191
    },
    {
      "epoch": 0.3001917282936182,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008999319781972246,
      "loss": 5.4816,
      "step": 2192
    },
    {
      "epoch": 0.300328677074774,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008999312640806375,
      "loss": 5.4542,
      "step": 2193
    },
    {
      "epoch": 0.3004656258559299,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008999305462353631,
      "loss": 5.3859,
      "step": 2194
    },
    {
      "epoch": 0.30060257463708573,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008999298246614069,
      "loss": 5.4429,
      "step": 2195
    },
    {
      "epoch": 0.30073952341824156,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000899929099358775,
      "loss": 5.53,
      "step": 2196
    },
    {
      "epoch": 0.30087647219939745,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008999283703274734,
      "loss": 5.4218,
      "step": 2197
    },
    {
      "epoch": 0.3010134209805533,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008999276375675082,
      "loss": 5.5279,
      "step": 2198
    },
    {
      "epoch": 0.3011503697617091,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008999269010788856,
      "loss": 5.5191,
      "step": 2199
    },
    {
      "epoch": 0.30128731854286495,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008999261608616115,
      "loss": 5.4673,
      "step": 2200
    },
    {
      "epoch": 0.30142426732402083,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008999254169156921,
      "loss": 5.4569,
      "step": 2201
    },
    {
      "epoch": 0.30156121610517667,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008999246692411334,
      "loss": 5.4904,
      "step": 2202
    },
    {
      "epoch": 0.3016981648863325,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008999239178379419,
      "loss": 5.4579,
      "step": 2203
    },
    {
      "epoch": 0.3018351136674884,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008999231627061236,
      "loss": 5.4626,
      "step": 2204
    },
    {
      "epoch": 0.3019720624486442,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008999224038456849,
      "loss": 5.4479,
      "step": 2205
    },
    {
      "epoch": 0.30210901122980005,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000899921641256632,
      "loss": 5.4775,
      "step": 2206
    },
    {
      "epoch": 0.3022459600109559,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008999208749389713,
      "loss": 5.4586,
      "step": 2207
    },
    {
      "epoch": 0.30238290879211177,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008999201048927091,
      "loss": 5.4894,
      "step": 2208
    },
    {
      "epoch": 0.3025198575732676,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008999193311178518,
      "loss": 5.3206,
      "step": 2209
    },
    {
      "epoch": 0.30265680635442344,
      "grad_norm": 0.5,
      "learning_rate": 0.0008999185536144057,
      "loss": 5.422,
      "step": 2210
    },
    {
      "epoch": 0.30279375513557927,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008999177723823775,
      "loss": 5.4155,
      "step": 2211
    },
    {
      "epoch": 0.30293070391673516,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008999169874217734,
      "loss": 5.4668,
      "step": 2212
    },
    {
      "epoch": 0.303067652697891,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008999161987326,
      "loss": 5.3753,
      "step": 2213
    },
    {
      "epoch": 0.3032046014790468,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008999154063148637,
      "loss": 5.3688,
      "step": 2214
    },
    {
      "epoch": 0.3033415502602027,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008999146101685715,
      "loss": 5.5083,
      "step": 2215
    },
    {
      "epoch": 0.30347849904135854,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008999138102937295,
      "loss": 5.408,
      "step": 2216
    },
    {
      "epoch": 0.3036154478225144,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008999130066903447,
      "loss": 5.389,
      "step": 2217
    },
    {
      "epoch": 0.3037523966036702,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008999121993584235,
      "loss": 5.452,
      "step": 2218
    },
    {
      "epoch": 0.3038893453848261,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008999113882979728,
      "loss": 5.3907,
      "step": 2219
    },
    {
      "epoch": 0.3040262941659819,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008999105735089992,
      "loss": 5.4747,
      "step": 2220
    },
    {
      "epoch": 0.30416324294713776,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008999097549915094,
      "loss": 5.5102,
      "step": 2221
    },
    {
      "epoch": 0.30430019172829365,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008999089327455103,
      "loss": 5.4831,
      "step": 2222
    },
    {
      "epoch": 0.3044371405094495,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008999081067710086,
      "loss": 5.4753,
      "step": 2223
    },
    {
      "epoch": 0.3045740892906053,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008999072770680112,
      "loss": 5.4335,
      "step": 2224
    },
    {
      "epoch": 0.30471103807176114,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008999064436365251,
      "loss": 5.431,
      "step": 2225
    },
    {
      "epoch": 0.30484798685291703,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008999056064765571,
      "loss": 5.4617,
      "step": 2226
    },
    {
      "epoch": 0.30498493563407286,
      "grad_norm": 0.578125,
      "learning_rate": 0.000899904765588114,
      "loss": 5.462,
      "step": 2227
    },
    {
      "epoch": 0.3051218844152287,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000899903920971203,
      "loss": 5.4496,
      "step": 2228
    },
    {
      "epoch": 0.3052588331963845,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008999030726258311,
      "loss": 5.4723,
      "step": 2229
    },
    {
      "epoch": 0.3053957819775404,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000899902220552005,
      "loss": 5.418,
      "step": 2230
    },
    {
      "epoch": 0.30553273075869625,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008999013647497322,
      "loss": 5.4386,
      "step": 2231
    },
    {
      "epoch": 0.3056696795398521,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008999005052190194,
      "loss": 5.4591,
      "step": 2232
    },
    {
      "epoch": 0.30580662832100797,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008998996419598739,
      "loss": 5.4325,
      "step": 2233
    },
    {
      "epoch": 0.3059435771021638,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000899898774972303,
      "loss": 5.4666,
      "step": 2234
    },
    {
      "epoch": 0.30608052588331963,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008998979042563137,
      "loss": 5.4379,
      "step": 2235
    },
    {
      "epoch": 0.30621747466447546,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008998970298119133,
      "loss": 5.4109,
      "step": 2236
    },
    {
      "epoch": 0.30635442344563135,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000899896151639109,
      "loss": 5.4623,
      "step": 2237
    },
    {
      "epoch": 0.3064913722267872,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008998952697379081,
      "loss": 5.4314,
      "step": 2238
    },
    {
      "epoch": 0.306628321007943,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008998943841083178,
      "loss": 5.503,
      "step": 2239
    },
    {
      "epoch": 0.3067652697890989,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008998934947503455,
      "loss": 5.4615,
      "step": 2240
    },
    {
      "epoch": 0.30690221857025474,
      "grad_norm": 0.625,
      "learning_rate": 0.0008998926016639989,
      "loss": 5.4642,
      "step": 2241
    },
    {
      "epoch": 0.30703916735141057,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008998917048492848,
      "loss": 5.3618,
      "step": 2242
    },
    {
      "epoch": 0.3071761161325664,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008998908043062111,
      "loss": 5.4357,
      "step": 2243
    },
    {
      "epoch": 0.3073130649137223,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008998899000347849,
      "loss": 5.4285,
      "step": 2244
    },
    {
      "epoch": 0.3074500136948781,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008998889920350142,
      "loss": 5.3958,
      "step": 2245
    },
    {
      "epoch": 0.30758696247603395,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008998880803069059,
      "loss": 5.4499,
      "step": 2246
    },
    {
      "epoch": 0.3077239112571898,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008998871648504681,
      "loss": 5.4539,
      "step": 2247
    },
    {
      "epoch": 0.3078608600383457,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008998862456657082,
      "loss": 5.4479,
      "step": 2248
    },
    {
      "epoch": 0.3079978088195015,
      "grad_norm": 0.8125,
      "learning_rate": 0.0008998853227526337,
      "loss": 5.4079,
      "step": 2249
    },
    {
      "epoch": 0.30813475760065734,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008998843961112522,
      "loss": 5.4864,
      "step": 2250
    },
    {
      "epoch": 0.3082717063818132,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008998834657415716,
      "loss": 5.4373,
      "step": 2251
    },
    {
      "epoch": 0.30840865516296906,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008998825316435995,
      "loss": 5.4295,
      "step": 2252
    },
    {
      "epoch": 0.3085456039441249,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008998815938173437,
      "loss": 5.4181,
      "step": 2253
    },
    {
      "epoch": 0.3086825527252807,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008998806522628119,
      "loss": 5.4189,
      "step": 2254
    },
    {
      "epoch": 0.3088195015064366,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008998797069800121,
      "loss": 5.4376,
      "step": 2255
    },
    {
      "epoch": 0.30895645028759244,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008998787579689518,
      "loss": 5.3144,
      "step": 2256
    },
    {
      "epoch": 0.3090933990687483,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000899877805229639,
      "loss": 5.375,
      "step": 2257
    },
    {
      "epoch": 0.30923034784990416,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008998768487620818,
      "loss": 5.355,
      "step": 2258
    },
    {
      "epoch": 0.30936729663106,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008998758885662877,
      "loss": 5.4038,
      "step": 2259
    },
    {
      "epoch": 0.30950424541221583,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008998749246422651,
      "loss": 5.484,
      "step": 2260
    },
    {
      "epoch": 0.30964119419337166,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008998739569900218,
      "loss": 5.4331,
      "step": 2261
    },
    {
      "epoch": 0.30977814297452755,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008998729856095657,
      "loss": 5.4103,
      "step": 2262
    },
    {
      "epoch": 0.3099150917556834,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008998720105009051,
      "loss": 5.4615,
      "step": 2263
    },
    {
      "epoch": 0.3100520405368392,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008998710316640479,
      "loss": 5.3828,
      "step": 2264
    },
    {
      "epoch": 0.31018898931799505,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008998700490990022,
      "loss": 5.3622,
      "step": 2265
    },
    {
      "epoch": 0.31032593809915093,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008998690628057763,
      "loss": 5.4334,
      "step": 2266
    },
    {
      "epoch": 0.31046288688030677,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008998680727843782,
      "loss": 5.4095,
      "step": 2267
    },
    {
      "epoch": 0.3105998356614626,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008998670790348162,
      "loss": 5.4879,
      "step": 2268
    },
    {
      "epoch": 0.3107367844426185,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008998660815570986,
      "loss": 5.4484,
      "step": 2269
    },
    {
      "epoch": 0.3108737332237743,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008998650803512336,
      "loss": 5.392,
      "step": 2270
    },
    {
      "epoch": 0.31101068200493015,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008998640754172293,
      "loss": 5.4076,
      "step": 2271
    },
    {
      "epoch": 0.311147630786086,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008998630667550943,
      "loss": 5.4017,
      "step": 2272
    },
    {
      "epoch": 0.31128457956724187,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008998620543648369,
      "loss": 5.3635,
      "step": 2273
    },
    {
      "epoch": 0.3114215283483977,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008998610382464654,
      "loss": 5.4821,
      "step": 2274
    },
    {
      "epoch": 0.31155847712955353,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008998600183999884,
      "loss": 5.3926,
      "step": 2275
    },
    {
      "epoch": 0.31169542591070937,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008998589948254141,
      "loss": 5.3793,
      "step": 2276
    },
    {
      "epoch": 0.31183237469186526,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008998579675227512,
      "loss": 5.3863,
      "step": 2277
    },
    {
      "epoch": 0.3119693234730211,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008998569364920081,
      "loss": 5.3832,
      "step": 2278
    },
    {
      "epoch": 0.3121062722541769,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008998559017331933,
      "loss": 5.5255,
      "step": 2279
    },
    {
      "epoch": 0.3122432210353328,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008998548632463155,
      "loss": 5.3949,
      "step": 2280
    },
    {
      "epoch": 0.31238016981648864,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008998538210313833,
      "loss": 5.3853,
      "step": 2281
    },
    {
      "epoch": 0.31251711859764447,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008998527750884053,
      "loss": 5.457,
      "step": 2282
    },
    {
      "epoch": 0.3126540673788003,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008998517254173901,
      "loss": 5.3356,
      "step": 2283
    },
    {
      "epoch": 0.3127910161599562,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008998506720183466,
      "loss": 5.3684,
      "step": 2284
    },
    {
      "epoch": 0.312927964941112,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008998496148912832,
      "loss": 5.4314,
      "step": 2285
    },
    {
      "epoch": 0.31306491372226786,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008998485540362089,
      "loss": 5.4109,
      "step": 2286
    },
    {
      "epoch": 0.31320186250342374,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008998474894531324,
      "loss": 5.3991,
      "step": 2287
    },
    {
      "epoch": 0.3133388112845796,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008998464211420626,
      "loss": 5.4859,
      "step": 2288
    },
    {
      "epoch": 0.3134757600657354,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008998453491030084,
      "loss": 5.4728,
      "step": 2289
    },
    {
      "epoch": 0.31361270884689124,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008998442733359785,
      "loss": 5.4224,
      "step": 2290
    },
    {
      "epoch": 0.31374965762804713,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000899843193840982,
      "loss": 5.4427,
      "step": 2291
    },
    {
      "epoch": 0.31388660640920296,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008998421106180276,
      "loss": 5.4171,
      "step": 2292
    },
    {
      "epoch": 0.3140235551903588,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008998410236671245,
      "loss": 5.4204,
      "step": 2293
    },
    {
      "epoch": 0.3141605039715146,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008998399329882817,
      "loss": 5.3655,
      "step": 2294
    },
    {
      "epoch": 0.3142974527526705,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008998388385815083,
      "loss": 5.4026,
      "step": 2295
    },
    {
      "epoch": 0.31443440153382635,
      "grad_norm": 0.71875,
      "learning_rate": 0.000899837740446813,
      "loss": 5.3957,
      "step": 2296
    },
    {
      "epoch": 0.3145713503149822,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008998366385842054,
      "loss": 5.4768,
      "step": 2297
    },
    {
      "epoch": 0.31470829909613807,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008998355329936943,
      "loss": 5.4255,
      "step": 2298
    },
    {
      "epoch": 0.3148452478772939,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008998344236752888,
      "loss": 5.468,
      "step": 2299
    },
    {
      "epoch": 0.31498219665844973,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0008998333106289984,
      "loss": 5.4089,
      "step": 2300
    },
    {
      "epoch": 0.31511914543960556,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008998321938548321,
      "loss": 5.428,
      "step": 2301
    },
    {
      "epoch": 0.31525609422076145,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008998310733527993,
      "loss": 5.4032,
      "step": 2302
    },
    {
      "epoch": 0.3153930430019173,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008998299491229092,
      "loss": 5.4719,
      "step": 2303
    },
    {
      "epoch": 0.3155299917830731,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000899828821165171,
      "loss": 5.452,
      "step": 2304
    },
    {
      "epoch": 0.315666940564229,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008998276894795943,
      "loss": 5.4545,
      "step": 2305
    },
    {
      "epoch": 0.31580388934538484,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008998265540661882,
      "loss": 5.4619,
      "step": 2306
    },
    {
      "epoch": 0.31594083812654067,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008998254149249624,
      "loss": 5.4329,
      "step": 2307
    },
    {
      "epoch": 0.3160777869076965,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008998242720559261,
      "loss": 5.4231,
      "step": 2308
    },
    {
      "epoch": 0.3162147356888524,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000899823125459089,
      "loss": 5.4466,
      "step": 2309
    },
    {
      "epoch": 0.3163516844700082,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008998219751344604,
      "loss": 5.3893,
      "step": 2310
    },
    {
      "epoch": 0.31648863325116405,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008998208210820499,
      "loss": 5.4184,
      "step": 2311
    },
    {
      "epoch": 0.3166255820323199,
      "grad_norm": 1.078125,
      "learning_rate": 0.000899819663301867,
      "loss": 5.4039,
      "step": 2312
    },
    {
      "epoch": 0.3167625308134758,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008998185017939214,
      "loss": 5.4952,
      "step": 2313
    },
    {
      "epoch": 0.3168994795946316,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008998173365582227,
      "loss": 5.46,
      "step": 2314
    },
    {
      "epoch": 0.31703642837578744,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008998161675947806,
      "loss": 5.4165,
      "step": 2315
    },
    {
      "epoch": 0.3171733771569433,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008998149949036047,
      "loss": 5.3911,
      "step": 2316
    },
    {
      "epoch": 0.31731032593809916,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008998138184847048,
      "loss": 5.4791,
      "step": 2317
    },
    {
      "epoch": 0.317447274719255,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008998126383380906,
      "loss": 5.3835,
      "step": 2318
    },
    {
      "epoch": 0.3175842235004108,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008998114544637718,
      "loss": 5.4546,
      "step": 2319
    },
    {
      "epoch": 0.3177211722815667,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008998102668617582,
      "loss": 5.4276,
      "step": 2320
    },
    {
      "epoch": 0.31785812106272254,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008998090755320599,
      "loss": 5.4551,
      "step": 2321
    },
    {
      "epoch": 0.3179950698438784,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008998078804746866,
      "loss": 5.4144,
      "step": 2322
    },
    {
      "epoch": 0.31813201862503426,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008998066816896483,
      "loss": 5.4408,
      "step": 2323
    },
    {
      "epoch": 0.3182689674061901,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008998054791769546,
      "loss": 5.3317,
      "step": 2324
    },
    {
      "epoch": 0.3184059161873459,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008998042729366159,
      "loss": 5.4526,
      "step": 2325
    },
    {
      "epoch": 0.31854286496850176,
      "grad_norm": 0.5625,
      "learning_rate": 0.000899803062968642,
      "loss": 5.3922,
      "step": 2326
    },
    {
      "epoch": 0.31867981374965765,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008998018492730427,
      "loss": 5.4358,
      "step": 2327
    },
    {
      "epoch": 0.3188167625308135,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008998006318498285,
      "loss": 5.4449,
      "step": 2328
    },
    {
      "epoch": 0.3189537113119693,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008997994106990093,
      "loss": 5.4425,
      "step": 2329
    },
    {
      "epoch": 0.31909066009312514,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008997981858205952,
      "loss": 5.399,
      "step": 2330
    },
    {
      "epoch": 0.31922760887428103,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008997969572145963,
      "loss": 5.411,
      "step": 2331
    },
    {
      "epoch": 0.31936455765543686,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008997957248810228,
      "loss": 5.4784,
      "step": 2332
    },
    {
      "epoch": 0.3195015064365927,
      "grad_norm": 0.5,
      "learning_rate": 0.0008997944888198849,
      "loss": 5.388,
      "step": 2333
    },
    {
      "epoch": 0.3196384552177486,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008997932490311931,
      "loss": 5.3872,
      "step": 2334
    },
    {
      "epoch": 0.3197754039989044,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008997920055149573,
      "loss": 5.3906,
      "step": 2335
    },
    {
      "epoch": 0.31991235278006025,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008997907582711881,
      "loss": 5.4107,
      "step": 2336
    },
    {
      "epoch": 0.3200493015612161,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008997895072998957,
      "loss": 5.4261,
      "step": 2337
    },
    {
      "epoch": 0.32018625034237197,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008997882526010903,
      "loss": 5.4454,
      "step": 2338
    },
    {
      "epoch": 0.3203231991235278,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008997869941747827,
      "loss": 5.4149,
      "step": 2339
    },
    {
      "epoch": 0.32046014790468363,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008997857320209828,
      "loss": 5.3559,
      "step": 2340
    },
    {
      "epoch": 0.3205970966858395,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008997844661397016,
      "loss": 5.405,
      "step": 2341
    },
    {
      "epoch": 0.32073404546699535,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008997831965309493,
      "loss": 5.394,
      "step": 2342
    },
    {
      "epoch": 0.3208709942481512,
      "grad_norm": 0.5,
      "learning_rate": 0.0008997819231947364,
      "loss": 5.4147,
      "step": 2343
    },
    {
      "epoch": 0.321007943029307,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008997806461310735,
      "loss": 5.4252,
      "step": 2344
    },
    {
      "epoch": 0.3211448918104629,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008997793653399713,
      "loss": 5.4427,
      "step": 2345
    },
    {
      "epoch": 0.32128184059161874,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008997780808214404,
      "loss": 5.4236,
      "step": 2346
    },
    {
      "epoch": 0.32141878937277457,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008997767925754914,
      "loss": 5.3837,
      "step": 2347
    },
    {
      "epoch": 0.3215557381539304,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008997755006021347,
      "loss": 5.3651,
      "step": 2348
    },
    {
      "epoch": 0.3216926869350863,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008997742049013814,
      "loss": 5.3632,
      "step": 2349
    },
    {
      "epoch": 0.3218296357162421,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000899772905473242,
      "loss": 5.3938,
      "step": 2350
    },
    {
      "epoch": 0.32196658449739796,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008997716023177275,
      "loss": 5.3876,
      "step": 2351
    },
    {
      "epoch": 0.32210353327855384,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008997702954348485,
      "loss": 5.3863,
      "step": 2352
    },
    {
      "epoch": 0.3222404820597097,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000899768984824616,
      "loss": 5.4736,
      "step": 2353
    },
    {
      "epoch": 0.3223774308408655,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008997676704870406,
      "loss": 5.3911,
      "step": 2354
    },
    {
      "epoch": 0.32251437962202134,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008997663524221333,
      "loss": 5.3205,
      "step": 2355
    },
    {
      "epoch": 0.32265132840317723,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000899765030629905,
      "loss": 5.3864,
      "step": 2356
    },
    {
      "epoch": 0.32278827718433306,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008997637051103669,
      "loss": 5.4137,
      "step": 2357
    },
    {
      "epoch": 0.3229252259654889,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008997623758635297,
      "loss": 5.356,
      "step": 2358
    },
    {
      "epoch": 0.3230621747466448,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008997610428894046,
      "loss": 5.3253,
      "step": 2359
    },
    {
      "epoch": 0.3231991235278006,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008997597061880026,
      "loss": 5.401,
      "step": 2360
    },
    {
      "epoch": 0.32333607230895645,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008997583657593346,
      "loss": 5.3983,
      "step": 2361
    },
    {
      "epoch": 0.3234730210901123,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008997570216034119,
      "loss": 5.3513,
      "step": 2362
    },
    {
      "epoch": 0.32360996987126817,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008997556737202456,
      "loss": 5.3506,
      "step": 2363
    },
    {
      "epoch": 0.323746918652424,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008997543221098469,
      "loss": 5.342,
      "step": 2364
    },
    {
      "epoch": 0.32388386743357983,
      "grad_norm": 0.7421875,
      "learning_rate": 0.000899752966772227,
      "loss": 5.4356,
      "step": 2365
    },
    {
      "epoch": 0.32402081621473566,
      "grad_norm": 0.5703125,
      "learning_rate": 0.000899751607707397,
      "loss": 5.3732,
      "step": 2366
    },
    {
      "epoch": 0.32415776499589155,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008997502449153683,
      "loss": 5.387,
      "step": 2367
    },
    {
      "epoch": 0.3242947137770474,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008997488783961522,
      "loss": 5.3795,
      "step": 2368
    },
    {
      "epoch": 0.3244316625582032,
      "grad_norm": 0.8125,
      "learning_rate": 0.0008997475081497598,
      "loss": 5.4826,
      "step": 2369
    },
    {
      "epoch": 0.3245686113393591,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008997461341762028,
      "loss": 5.3405,
      "step": 2370
    },
    {
      "epoch": 0.32470556012051494,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008997447564754923,
      "loss": 5.4,
      "step": 2371
    },
    {
      "epoch": 0.32484250890167077,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008997433750476399,
      "loss": 5.3922,
      "step": 2372
    },
    {
      "epoch": 0.3249794576828266,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000899741989892657,
      "loss": 5.4687,
      "step": 2373
    },
    {
      "epoch": 0.3251164064639825,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000899740601010555,
      "loss": 5.4162,
      "step": 2374
    },
    {
      "epoch": 0.3252533552451383,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008997392084013455,
      "loss": 5.2952,
      "step": 2375
    },
    {
      "epoch": 0.32539030402629415,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008997378120650401,
      "loss": 5.4357,
      "step": 2376
    },
    {
      "epoch": 0.32552725280745004,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008997364120016503,
      "loss": 5.4096,
      "step": 2377
    },
    {
      "epoch": 0.3256642015886059,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008997350082111875,
      "loss": 5.3435,
      "step": 2378
    },
    {
      "epoch": 0.3258011503697617,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008997336006936636,
      "loss": 5.4012,
      "step": 2379
    },
    {
      "epoch": 0.32593809915091754,
      "grad_norm": 0.875,
      "learning_rate": 0.0008997321894490902,
      "loss": 5.3996,
      "step": 2380
    },
    {
      "epoch": 0.3260750479320734,
      "grad_norm": 1.0625,
      "learning_rate": 0.000899730774477479,
      "loss": 5.3838,
      "step": 2381
    },
    {
      "epoch": 0.32621199671322926,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008997293557788417,
      "loss": 5.4191,
      "step": 2382
    },
    {
      "epoch": 0.3263489454943851,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008997279333531901,
      "loss": 5.3416,
      "step": 2383
    },
    {
      "epoch": 0.3264858942755409,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008997265072005358,
      "loss": 5.4556,
      "step": 2384
    },
    {
      "epoch": 0.3266228430566968,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008997250773208909,
      "loss": 5.4047,
      "step": 2385
    },
    {
      "epoch": 0.32675979183785264,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008997236437142672,
      "loss": 5.4041,
      "step": 2386
    },
    {
      "epoch": 0.3268967406190085,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008997222063806763,
      "loss": 5.3847,
      "step": 2387
    },
    {
      "epoch": 0.32703368940016436,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008997207653201304,
      "loss": 5.3714,
      "step": 2388
    },
    {
      "epoch": 0.3271706381813202,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008997193205326413,
      "loss": 5.3642,
      "step": 2389
    },
    {
      "epoch": 0.327307586962476,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008997178720182212,
      "loss": 5.3492,
      "step": 2390
    },
    {
      "epoch": 0.32744453574363186,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008997164197768817,
      "loss": 5.4397,
      "step": 2391
    },
    {
      "epoch": 0.32758148452478775,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008997149638086354,
      "loss": 5.3912,
      "step": 2392
    },
    {
      "epoch": 0.3277184333059436,
      "grad_norm": 0.625,
      "learning_rate": 0.0008997135041134937,
      "loss": 5.3704,
      "step": 2393
    },
    {
      "epoch": 0.3278553820870994,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008997120406914693,
      "loss": 5.3511,
      "step": 2394
    },
    {
      "epoch": 0.3279923308682553,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008997105735425739,
      "loss": 5.3022,
      "step": 2395
    },
    {
      "epoch": 0.32812927964941113,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008997091026668198,
      "loss": 5.4608,
      "step": 2396
    },
    {
      "epoch": 0.32826622843056696,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008997076280642193,
      "loss": 5.4186,
      "step": 2397
    },
    {
      "epoch": 0.3284031772117228,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008997061497347844,
      "loss": 5.3854,
      "step": 2398
    },
    {
      "epoch": 0.3285401259928787,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008997046676785275,
      "loss": 5.3539,
      "step": 2399
    },
    {
      "epoch": 0.3286770747740345,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008997031818954609,
      "loss": 5.4526,
      "step": 2400
    },
    {
      "epoch": 0.32881402355519035,
      "grad_norm": 0.625,
      "learning_rate": 0.0008997016923855969,
      "loss": 5.4021,
      "step": 2401
    },
    {
      "epoch": 0.3289509723363462,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008997001991489478,
      "loss": 5.3654,
      "step": 2402
    },
    {
      "epoch": 0.32908792111750207,
      "grad_norm": 0.52734375,
      "learning_rate": 0.000899698702185526,
      "loss": 5.3893,
      "step": 2403
    },
    {
      "epoch": 0.3292248698986579,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008996972014953439,
      "loss": 5.462,
      "step": 2404
    },
    {
      "epoch": 0.32936181867981373,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008996956970784139,
      "loss": 5.3883,
      "step": 2405
    },
    {
      "epoch": 0.3294987674609696,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008996941889347486,
      "loss": 5.4358,
      "step": 2406
    },
    {
      "epoch": 0.32963571624212545,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008996926770643603,
      "loss": 5.3343,
      "step": 2407
    },
    {
      "epoch": 0.3297726650232813,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008996911614672616,
      "loss": 5.3461,
      "step": 2408
    },
    {
      "epoch": 0.3299096138044371,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008996896421434651,
      "loss": 5.4057,
      "step": 2409
    },
    {
      "epoch": 0.330046562585593,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008996881190929835,
      "loss": 5.3718,
      "step": 2410
    },
    {
      "epoch": 0.33018351136674884,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008996865923158292,
      "loss": 5.3013,
      "step": 2411
    },
    {
      "epoch": 0.33032046014790467,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000899685061812015,
      "loss": 5.413,
      "step": 2412
    },
    {
      "epoch": 0.33045740892906056,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008996835275815535,
      "loss": 5.3169,
      "step": 2413
    },
    {
      "epoch": 0.3305943577102164,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008996819896244575,
      "loss": 5.2881,
      "step": 2414
    },
    {
      "epoch": 0.3307313064913722,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008996804479407397,
      "loss": 5.3898,
      "step": 2415
    },
    {
      "epoch": 0.33086825527252806,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008996789025304128,
      "loss": 5.3129,
      "step": 2416
    },
    {
      "epoch": 0.33100520405368394,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008996773533934897,
      "loss": 5.3881,
      "step": 2417
    },
    {
      "epoch": 0.3311421528348398,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008996758005299833,
      "loss": 5.4124,
      "step": 2418
    },
    {
      "epoch": 0.3312791016159956,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008996742439399063,
      "loss": 5.3441,
      "step": 2419
    },
    {
      "epoch": 0.33141605039715144,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008996726836232717,
      "loss": 5.3915,
      "step": 2420
    },
    {
      "epoch": 0.33155299917830733,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008996711195800925,
      "loss": 5.4,
      "step": 2421
    },
    {
      "epoch": 0.33168994795946316,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0008996695518103815,
      "loss": 5.3422,
      "step": 2422
    },
    {
      "epoch": 0.331826896740619,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008996679803141518,
      "loss": 5.3655,
      "step": 2423
    },
    {
      "epoch": 0.3319638455217749,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008996664050914164,
      "loss": 5.4168,
      "step": 2424
    },
    {
      "epoch": 0.3321007943029307,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008996648261421883,
      "loss": 5.3853,
      "step": 2425
    },
    {
      "epoch": 0.33223774308408655,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008996632434664808,
      "loss": 5.3655,
      "step": 2426
    },
    {
      "epoch": 0.3323746918652424,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008996616570643067,
      "loss": 5.4116,
      "step": 2427
    },
    {
      "epoch": 0.33251164064639827,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008996600669356794,
      "loss": 5.4168,
      "step": 2428
    },
    {
      "epoch": 0.3326485894275541,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0008996584730806118,
      "loss": 5.4296,
      "step": 2429
    },
    {
      "epoch": 0.33278553820870993,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008996568754991174,
      "loss": 5.3768,
      "step": 2430
    },
    {
      "epoch": 0.3329224869898658,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008996552741912093,
      "loss": 5.4247,
      "step": 2431
    },
    {
      "epoch": 0.33305943577102165,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008996536691569009,
      "loss": 5.343,
      "step": 2432
    },
    {
      "epoch": 0.3331963845521775,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008996520603962053,
      "loss": 5.3919,
      "step": 2433
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008996504479091359,
      "loss": 5.4765,
      "step": 2434
    },
    {
      "epoch": 0.3334702821144892,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008996488316957062,
      "loss": 5.353,
      "step": 2435
    },
    {
      "epoch": 0.33360723089564503,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008996472117559295,
      "loss": 5.3644,
      "step": 2436
    },
    {
      "epoch": 0.33374417967680087,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000899645588089819,
      "loss": 5.4398,
      "step": 2437
    },
    {
      "epoch": 0.3338811284579567,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008996439606973885,
      "loss": 5.3241,
      "step": 2438
    },
    {
      "epoch": 0.3340180772391126,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008996423295786513,
      "loss": 5.3483,
      "step": 2439
    },
    {
      "epoch": 0.3341550260202684,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000899640694733621,
      "loss": 5.3391,
      "step": 2440
    },
    {
      "epoch": 0.33429197480142425,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008996390561623112,
      "loss": 5.4384,
      "step": 2441
    },
    {
      "epoch": 0.33442892358258014,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008996374138647355,
      "loss": 5.371,
      "step": 2442
    },
    {
      "epoch": 0.33456587236373597,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008996357678409073,
      "loss": 5.3542,
      "step": 2443
    },
    {
      "epoch": 0.3347028211448918,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008996341180908402,
      "loss": 5.3759,
      "step": 2444
    },
    {
      "epoch": 0.33483976992604764,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008996324646145482,
      "loss": 5.3875,
      "step": 2445
    },
    {
      "epoch": 0.3349767187072035,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008996308074120446,
      "loss": 5.348,
      "step": 2446
    },
    {
      "epoch": 0.33511366748835936,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008996291464833436,
      "loss": 5.3642,
      "step": 2447
    },
    {
      "epoch": 0.3352506162695152,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008996274818284586,
      "loss": 5.4178,
      "step": 2448
    },
    {
      "epoch": 0.335387565050671,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008996258134474035,
      "loss": 5.3267,
      "step": 2449
    },
    {
      "epoch": 0.3355245138318269,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008996241413401923,
      "loss": 5.4153,
      "step": 2450
    },
    {
      "epoch": 0.33566146261298274,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008996224655068386,
      "loss": 5.3611,
      "step": 2451
    },
    {
      "epoch": 0.3357984113941386,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008996207859473563,
      "loss": 5.3698,
      "step": 2452
    },
    {
      "epoch": 0.33593536017529446,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008996191026617595,
      "loss": 5.3662,
      "step": 2453
    },
    {
      "epoch": 0.3360723089564503,
      "grad_norm": 0.546875,
      "learning_rate": 0.000899617415650062,
      "loss": 5.4537,
      "step": 2454
    },
    {
      "epoch": 0.3362092577376061,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008996157249122779,
      "loss": 5.3159,
      "step": 2455
    },
    {
      "epoch": 0.33634620651876196,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008996140304484211,
      "loss": 5.361,
      "step": 2456
    },
    {
      "epoch": 0.33648315529991785,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008996123322585057,
      "loss": 5.374,
      "step": 2457
    },
    {
      "epoch": 0.3366201040810737,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008996106303425459,
      "loss": 5.3406,
      "step": 2458
    },
    {
      "epoch": 0.3367570528622295,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008996089247005556,
      "loss": 5.3541,
      "step": 2459
    },
    {
      "epoch": 0.3368940016433854,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008996072153325489,
      "loss": 5.4103,
      "step": 2460
    },
    {
      "epoch": 0.33703095042454123,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008996055022385402,
      "loss": 5.3779,
      "step": 2461
    },
    {
      "epoch": 0.33716789920569706,
      "grad_norm": 0.625,
      "learning_rate": 0.0008996037854185434,
      "loss": 5.4171,
      "step": 2462
    },
    {
      "epoch": 0.3373048479868529,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008996020648725731,
      "loss": 5.3378,
      "step": 2463
    },
    {
      "epoch": 0.3374417967680088,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008996003406006432,
      "loss": 5.3345,
      "step": 2464
    },
    {
      "epoch": 0.3375787455491646,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008995986126027683,
      "loss": 5.384,
      "step": 2465
    },
    {
      "epoch": 0.33771569433032045,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008995968808789623,
      "loss": 5.3541,
      "step": 2466
    },
    {
      "epoch": 0.3378526431114763,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008995951454292399,
      "loss": 5.3902,
      "step": 2467
    },
    {
      "epoch": 0.33798959189263217,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008995934062536157,
      "loss": 5.3276,
      "step": 2468
    },
    {
      "epoch": 0.338126540673788,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008995916633521035,
      "loss": 5.3373,
      "step": 2469
    },
    {
      "epoch": 0.33826348945494383,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008995899167247182,
      "loss": 5.4234,
      "step": 2470
    },
    {
      "epoch": 0.3384004382360997,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008995881663714741,
      "loss": 5.3622,
      "step": 2471
    },
    {
      "epoch": 0.33853738701725555,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008995864122923857,
      "loss": 5.2916,
      "step": 2472
    },
    {
      "epoch": 0.3386743357984114,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008995846544874675,
      "loss": 5.3577,
      "step": 2473
    },
    {
      "epoch": 0.3388112845795672,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008995828929567342,
      "loss": 5.3972,
      "step": 2474
    },
    {
      "epoch": 0.3389482333607231,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008995811277002003,
      "loss": 5.316,
      "step": 2475
    },
    {
      "epoch": 0.33908518214187894,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008995793587178807,
      "loss": 5.3854,
      "step": 2476
    },
    {
      "epoch": 0.33922213092303477,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008995775860097895,
      "loss": 5.2465,
      "step": 2477
    },
    {
      "epoch": 0.33935907970419066,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008995758095759419,
      "loss": 5.3674,
      "step": 2478
    },
    {
      "epoch": 0.3394960284853465,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008995740294163524,
      "loss": 5.2634,
      "step": 2479
    },
    {
      "epoch": 0.3396329772665023,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008995722455310357,
      "loss": 5.323,
      "step": 2480
    },
    {
      "epoch": 0.33976992604765816,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008995704579200069,
      "loss": 5.3226,
      "step": 2481
    },
    {
      "epoch": 0.33990687482881404,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008995686665832803,
      "loss": 5.4594,
      "step": 2482
    },
    {
      "epoch": 0.3400438236099699,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008995668715208712,
      "loss": 5.3502,
      "step": 2483
    },
    {
      "epoch": 0.3401807723911257,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008995650727327944,
      "loss": 5.362,
      "step": 2484
    },
    {
      "epoch": 0.34031772117228154,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008995632702190645,
      "loss": 5.3259,
      "step": 2485
    },
    {
      "epoch": 0.3404546699534374,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008995614639796968,
      "loss": 5.3791,
      "step": 2486
    },
    {
      "epoch": 0.34059161873459326,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008995596540147061,
      "loss": 5.3179,
      "step": 2487
    },
    {
      "epoch": 0.3407285675157491,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008995578403241073,
      "loss": 5.4355,
      "step": 2488
    },
    {
      "epoch": 0.340865516296905,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008995560229079157,
      "loss": 5.3131,
      "step": 2489
    },
    {
      "epoch": 0.3410024650780608,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008995542017661462,
      "loss": 5.3928,
      "step": 2490
    },
    {
      "epoch": 0.34113941385921664,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008995523768988139,
      "loss": 5.3708,
      "step": 2491
    },
    {
      "epoch": 0.3412763626403725,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008995505483059339,
      "loss": 5.4778,
      "step": 2492
    },
    {
      "epoch": 0.34141331142152836,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008995487159875213,
      "loss": 5.3661,
      "step": 2493
    },
    {
      "epoch": 0.3415502602026842,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008995468799435916,
      "loss": 5.357,
      "step": 2494
    },
    {
      "epoch": 0.34168720898384003,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008995450401741595,
      "loss": 5.4071,
      "step": 2495
    },
    {
      "epoch": 0.3418241577649959,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008995431966792409,
      "loss": 5.3611,
      "step": 2496
    },
    {
      "epoch": 0.34196110654615175,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008995413494588504,
      "loss": 5.2804,
      "step": 2497
    },
    {
      "epoch": 0.3420980553273076,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008995394985130036,
      "loss": 5.3603,
      "step": 2498
    },
    {
      "epoch": 0.3422350041084634,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008995376438417161,
      "loss": 5.3131,
      "step": 2499
    },
    {
      "epoch": 0.3423719528896193,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008995357854450029,
      "loss": 5.3605,
      "step": 2500
    },
    {
      "epoch": 0.34250890167077513,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008995339233228795,
      "loss": 5.3849,
      "step": 2501
    },
    {
      "epoch": 0.34264585045193097,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008995320574753614,
      "loss": 5.3872,
      "step": 2502
    },
    {
      "epoch": 0.3427827992330868,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008995301879024639,
      "loss": 5.466,
      "step": 2503
    },
    {
      "epoch": 0.3429197480142427,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008995283146042027,
      "loss": 5.3799,
      "step": 2504
    },
    {
      "epoch": 0.3430566967953985,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008995264375805933,
      "loss": 5.398,
      "step": 2505
    },
    {
      "epoch": 0.34319364557655435,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008995245568316513,
      "loss": 5.439,
      "step": 2506
    },
    {
      "epoch": 0.34333059435771024,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008995226723573919,
      "loss": 5.3658,
      "step": 2507
    },
    {
      "epoch": 0.34346754313886607,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008995207841578311,
      "loss": 5.3468,
      "step": 2508
    },
    {
      "epoch": 0.3436044919200219,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008995188922329845,
      "loss": 5.3927,
      "step": 2509
    },
    {
      "epoch": 0.34374144070117774,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008995169965828677,
      "loss": 5.437,
      "step": 2510
    },
    {
      "epoch": 0.3438783894823336,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008995150972074964,
      "loss": 5.3648,
      "step": 2511
    },
    {
      "epoch": 0.34401533826348946,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008995131941068865,
      "loss": 5.3489,
      "step": 2512
    },
    {
      "epoch": 0.3441522870446453,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008995112872810535,
      "loss": 5.3335,
      "step": 2513
    },
    {
      "epoch": 0.3442892358258012,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008995093767300134,
      "loss": 5.3249,
      "step": 2514
    },
    {
      "epoch": 0.344426184606957,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000899507462453782,
      "loss": 5.3853,
      "step": 2515
    },
    {
      "epoch": 0.34456313338811284,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008995055444523751,
      "loss": 5.3185,
      "step": 2516
    },
    {
      "epoch": 0.3447000821692687,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008995036227258087,
      "loss": 5.3899,
      "step": 2517
    },
    {
      "epoch": 0.34483703095042456,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008995016972740987,
      "loss": 5.3743,
      "step": 2518
    },
    {
      "epoch": 0.3449739797315804,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000899499768097261,
      "loss": 5.3438,
      "step": 2519
    },
    {
      "epoch": 0.3451109285127362,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008994978351953115,
      "loss": 5.3674,
      "step": 2520
    },
    {
      "epoch": 0.34524787729389206,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008994958985682665,
      "loss": 5.3397,
      "step": 2521
    },
    {
      "epoch": 0.34538482607504795,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008994939582161418,
      "loss": 5.3434,
      "step": 2522
    },
    {
      "epoch": 0.3455217748562038,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008994920141389536,
      "loss": 5.4006,
      "step": 2523
    },
    {
      "epoch": 0.3456587236373596,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008994900663367179,
      "loss": 5.341,
      "step": 2524
    },
    {
      "epoch": 0.3457956724185155,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008994881148094509,
      "loss": 5.3294,
      "step": 2525
    },
    {
      "epoch": 0.34593262119967133,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008994861595571689,
      "loss": 5.3384,
      "step": 2526
    },
    {
      "epoch": 0.34606956998082716,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008994842005798879,
      "loss": 5.369,
      "step": 2527
    },
    {
      "epoch": 0.346206518761983,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008994822378776241,
      "loss": 5.3329,
      "step": 2528
    },
    {
      "epoch": 0.3463434675431389,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008994802714503941,
      "loss": 5.354,
      "step": 2529
    },
    {
      "epoch": 0.3464804163242947,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008994783012982139,
      "loss": 5.4144,
      "step": 2530
    },
    {
      "epoch": 0.34661736510545055,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008994763274211,
      "loss": 5.3339,
      "step": 2531
    },
    {
      "epoch": 0.34675431388660644,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008994743498190685,
      "loss": 5.4126,
      "step": 2532
    },
    {
      "epoch": 0.34689126266776227,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008994723684921361,
      "loss": 5.3735,
      "step": 2533
    },
    {
      "epoch": 0.3470282114489181,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000899470383440319,
      "loss": 5.3872,
      "step": 2534
    },
    {
      "epoch": 0.34716516023007393,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008994683946636338,
      "loss": 5.3805,
      "step": 2535
    },
    {
      "epoch": 0.3473021090112298,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008994664021620969,
      "loss": 5.3193,
      "step": 2536
    },
    {
      "epoch": 0.34743905779238565,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008994644059357247,
      "loss": 5.355,
      "step": 2537
    },
    {
      "epoch": 0.3475760065735415,
      "grad_norm": 0.59375,
      "learning_rate": 0.000899462405984534,
      "loss": 5.257,
      "step": 2538
    },
    {
      "epoch": 0.3477129553546973,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008994604023085412,
      "loss": 5.3003,
      "step": 2539
    },
    {
      "epoch": 0.3478499041358532,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000899458394907763,
      "loss": 5.3529,
      "step": 2540
    },
    {
      "epoch": 0.34798685291700904,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008994563837822159,
      "loss": 5.4364,
      "step": 2541
    },
    {
      "epoch": 0.34812380169816487,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008994543689319166,
      "loss": 5.3656,
      "step": 2542
    },
    {
      "epoch": 0.34826075047932076,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000899452350356882,
      "loss": 5.3081,
      "step": 2543
    },
    {
      "epoch": 0.3483976992604766,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008994503280571287,
      "loss": 5.3459,
      "step": 2544
    },
    {
      "epoch": 0.3485346480416324,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008994483020326732,
      "loss": 5.3434,
      "step": 2545
    },
    {
      "epoch": 0.34867159682278825,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008994462722835327,
      "loss": 5.3296,
      "step": 2546
    },
    {
      "epoch": 0.34880854560394414,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008994442388097237,
      "loss": 5.3905,
      "step": 2547
    },
    {
      "epoch": 0.3489454943851,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008994422016112634,
      "loss": 5.3372,
      "step": 2548
    },
    {
      "epoch": 0.3490824431662558,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008994401606881682,
      "loss": 5.3852,
      "step": 2549
    },
    {
      "epoch": 0.3492193919474117,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008994381160404556,
      "loss": 5.3593,
      "step": 2550
    },
    {
      "epoch": 0.3493563407285675,
      "grad_norm": 0.75,
      "learning_rate": 0.0008994360676681421,
      "loss": 5.3779,
      "step": 2551
    },
    {
      "epoch": 0.34949328950972336,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008994340155712448,
      "loss": 5.2999,
      "step": 2552
    },
    {
      "epoch": 0.3496302382908792,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008994319597497806,
      "loss": 5.3835,
      "step": 2553
    },
    {
      "epoch": 0.3497671870720351,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000899429900203767,
      "loss": 5.4034,
      "step": 2554
    },
    {
      "epoch": 0.3499041358531909,
      "grad_norm": 0.5,
      "learning_rate": 0.0008994278369332205,
      "loss": 5.2953,
      "step": 2555
    },
    {
      "epoch": 0.35004108463434674,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008994257699381584,
      "loss": 5.3612,
      "step": 2556
    },
    {
      "epoch": 0.3501780334155026,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008994236992185978,
      "loss": 5.3875,
      "step": 2557
    },
    {
      "epoch": 0.35031498219665846,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008994216247745561,
      "loss": 5.4053,
      "step": 2558
    },
    {
      "epoch": 0.3504519309778143,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008994195466060502,
      "loss": 5.3426,
      "step": 2559
    },
    {
      "epoch": 0.35058887975897013,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008994174647130973,
      "loss": 5.3506,
      "step": 2560
    },
    {
      "epoch": 0.350725828540126,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008994153790957151,
      "loss": 5.3359,
      "step": 2561
    },
    {
      "epoch": 0.35086277732128185,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008994132897539204,
      "loss": 5.3359,
      "step": 2562
    },
    {
      "epoch": 0.3509997261024377,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008994111966877306,
      "loss": 5.2485,
      "step": 2563
    },
    {
      "epoch": 0.3511366748835935,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008994090998971631,
      "loss": 5.2925,
      "step": 2564
    },
    {
      "epoch": 0.3512736236647494,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008994069993822354,
      "loss": 5.3621,
      "step": 2565
    },
    {
      "epoch": 0.35141057244590523,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008994048951429648,
      "loss": 5.3995,
      "step": 2566
    },
    {
      "epoch": 0.35154752122706107,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008994027871793688,
      "loss": 5.3969,
      "step": 2567
    },
    {
      "epoch": 0.35168447000821695,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008994006754914647,
      "loss": 5.2988,
      "step": 2568
    },
    {
      "epoch": 0.3518214187893728,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008993985600792702,
      "loss": 5.3739,
      "step": 2569
    },
    {
      "epoch": 0.3519583675705286,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008993964409428027,
      "loss": 5.35,
      "step": 2570
    },
    {
      "epoch": 0.35209531635168445,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008993943180820797,
      "loss": 5.4079,
      "step": 2571
    },
    {
      "epoch": 0.35223226513284034,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008993921914971191,
      "loss": 5.3067,
      "step": 2572
    },
    {
      "epoch": 0.35236921391399617,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008993900611879383,
      "loss": 5.2957,
      "step": 2573
    },
    {
      "epoch": 0.352506162695152,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008993879271545549,
      "loss": 5.3632,
      "step": 2574
    },
    {
      "epoch": 0.35264311147630784,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008993857893969867,
      "loss": 5.299,
      "step": 2575
    },
    {
      "epoch": 0.3527800602574637,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008993836479152514,
      "loss": 5.3427,
      "step": 2576
    },
    {
      "epoch": 0.35291700903861956,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008993815027093666,
      "loss": 5.332,
      "step": 2577
    },
    {
      "epoch": 0.3530539578197754,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008993793537793503,
      "loss": 5.3874,
      "step": 2578
    },
    {
      "epoch": 0.3531909066009313,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008993772011252202,
      "loss": 5.3623,
      "step": 2579
    },
    {
      "epoch": 0.3533278553820871,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008993750447469942,
      "loss": 5.3968,
      "step": 2580
    },
    {
      "epoch": 0.35346480416324294,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008993728846446901,
      "loss": 5.3844,
      "step": 2581
    },
    {
      "epoch": 0.3536017529443988,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008993707208183258,
      "loss": 5.4101,
      "step": 2582
    },
    {
      "epoch": 0.35373870172555466,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008993685532679193,
      "loss": 5.3174,
      "step": 2583
    },
    {
      "epoch": 0.3538756505067105,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008993663819934885,
      "loss": 5.266,
      "step": 2584
    },
    {
      "epoch": 0.3540125992878663,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008993642069950514,
      "loss": 5.3843,
      "step": 2585
    },
    {
      "epoch": 0.3541495480690222,
      "grad_norm": 1.1875,
      "learning_rate": 0.000899362028272626,
      "loss": 5.3755,
      "step": 2586
    },
    {
      "epoch": 0.35428649685017805,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008993598458262305,
      "loss": 5.2812,
      "step": 2587
    },
    {
      "epoch": 0.3544234456313339,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008993576596558829,
      "loss": 5.3701,
      "step": 2588
    },
    {
      "epoch": 0.3545603944124897,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008993554697616012,
      "loss": 5.3825,
      "step": 2589
    },
    {
      "epoch": 0.3546973431936456,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008993532761434037,
      "loss": 5.4528,
      "step": 2590
    },
    {
      "epoch": 0.35483429197480143,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008993510788013084,
      "loss": 5.3186,
      "step": 2591
    },
    {
      "epoch": 0.35497124075595726,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008993488777353338,
      "loss": 5.3496,
      "step": 2592
    },
    {
      "epoch": 0.3551081895371131,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008993466729454978,
      "loss": 5.3685,
      "step": 2593
    },
    {
      "epoch": 0.355245138318269,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008993444644318191,
      "loss": 5.284,
      "step": 2594
    },
    {
      "epoch": 0.3553820870994248,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008993422521943155,
      "loss": 5.2998,
      "step": 2595
    },
    {
      "epoch": 0.35551903588058065,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008993400362330056,
      "loss": 5.4061,
      "step": 2596
    },
    {
      "epoch": 0.35565598466173654,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008993378165479079,
      "loss": 5.3299,
      "step": 2597
    },
    {
      "epoch": 0.35579293344289237,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008993355931390407,
      "loss": 5.2962,
      "step": 2598
    },
    {
      "epoch": 0.3559298822240482,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008993333660064222,
      "loss": 5.296,
      "step": 2599
    },
    {
      "epoch": 0.35606683100520403,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008993311351500711,
      "loss": 5.2983,
      "step": 2600
    },
    {
      "epoch": 0.3562037797863599,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008993289005700057,
      "loss": 5.3313,
      "step": 2601
    },
    {
      "epoch": 0.35634072856751575,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008993266622662448,
      "loss": 5.3262,
      "step": 2602
    },
    {
      "epoch": 0.3564776773486716,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008993244202388067,
      "loss": 5.3165,
      "step": 2603
    },
    {
      "epoch": 0.35661462612982747,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00089932217448771,
      "loss": 5.3109,
      "step": 2604
    },
    {
      "epoch": 0.3567515749109833,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008993199250129735,
      "loss": 5.3129,
      "step": 2605
    },
    {
      "epoch": 0.35688852369213914,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008993176718146156,
      "loss": 5.4012,
      "step": 2606
    },
    {
      "epoch": 0.35702547247329497,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008993154148926551,
      "loss": 5.3044,
      "step": 2607
    },
    {
      "epoch": 0.35716242125445086,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008993131542471107,
      "loss": 5.3356,
      "step": 2608
    },
    {
      "epoch": 0.3572993700356067,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008993108898780012,
      "loss": 5.3627,
      "step": 2609
    },
    {
      "epoch": 0.3574363188167625,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008993086217853452,
      "loss": 5.339,
      "step": 2610
    },
    {
      "epoch": 0.35757326759791835,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008993063499691617,
      "loss": 5.3668,
      "step": 2611
    },
    {
      "epoch": 0.35771021637907424,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008993040744294694,
      "loss": 5.3498,
      "step": 2612
    },
    {
      "epoch": 0.3578471651602301,
      "grad_norm": 0.62109375,
      "learning_rate": 0.000899301795166287,
      "loss": 5.3355,
      "step": 2613
    },
    {
      "epoch": 0.3579841139413859,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008992995121796337,
      "loss": 5.3116,
      "step": 2614
    },
    {
      "epoch": 0.3581210627225418,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008992972254695282,
      "loss": 5.3419,
      "step": 2615
    },
    {
      "epoch": 0.3582580115036976,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008992949350359896,
      "loss": 5.3424,
      "step": 2616
    },
    {
      "epoch": 0.35839496028485346,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008992926408790369,
      "loss": 5.2967,
      "step": 2617
    },
    {
      "epoch": 0.3585319090660093,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008992903429986889,
      "loss": 5.323,
      "step": 2618
    },
    {
      "epoch": 0.3586688578471652,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008992880413949648,
      "loss": 5.3099,
      "step": 2619
    },
    {
      "epoch": 0.358805806628321,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008992857360678836,
      "loss": 5.3397,
      "step": 2620
    },
    {
      "epoch": 0.35894275540947684,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008992834270174644,
      "loss": 5.3019,
      "step": 2621
    },
    {
      "epoch": 0.3590797041906327,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008992811142437265,
      "loss": 5.3226,
      "step": 2622
    },
    {
      "epoch": 0.35921665297178856,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008992787977466889,
      "loss": 5.3149,
      "step": 2623
    },
    {
      "epoch": 0.3593536017529444,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008992764775263708,
      "loss": 5.4041,
      "step": 2624
    },
    {
      "epoch": 0.35949055053410023,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008992741535827917,
      "loss": 5.3563,
      "step": 2625
    },
    {
      "epoch": 0.3596274993152561,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008992718259159704,
      "loss": 5.3135,
      "step": 2626
    },
    {
      "epoch": 0.35976444809641195,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008992694945259265,
      "loss": 5.3756,
      "step": 2627
    },
    {
      "epoch": 0.3599013968775678,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008992671594126793,
      "loss": 5.3088,
      "step": 2628
    },
    {
      "epoch": 0.3600383456587236,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008992648205762479,
      "loss": 5.3544,
      "step": 2629
    },
    {
      "epoch": 0.3601752944398795,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000899262478016652,
      "loss": 5.3083,
      "step": 2630
    },
    {
      "epoch": 0.36031224322103533,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008992601317339109,
      "loss": 5.2834,
      "step": 2631
    },
    {
      "epoch": 0.36044919200219117,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000899257781728044,
      "loss": 5.3368,
      "step": 2632
    },
    {
      "epoch": 0.36058614078334705,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008992554279990709,
      "loss": 5.377,
      "step": 2633
    },
    {
      "epoch": 0.3607230895645029,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008992530705470109,
      "loss": 5.3222,
      "step": 2634
    },
    {
      "epoch": 0.3608600383456587,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008992507093718838,
      "loss": 5.2838,
      "step": 2635
    },
    {
      "epoch": 0.36099698712681455,
      "grad_norm": 0.71875,
      "learning_rate": 0.000899248344473709,
      "loss": 5.3929,
      "step": 2636
    },
    {
      "epoch": 0.36113393590797044,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008992459758525059,
      "loss": 5.3191,
      "step": 2637
    },
    {
      "epoch": 0.36127088468912627,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008992436035082945,
      "loss": 5.3819,
      "step": 2638
    },
    {
      "epoch": 0.3614078334702821,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008992412274410943,
      "loss": 5.409,
      "step": 2639
    },
    {
      "epoch": 0.36154478225143794,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000899238847650925,
      "loss": 5.3938,
      "step": 2640
    },
    {
      "epoch": 0.3616817310325938,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008992364641378064,
      "loss": 5.3494,
      "step": 2641
    },
    {
      "epoch": 0.36181867981374966,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008992340769017581,
      "loss": 5.3157,
      "step": 2642
    },
    {
      "epoch": 0.3619556285949055,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008992316859427999,
      "loss": 5.4071,
      "step": 2643
    },
    {
      "epoch": 0.3620925773760614,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008992292912609518,
      "loss": 5.354,
      "step": 2644
    },
    {
      "epoch": 0.3622295261572172,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008992268928562334,
      "loss": 5.3284,
      "step": 2645
    },
    {
      "epoch": 0.36236647493837304,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008992244907286647,
      "loss": 5.3167,
      "step": 2646
    },
    {
      "epoch": 0.36250342371952887,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008992220848782657,
      "loss": 5.382,
      "step": 2647
    },
    {
      "epoch": 0.36264037250068476,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008992196753050563,
      "loss": 5.2958,
      "step": 2648
    },
    {
      "epoch": 0.3627773212818406,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008992172620090562,
      "loss": 5.3832,
      "step": 2649
    },
    {
      "epoch": 0.3629142700629964,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008992148449902858,
      "loss": 5.2926,
      "step": 2650
    },
    {
      "epoch": 0.3630512188441523,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008992124242487647,
      "loss": 5.31,
      "step": 2651
    },
    {
      "epoch": 0.36318816762530814,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008992099997845134,
      "loss": 5.2913,
      "step": 2652
    },
    {
      "epoch": 0.363325116406464,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008992075715975517,
      "loss": 5.3557,
      "step": 2653
    },
    {
      "epoch": 0.3634620651876198,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008992051396878998,
      "loss": 5.4022,
      "step": 2654
    },
    {
      "epoch": 0.3635990139687757,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008992027040555779,
      "loss": 5.2959,
      "step": 2655
    },
    {
      "epoch": 0.36373596274993153,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008992002647006061,
      "loss": 5.3344,
      "step": 2656
    },
    {
      "epoch": 0.36387291153108736,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008991978216230049,
      "loss": 5.3436,
      "step": 2657
    },
    {
      "epoch": 0.3640098603122432,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008991953748227941,
      "loss": 5.3983,
      "step": 2658
    },
    {
      "epoch": 0.3641468090933991,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008991929242999942,
      "loss": 5.2982,
      "step": 2659
    },
    {
      "epoch": 0.3642837578745549,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0008991904700546254,
      "loss": 5.3253,
      "step": 2660
    },
    {
      "epoch": 0.36442070665571075,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008991880120867085,
      "loss": 5.3592,
      "step": 2661
    },
    {
      "epoch": 0.36455765543686663,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008991855503962632,
      "loss": 5.32,
      "step": 2662
    },
    {
      "epoch": 0.36469460421802247,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008991830849833102,
      "loss": 5.3721,
      "step": 2663
    },
    {
      "epoch": 0.3648315529991783,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008991806158478701,
      "loss": 5.3752,
      "step": 2664
    },
    {
      "epoch": 0.36496850178033413,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008991781429899631,
      "loss": 5.3465,
      "step": 2665
    },
    {
      "epoch": 0.36510545056149,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008991756664096099,
      "loss": 5.3313,
      "step": 2666
    },
    {
      "epoch": 0.36524239934264585,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008991731861068308,
      "loss": 5.323,
      "step": 2667
    },
    {
      "epoch": 0.3653793481238017,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008991707020816465,
      "loss": 5.4047,
      "step": 2668
    },
    {
      "epoch": 0.36551629690495757,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008991682143340776,
      "loss": 5.3632,
      "step": 2669
    },
    {
      "epoch": 0.3656532456861134,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008991657228641447,
      "loss": 5.3519,
      "step": 2670
    },
    {
      "epoch": 0.36579019446726924,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008991632276718684,
      "loss": 5.3166,
      "step": 2671
    },
    {
      "epoch": 0.36592714324842507,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008991607287572694,
      "loss": 5.3258,
      "step": 2672
    },
    {
      "epoch": 0.36606409202958096,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008991582261203683,
      "loss": 5.3047,
      "step": 2673
    },
    {
      "epoch": 0.3662010408107368,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008991557197611861,
      "loss": 5.3322,
      "step": 2674
    },
    {
      "epoch": 0.3663379895918926,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008991532096797434,
      "loss": 5.3288,
      "step": 2675
    },
    {
      "epoch": 0.36647493837304845,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000899150695876061,
      "loss": 5.3029,
      "step": 2676
    },
    {
      "epoch": 0.36661188715420434,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008991481783501596,
      "loss": 5.3649,
      "step": 2677
    },
    {
      "epoch": 0.3667488359353602,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008991456571020605,
      "loss": 5.3325,
      "step": 2678
    },
    {
      "epoch": 0.366885784716516,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008991431321317842,
      "loss": 5.3513,
      "step": 2679
    },
    {
      "epoch": 0.3670227334976719,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008991406034393516,
      "loss": 5.306,
      "step": 2680
    },
    {
      "epoch": 0.3671596822788277,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000899138071024784,
      "loss": 5.2903,
      "step": 2681
    },
    {
      "epoch": 0.36729663105998356,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008991355348881021,
      "loss": 5.3495,
      "step": 2682
    },
    {
      "epoch": 0.3674335798411394,
      "grad_norm": 0.578125,
      "learning_rate": 0.000899132995029327,
      "loss": 5.2951,
      "step": 2683
    },
    {
      "epoch": 0.3675705286222953,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008991304514484797,
      "loss": 5.3059,
      "step": 2684
    },
    {
      "epoch": 0.3677074774034511,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008991279041455814,
      "loss": 5.3178,
      "step": 2685
    },
    {
      "epoch": 0.36784442618460694,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008991253531206531,
      "loss": 5.3087,
      "step": 2686
    },
    {
      "epoch": 0.36798137496576283,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008991227983737159,
      "loss": 5.3882,
      "step": 2687
    },
    {
      "epoch": 0.36811832374691866,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008991202399047912,
      "loss": 5.3023,
      "step": 2688
    },
    {
      "epoch": 0.3682552725280745,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008991176777139,
      "loss": 5.2843,
      "step": 2689
    },
    {
      "epoch": 0.3683922213092303,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008991151118010636,
      "loss": 5.3244,
      "step": 2690
    },
    {
      "epoch": 0.3685291700903862,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008991125421663032,
      "loss": 5.2217,
      "step": 2691
    },
    {
      "epoch": 0.36866611887154205,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008991099688096402,
      "loss": 5.2759,
      "step": 2692
    },
    {
      "epoch": 0.3688030676526979,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008991073917310959,
      "loss": 5.2821,
      "step": 2693
    },
    {
      "epoch": 0.3689400164338537,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008991048109306915,
      "loss": 5.3536,
      "step": 2694
    },
    {
      "epoch": 0.3690769652150096,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008991022264084487,
      "loss": 5.3086,
      "step": 2695
    },
    {
      "epoch": 0.36921391399616543,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008990996381643886,
      "loss": 5.2593,
      "step": 2696
    },
    {
      "epoch": 0.36935086277732126,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008990970461985328,
      "loss": 5.3158,
      "step": 2697
    },
    {
      "epoch": 0.36948781155847715,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008990944505109027,
      "loss": 5.3612,
      "step": 2698
    },
    {
      "epoch": 0.369624760339633,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00089909185110152,
      "loss": 5.2918,
      "step": 2699
    },
    {
      "epoch": 0.3697617091207888,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000899089247970406,
      "loss": 5.3018,
      "step": 2700
    },
    {
      "epoch": 0.36989865790194465,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008990866411175825,
      "loss": 5.3369,
      "step": 2701
    },
    {
      "epoch": 0.37003560668310054,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000899084030543071,
      "loss": 5.3359,
      "step": 2702
    },
    {
      "epoch": 0.37017255546425637,
      "grad_norm": 0.84375,
      "learning_rate": 0.000899081416246893,
      "loss": 5.2866,
      "step": 2703
    },
    {
      "epoch": 0.3703095042454122,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008990787982290704,
      "loss": 5.3124,
      "step": 2704
    },
    {
      "epoch": 0.3704464530265681,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008990761764896247,
      "loss": 5.2754,
      "step": 2705
    },
    {
      "epoch": 0.3705834018077239,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008990735510285778,
      "loss": 5.3228,
      "step": 2706
    },
    {
      "epoch": 0.37072035058887975,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008990709218459513,
      "loss": 5.2929,
      "step": 2707
    },
    {
      "epoch": 0.3708572993700356,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008990682889417672,
      "loss": 5.3549,
      "step": 2708
    },
    {
      "epoch": 0.3709942481511915,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000899065652316047,
      "loss": 5.2885,
      "step": 2709
    },
    {
      "epoch": 0.3711311969323473,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008990630119688128,
      "loss": 5.3723,
      "step": 2710
    },
    {
      "epoch": 0.37126814571350314,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008990603679000865,
      "loss": 5.3011,
      "step": 2711
    },
    {
      "epoch": 0.37140509449465897,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008990577201098899,
      "loss": 5.3229,
      "step": 2712
    },
    {
      "epoch": 0.37154204327581486,
      "grad_norm": 0.5,
      "learning_rate": 0.0008990550685982449,
      "loss": 5.2645,
      "step": 2713
    },
    {
      "epoch": 0.3716789920569707,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008990524133651737,
      "loss": 5.2603,
      "step": 2714
    },
    {
      "epoch": 0.3718159408381265,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008990497544106979,
      "loss": 5.2948,
      "step": 2715
    },
    {
      "epoch": 0.3719528896192824,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008990470917348399,
      "loss": 5.2889,
      "step": 2716
    },
    {
      "epoch": 0.37208983840043824,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0008990444253376217,
      "loss": 5.3767,
      "step": 2717
    },
    {
      "epoch": 0.3722267871815941,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008990417552190654,
      "loss": 5.2386,
      "step": 2718
    },
    {
      "epoch": 0.3723637359627499,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008990390813791929,
      "loss": 5.2843,
      "step": 2719
    },
    {
      "epoch": 0.3725006847439058,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008990364038180265,
      "loss": 5.4081,
      "step": 2720
    },
    {
      "epoch": 0.37263763352506163,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008990337225355886,
      "loss": 5.3131,
      "step": 2721
    },
    {
      "epoch": 0.37277458230621746,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008990310375319012,
      "loss": 5.3101,
      "step": 2722
    },
    {
      "epoch": 0.37291153108737335,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008990283488069865,
      "loss": 5.2881,
      "step": 2723
    },
    {
      "epoch": 0.3730484798685292,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000899025656360867,
      "loss": 5.2285,
      "step": 2724
    },
    {
      "epoch": 0.373185428649685,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008990229601935648,
      "loss": 5.4199,
      "step": 2725
    },
    {
      "epoch": 0.37332237743084085,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008990202603051024,
      "loss": 5.3601,
      "step": 2726
    },
    {
      "epoch": 0.37345932621199673,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000899017556695502,
      "loss": 5.3981,
      "step": 2727
    },
    {
      "epoch": 0.37359627499315257,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008990148493647862,
      "loss": 5.365,
      "step": 2728
    },
    {
      "epoch": 0.3737332237743084,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008990121383129774,
      "loss": 5.3075,
      "step": 2729
    },
    {
      "epoch": 0.37387017255546423,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008990094235400978,
      "loss": 5.3247,
      "step": 2730
    },
    {
      "epoch": 0.3740071213366201,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008990067050461703,
      "loss": 5.35,
      "step": 2731
    },
    {
      "epoch": 0.37414407011777595,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008990039828312172,
      "loss": 5.3409,
      "step": 2732
    },
    {
      "epoch": 0.3742810188989318,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008990012568952612,
      "loss": 5.3115,
      "step": 2733
    },
    {
      "epoch": 0.37441796768008767,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008989985272383248,
      "loss": 5.3784,
      "step": 2734
    },
    {
      "epoch": 0.3745549164612435,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008989957938604304,
      "loss": 5.2886,
      "step": 2735
    },
    {
      "epoch": 0.37469186524239934,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000898993056761601,
      "loss": 5.2581,
      "step": 2736
    },
    {
      "epoch": 0.37482881402355517,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008989903159418592,
      "loss": 5.3174,
      "step": 2737
    },
    {
      "epoch": 0.37496576280471106,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008989875714012275,
      "loss": 5.2544,
      "step": 2738
    },
    {
      "epoch": 0.3751027115858669,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008989848231397289,
      "loss": 5.3101,
      "step": 2739
    },
    {
      "epoch": 0.3752396603670227,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008989820711573861,
      "loss": 5.2315,
      "step": 2740
    },
    {
      "epoch": 0.3753766091481786,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008989793154542218,
      "loss": 5.2557,
      "step": 2741
    },
    {
      "epoch": 0.37551355792933444,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008989765560302591,
      "loss": 5.3522,
      "step": 2742
    },
    {
      "epoch": 0.3756505067104903,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008989737928855204,
      "loss": 5.2505,
      "step": 2743
    },
    {
      "epoch": 0.3757874554916461,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008989710260200293,
      "loss": 5.2732,
      "step": 2744
    },
    {
      "epoch": 0.375924404272802,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008989682554338081,
      "loss": 5.3537,
      "step": 2745
    },
    {
      "epoch": 0.3760613530539578,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008989654811268799,
      "loss": 5.3185,
      "step": 2746
    },
    {
      "epoch": 0.37619830183511366,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008989627030992679,
      "loss": 5.3007,
      "step": 2747
    },
    {
      "epoch": 0.3763352506162695,
      "grad_norm": 0.6015625,
      "learning_rate": 0.000898959921350995,
      "loss": 5.2379,
      "step": 2748
    },
    {
      "epoch": 0.3764721993974254,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008989571358820843,
      "loss": 5.3603,
      "step": 2749
    },
    {
      "epoch": 0.3766091481785812,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008989543466925587,
      "loss": 5.3476,
      "step": 2750
    },
    {
      "epoch": 0.37674609695973704,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008989515537824414,
      "loss": 5.2904,
      "step": 2751
    },
    {
      "epoch": 0.37688304574089293,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008989487571517558,
      "loss": 5.3475,
      "step": 2752
    },
    {
      "epoch": 0.37701999452204876,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008989459568005247,
      "loss": 5.2831,
      "step": 2753
    },
    {
      "epoch": 0.3771569433032046,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008989431527287716,
      "loss": 5.2366,
      "step": 2754
    },
    {
      "epoch": 0.3772938920843604,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008989403449365195,
      "loss": 5.3503,
      "step": 2755
    },
    {
      "epoch": 0.3774308408655163,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008989375334237919,
      "loss": 5.3103,
      "step": 2756
    },
    {
      "epoch": 0.37756778964667215,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000898934718190612,
      "loss": 5.3576,
      "step": 2757
    },
    {
      "epoch": 0.377704738427828,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008989318992370031,
      "loss": 5.3093,
      "step": 2758
    },
    {
      "epoch": 0.37784168720898387,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008989290765629884,
      "loss": 5.3232,
      "step": 2759
    },
    {
      "epoch": 0.3779786359901397,
      "grad_norm": 0.5,
      "learning_rate": 0.0008989262501685916,
      "loss": 5.2135,
      "step": 2760
    },
    {
      "epoch": 0.37811558477129553,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000898923420053836,
      "loss": 5.2468,
      "step": 2761
    },
    {
      "epoch": 0.37825253355245136,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008989205862187451,
      "loss": 5.3089,
      "step": 2762
    },
    {
      "epoch": 0.37838948233360725,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008989177486633422,
      "loss": 5.3746,
      "step": 2763
    },
    {
      "epoch": 0.3785264311147631,
      "grad_norm": 0.484375,
      "learning_rate": 0.000898914907387651,
      "loss": 5.2956,
      "step": 2764
    },
    {
      "epoch": 0.3786633798959189,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000898912062391695,
      "loss": 5.3001,
      "step": 2765
    },
    {
      "epoch": 0.37880032867707475,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008989092136754977,
      "loss": 5.3228,
      "step": 2766
    },
    {
      "epoch": 0.37893727745823064,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008989063612390828,
      "loss": 5.334,
      "step": 2767
    },
    {
      "epoch": 0.37907422623938647,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000898903505082474,
      "loss": 5.3265,
      "step": 2768
    },
    {
      "epoch": 0.3792111750205423,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008989006452056949,
      "loss": 5.3487,
      "step": 2769
    },
    {
      "epoch": 0.3793481238016982,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008988977816087691,
      "loss": 5.2994,
      "step": 2770
    },
    {
      "epoch": 0.379485072582854,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008988949142917205,
      "loss": 5.3084,
      "step": 2771
    },
    {
      "epoch": 0.37962202136400985,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008988920432545727,
      "loss": 5.3481,
      "step": 2772
    },
    {
      "epoch": 0.3797589701451657,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008988891684973496,
      "loss": 5.1979,
      "step": 2773
    },
    {
      "epoch": 0.3798959189263216,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000898886290020075,
      "loss": 5.3154,
      "step": 2774
    },
    {
      "epoch": 0.3800328677074774,
      "grad_norm": 1.375,
      "learning_rate": 0.0008988834078227727,
      "loss": 5.3818,
      "step": 2775
    },
    {
      "epoch": 0.38016981648863324,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008988805219054667,
      "loss": 5.2622,
      "step": 2776
    },
    {
      "epoch": 0.3803067652697891,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008988776322681808,
      "loss": 5.3227,
      "step": 2777
    },
    {
      "epoch": 0.38044371405094496,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008988747389109391,
      "loss": 5.3839,
      "step": 2778
    },
    {
      "epoch": 0.3805806628321008,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008988718418337653,
      "loss": 5.31,
      "step": 2779
    },
    {
      "epoch": 0.3807176116132566,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008988689410366837,
      "loss": 5.3188,
      "step": 2780
    },
    {
      "epoch": 0.3808545603944125,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008988660365197182,
      "loss": 5.2586,
      "step": 2781
    },
    {
      "epoch": 0.38099150917556834,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008988631282828931,
      "loss": 5.3244,
      "step": 2782
    },
    {
      "epoch": 0.3811284579567242,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000898860216326232,
      "loss": 5.3773,
      "step": 2783
    },
    {
      "epoch": 0.38126540673788,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008988573006497595,
      "loss": 5.3092,
      "step": 2784
    },
    {
      "epoch": 0.3814023555190359,
      "grad_norm": 0.5,
      "learning_rate": 0.0008988543812534995,
      "loss": 5.2874,
      "step": 2785
    },
    {
      "epoch": 0.38153930430019173,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008988514581374764,
      "loss": 5.2907,
      "step": 2786
    },
    {
      "epoch": 0.38167625308134756,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008988485313017143,
      "loss": 5.3341,
      "step": 2787
    },
    {
      "epoch": 0.38181320186250345,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008988456007462374,
      "loss": 5.3638,
      "step": 2788
    },
    {
      "epoch": 0.3819501506436593,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008988426664710699,
      "loss": 5.3074,
      "step": 2789
    },
    {
      "epoch": 0.3820870994248151,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008988397284762365,
      "loss": 5.358,
      "step": 2790
    },
    {
      "epoch": 0.38222404820597095,
      "grad_norm": 0.5,
      "learning_rate": 0.0008988367867617613,
      "loss": 5.3426,
      "step": 2791
    },
    {
      "epoch": 0.38236099698712683,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008988338413276686,
      "loss": 5.3842,
      "step": 2792
    },
    {
      "epoch": 0.38249794576828267,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008988308921739829,
      "loss": 5.295,
      "step": 2793
    },
    {
      "epoch": 0.3826348945494385,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008988279393007287,
      "loss": 5.3255,
      "step": 2794
    },
    {
      "epoch": 0.38277184333059433,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008988249827079304,
      "loss": 5.3117,
      "step": 2795
    },
    {
      "epoch": 0.3829087921117502,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008988220223956126,
      "loss": 5.3006,
      "step": 2796
    },
    {
      "epoch": 0.38304574089290605,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008988190583637996,
      "loss": 5.3496,
      "step": 2797
    },
    {
      "epoch": 0.3831826896740619,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008988160906125162,
      "loss": 5.3037,
      "step": 2798
    },
    {
      "epoch": 0.38331963845521777,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000898813119141787,
      "loss": 5.3062,
      "step": 2799
    },
    {
      "epoch": 0.3834565872363736,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008988101439516364,
      "loss": 5.2552,
      "step": 2800
    },
    {
      "epoch": 0.38359353601752944,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008988071650420891,
      "loss": 5.2517,
      "step": 2801
    },
    {
      "epoch": 0.38373048479868527,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008988041824131701,
      "loss": 5.3918,
      "step": 2802
    },
    {
      "epoch": 0.38386743357984116,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008988011960649039,
      "loss": 5.2664,
      "step": 2803
    },
    {
      "epoch": 0.384004382360997,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008987982059973151,
      "loss": 5.2905,
      "step": 2804
    },
    {
      "epoch": 0.3841413311421528,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008987952122104286,
      "loss": 5.2816,
      "step": 2805
    },
    {
      "epoch": 0.3842782799233087,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008987922147042693,
      "loss": 5.3267,
      "step": 2806
    },
    {
      "epoch": 0.38441522870446454,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008987892134788621,
      "loss": 5.3639,
      "step": 2807
    },
    {
      "epoch": 0.3845521774856204,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008987862085342315,
      "loss": 5.3407,
      "step": 2808
    },
    {
      "epoch": 0.3846891262667762,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008987831998704028,
      "loss": 5.3409,
      "step": 2809
    },
    {
      "epoch": 0.3848260750479321,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008987801874874008,
      "loss": 5.2978,
      "step": 2810
    },
    {
      "epoch": 0.3849630238290879,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008987771713852503,
      "loss": 5.2196,
      "step": 2811
    },
    {
      "epoch": 0.38509997261024376,
      "grad_norm": 0.5,
      "learning_rate": 0.0008987741515639767,
      "loss": 5.2904,
      "step": 2812
    },
    {
      "epoch": 0.3852369213913996,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008987711280236047,
      "loss": 5.2624,
      "step": 2813
    },
    {
      "epoch": 0.3853738701725555,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008987681007641593,
      "loss": 5.2644,
      "step": 2814
    },
    {
      "epoch": 0.3855108189537113,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008987650697856657,
      "loss": 5.344,
      "step": 2815
    },
    {
      "epoch": 0.38564776773486714,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008987620350881491,
      "loss": 5.2891,
      "step": 2816
    },
    {
      "epoch": 0.38578471651602303,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008987589966716347,
      "loss": 5.3274,
      "step": 2817
    },
    {
      "epoch": 0.38592166529717886,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008987559545361474,
      "loss": 5.3604,
      "step": 2818
    },
    {
      "epoch": 0.3860586140783347,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008987529086817125,
      "loss": 5.3197,
      "step": 2819
    },
    {
      "epoch": 0.3861955628594905,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008987498591083554,
      "loss": 5.3345,
      "step": 2820
    },
    {
      "epoch": 0.3863325116406464,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008987468058161014,
      "loss": 5.273,
      "step": 2821
    },
    {
      "epoch": 0.38646946042180225,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008987437488049755,
      "loss": 5.3251,
      "step": 2822
    },
    {
      "epoch": 0.3866064092029581,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008987406880750034,
      "loss": 5.3674,
      "step": 2823
    },
    {
      "epoch": 0.38674335798411397,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008987376236262102,
      "loss": 5.2538,
      "step": 2824
    },
    {
      "epoch": 0.3868803067652698,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008987345554586214,
      "loss": 5.3059,
      "step": 2825
    },
    {
      "epoch": 0.38701725554642563,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008987314835722625,
      "loss": 5.3065,
      "step": 2826
    },
    {
      "epoch": 0.38715420432758146,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008987284079671587,
      "loss": 5.3348,
      "step": 2827
    },
    {
      "epoch": 0.38729115310873735,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008987253286433358,
      "loss": 5.3088,
      "step": 2828
    },
    {
      "epoch": 0.3874281018898932,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008987222456008191,
      "loss": 5.3548,
      "step": 2829
    },
    {
      "epoch": 0.387565050671049,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008987191588396342,
      "loss": 5.3242,
      "step": 2830
    },
    {
      "epoch": 0.38770199945220485,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008987160683598068,
      "loss": 5.3186,
      "step": 2831
    },
    {
      "epoch": 0.38783894823336074,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008987129741613623,
      "loss": 5.3511,
      "step": 2832
    },
    {
      "epoch": 0.38797589701451657,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008987098762443265,
      "loss": 5.1806,
      "step": 2833
    },
    {
      "epoch": 0.3881128457956724,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000898706774608725,
      "loss": 5.23,
      "step": 2834
    },
    {
      "epoch": 0.3882497945768283,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008987036692545837,
      "loss": 5.2783,
      "step": 2835
    },
    {
      "epoch": 0.3883867433579841,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000898700560181928,
      "loss": 5.2575,
      "step": 2836
    },
    {
      "epoch": 0.38852369213913995,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008986974473907839,
      "loss": 5.2632,
      "step": 2837
    },
    {
      "epoch": 0.3886606409202958,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008986943308811772,
      "loss": 5.2991,
      "step": 2838
    },
    {
      "epoch": 0.3887975897014517,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008986912106531336,
      "loss": 5.3046,
      "step": 2839
    },
    {
      "epoch": 0.3889345384826075,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000898688086706679,
      "loss": 5.3032,
      "step": 2840
    },
    {
      "epoch": 0.38907148726376334,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008986849590418393,
      "loss": 5.3814,
      "step": 2841
    },
    {
      "epoch": 0.3892084360449192,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008986818276586405,
      "loss": 5.3282,
      "step": 2842
    },
    {
      "epoch": 0.38934538482607506,
      "grad_norm": 0.890625,
      "learning_rate": 0.0008986786925571083,
      "loss": 5.356,
      "step": 2843
    },
    {
      "epoch": 0.3894823336072309,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008986755537372691,
      "loss": 5.3325,
      "step": 2844
    },
    {
      "epoch": 0.3896192823883867,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008986724111991486,
      "loss": 5.3312,
      "step": 2845
    },
    {
      "epoch": 0.3897562311695426,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008986692649427729,
      "loss": 5.2918,
      "step": 2846
    },
    {
      "epoch": 0.38989317995069844,
      "grad_norm": 0.65625,
      "learning_rate": 0.000898666114968168,
      "loss": 5.3164,
      "step": 2847
    },
    {
      "epoch": 0.3900301287318543,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008986629612753603,
      "loss": 5.2534,
      "step": 2848
    },
    {
      "epoch": 0.3901670775130101,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008986598038643755,
      "loss": 5.249,
      "step": 2849
    },
    {
      "epoch": 0.390304026294166,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008986566427352401,
      "loss": 5.2566,
      "step": 2850
    },
    {
      "epoch": 0.39044097507532183,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008986534778879802,
      "loss": 5.2981,
      "step": 2851
    },
    {
      "epoch": 0.39057792385647766,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008986503093226221,
      "loss": 5.2488,
      "step": 2852
    },
    {
      "epoch": 0.39071487263763355,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008986471370391918,
      "loss": 5.2462,
      "step": 2853
    },
    {
      "epoch": 0.3908518214187894,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008986439610377158,
      "loss": 5.2488,
      "step": 2854
    },
    {
      "epoch": 0.3909887701999452,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008986407813182204,
      "loss": 5.3137,
      "step": 2855
    },
    {
      "epoch": 0.39112571898110104,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008986375978807321,
      "loss": 5.2912,
      "step": 2856
    },
    {
      "epoch": 0.39126266776225693,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000898634410725277,
      "loss": 5.2069,
      "step": 2857
    },
    {
      "epoch": 0.39139961654341276,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008986312198518817,
      "loss": 5.3285,
      "step": 2858
    },
    {
      "epoch": 0.3915365653245686,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008986280252605724,
      "loss": 5.2292,
      "step": 2859
    },
    {
      "epoch": 0.3916735141057245,
      "grad_norm": 0.5,
      "learning_rate": 0.000898624826951376,
      "loss": 5.2363,
      "step": 2860
    },
    {
      "epoch": 0.3918104628868803,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008986216249243186,
      "loss": 5.2911,
      "step": 2861
    },
    {
      "epoch": 0.39194741166803615,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008986184191794269,
      "loss": 5.2676,
      "step": 2862
    },
    {
      "epoch": 0.392084360449192,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008986152097167275,
      "loss": 5.2624,
      "step": 2863
    },
    {
      "epoch": 0.39222130923034787,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008986119965362469,
      "loss": 5.3399,
      "step": 2864
    },
    {
      "epoch": 0.3923582580115037,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008986087796380118,
      "loss": 5.3415,
      "step": 2865
    },
    {
      "epoch": 0.39249520679265953,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008986055590220488,
      "loss": 5.263,
      "step": 2866
    },
    {
      "epoch": 0.39263215557381537,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008986023346883848,
      "loss": 5.285,
      "step": 2867
    },
    {
      "epoch": 0.39276910435497125,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008985991066370462,
      "loss": 5.2439,
      "step": 2868
    },
    {
      "epoch": 0.3929060531361271,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00089859587486806,
      "loss": 5.2611,
      "step": 2869
    },
    {
      "epoch": 0.3930430019172829,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000898592639381453,
      "loss": 5.3639,
      "step": 2870
    },
    {
      "epoch": 0.3931799506984388,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008985894001772517,
      "loss": 5.2881,
      "step": 2871
    },
    {
      "epoch": 0.39331689947959464,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008985861572554831,
      "loss": 5.2546,
      "step": 2872
    },
    {
      "epoch": 0.39345384826075047,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008985829106161744,
      "loss": 5.2382,
      "step": 2873
    },
    {
      "epoch": 0.3935907970419063,
      "grad_norm": 0.47265625,
      "learning_rate": 0.000898579660259352,
      "loss": 5.2659,
      "step": 2874
    },
    {
      "epoch": 0.3937277458230622,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008985764061850432,
      "loss": 5.2548,
      "step": 2875
    },
    {
      "epoch": 0.393864694604218,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008985731483932747,
      "loss": 5.2815,
      "step": 2876
    },
    {
      "epoch": 0.39400164338537386,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008985698868840737,
      "loss": 5.3438,
      "step": 2877
    },
    {
      "epoch": 0.39413859216652974,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008985666216574671,
      "loss": 5.2949,
      "step": 2878
    },
    {
      "epoch": 0.3942755409476856,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008985633527134821,
      "loss": 5.2705,
      "step": 2879
    },
    {
      "epoch": 0.3944124897288414,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008985600800521456,
      "loss": 5.3003,
      "step": 2880
    },
    {
      "epoch": 0.39454943850999724,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000898556803673485,
      "loss": 5.2887,
      "step": 2881
    },
    {
      "epoch": 0.39468638729115313,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008985535235775272,
      "loss": 5.3385,
      "step": 2882
    },
    {
      "epoch": 0.39482333607230896,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008985502397642995,
      "loss": 5.3542,
      "step": 2883
    },
    {
      "epoch": 0.3949602848534648,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008985469522338291,
      "loss": 5.3045,
      "step": 2884
    },
    {
      "epoch": 0.3950972336346206,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000898543660986143,
      "loss": 5.3544,
      "step": 2885
    },
    {
      "epoch": 0.3952341824157765,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008985403660212689,
      "loss": 5.2346,
      "step": 2886
    },
    {
      "epoch": 0.39537113119693235,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008985370673392338,
      "loss": 5.3404,
      "step": 2887
    },
    {
      "epoch": 0.3955080799780882,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008985337649400653,
      "loss": 5.2968,
      "step": 2888
    },
    {
      "epoch": 0.39564502875924407,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008985304588237905,
      "loss": 5.259,
      "step": 2889
    },
    {
      "epoch": 0.3957819775403999,
      "grad_norm": 0.625,
      "learning_rate": 0.0008985271489904369,
      "loss": 5.2226,
      "step": 2890
    },
    {
      "epoch": 0.39591892632155573,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008985238354400321,
      "loss": 5.3306,
      "step": 2891
    },
    {
      "epoch": 0.39605587510271156,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008985205181726031,
      "loss": 5.2851,
      "step": 2892
    },
    {
      "epoch": 0.39619282388386745,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008985171971881778,
      "loss": 5.334,
      "step": 2893
    },
    {
      "epoch": 0.3963297726650233,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008985138724867838,
      "loss": 5.2777,
      "step": 2894
    },
    {
      "epoch": 0.3964667214461791,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008985105440684483,
      "loss": 5.3161,
      "step": 2895
    },
    {
      "epoch": 0.396603670227335,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008985072119331991,
      "loss": 5.2713,
      "step": 2896
    },
    {
      "epoch": 0.39674061900849084,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008985038760810636,
      "loss": 5.2417,
      "step": 2897
    },
    {
      "epoch": 0.39687756778964667,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008985005365120698,
      "loss": 5.2547,
      "step": 2898
    },
    {
      "epoch": 0.3970145165708025,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008984971932262452,
      "loss": 5.2705,
      "step": 2899
    },
    {
      "epoch": 0.3971514653519584,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008984938462236173,
      "loss": 5.243,
      "step": 2900
    },
    {
      "epoch": 0.3972884141331142,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008984904955042141,
      "loss": 5.2397,
      "step": 2901
    },
    {
      "epoch": 0.39742536291427005,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008984871410680635,
      "loss": 5.2192,
      "step": 2902
    },
    {
      "epoch": 0.3975623116954259,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008984837829151929,
      "loss": 5.224,
      "step": 2903
    },
    {
      "epoch": 0.3976992604765818,
      "grad_norm": 0.5,
      "learning_rate": 0.0008984804210456303,
      "loss": 5.2131,
      "step": 2904
    },
    {
      "epoch": 0.3978362092577376,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008984770554594036,
      "loss": 5.2711,
      "step": 2905
    },
    {
      "epoch": 0.39797315803889344,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008984736861565408,
      "loss": 5.303,
      "step": 2906
    },
    {
      "epoch": 0.3981101068200493,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008984703131370696,
      "loss": 5.291,
      "step": 2907
    },
    {
      "epoch": 0.39824705560120516,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000898466936401018,
      "loss": 5.2751,
      "step": 2908
    },
    {
      "epoch": 0.398384004382361,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008984635559484141,
      "loss": 5.2864,
      "step": 2909
    },
    {
      "epoch": 0.3985209531635168,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008984601717792859,
      "loss": 5.3069,
      "step": 2910
    },
    {
      "epoch": 0.3986579019446727,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008984567838936612,
      "loss": 5.2877,
      "step": 2911
    },
    {
      "epoch": 0.39879485072582854,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008984533922915685,
      "loss": 5.2062,
      "step": 2912
    },
    {
      "epoch": 0.3989317995069844,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008984499969730355,
      "loss": 5.2214,
      "step": 2913
    },
    {
      "epoch": 0.39906874828814026,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008984465979380907,
      "loss": 5.2484,
      "step": 2914
    },
    {
      "epoch": 0.3992056970692961,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008984431951867619,
      "loss": 5.2317,
      "step": 2915
    },
    {
      "epoch": 0.3993426458504519,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008984397887190776,
      "loss": 5.2065,
      "step": 2916
    },
    {
      "epoch": 0.39947959463160776,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008984363785350658,
      "loss": 5.2422,
      "step": 2917
    },
    {
      "epoch": 0.39961654341276365,
      "grad_norm": 0.51171875,
      "learning_rate": 0.000898432964634755,
      "loss": 5.2559,
      "step": 2918
    },
    {
      "epoch": 0.3997534921939195,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008984295470181733,
      "loss": 5.2872,
      "step": 2919
    },
    {
      "epoch": 0.3998904409750753,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008984261256853492,
      "loss": 5.237,
      "step": 2920
    },
    {
      "epoch": 0.40002738975623114,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008984227006363108,
      "loss": 5.2307,
      "step": 2921
    },
    {
      "epoch": 0.40016433853738703,
      "grad_norm": 0.5,
      "learning_rate": 0.0008984192718710866,
      "loss": 5.2647,
      "step": 2922
    },
    {
      "epoch": 0.40030128731854286,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008984158393897053,
      "loss": 5.2022,
      "step": 2923
    },
    {
      "epoch": 0.4004382360996987,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008984124031921948,
      "loss": 5.2409,
      "step": 2924
    },
    {
      "epoch": 0.4005751848808546,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000898408963278584,
      "loss": 5.2991,
      "step": 2925
    },
    {
      "epoch": 0.4007121336620104,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008984055196489012,
      "loss": 5.2601,
      "step": 2926
    },
    {
      "epoch": 0.40084908244316625,
      "grad_norm": 0.53125,
      "learning_rate": 0.000898402072303175,
      "loss": 5.2173,
      "step": 2927
    },
    {
      "epoch": 0.4009860312243221,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008983986212414342,
      "loss": 5.3016,
      "step": 2928
    },
    {
      "epoch": 0.40112298000547797,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008983951664637068,
      "loss": 5.2062,
      "step": 2929
    },
    {
      "epoch": 0.4012599287866338,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008983917079700221,
      "loss": 5.2496,
      "step": 2930
    },
    {
      "epoch": 0.40139687756778963,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008983882457604084,
      "loss": 5.2607,
      "step": 2931
    },
    {
      "epoch": 0.4015338263489455,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008983847798348945,
      "loss": 5.2924,
      "step": 2932
    },
    {
      "epoch": 0.40167077513010135,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008983813101935091,
      "loss": 5.2819,
      "step": 2933
    },
    {
      "epoch": 0.4018077239112572,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008983778368362808,
      "loss": 5.332,
      "step": 2934
    },
    {
      "epoch": 0.401944672692413,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008983743597632385,
      "loss": 5.3071,
      "step": 2935
    },
    {
      "epoch": 0.4020816214735689,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008983708789744112,
      "loss": 5.1985,
      "step": 2936
    },
    {
      "epoch": 0.40221857025472474,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008983673944698274,
      "loss": 5.3015,
      "step": 2937
    },
    {
      "epoch": 0.40235551903588057,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008983639062495164,
      "loss": 5.2893,
      "step": 2938
    },
    {
      "epoch": 0.4024924678170364,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008983604143135067,
      "loss": 5.1879,
      "step": 2939
    },
    {
      "epoch": 0.4026294165981923,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008983569186618275,
      "loss": 5.2945,
      "step": 2940
    },
    {
      "epoch": 0.4027663653793481,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008983534192945077,
      "loss": 5.2756,
      "step": 2941
    },
    {
      "epoch": 0.40290331416050396,
      "grad_norm": 0.5,
      "learning_rate": 0.0008983499162115761,
      "loss": 5.2239,
      "step": 2942
    },
    {
      "epoch": 0.40304026294165984,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000898346409413062,
      "loss": 5.2736,
      "step": 2943
    },
    {
      "epoch": 0.4031772117228157,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008983428988989944,
      "loss": 5.3024,
      "step": 2944
    },
    {
      "epoch": 0.4033141605039715,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008983393846694024,
      "loss": 5.2681,
      "step": 2945
    },
    {
      "epoch": 0.40345110928512734,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008983358667243149,
      "loss": 5.3819,
      "step": 2946
    },
    {
      "epoch": 0.40358805806628323,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008983323450637615,
      "loss": 5.2787,
      "step": 2947
    },
    {
      "epoch": 0.40372500684743906,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008983288196877711,
      "loss": 5.2293,
      "step": 2948
    },
    {
      "epoch": 0.4038619556285949,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008983252905963728,
      "loss": 5.1809,
      "step": 2949
    },
    {
      "epoch": 0.4039989044097508,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000898321757789596,
      "loss": 5.2414,
      "step": 2950
    },
    {
      "epoch": 0.4041358531909066,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00089831822126747,
      "loss": 5.2775,
      "step": 2951
    },
    {
      "epoch": 0.40427280197206245,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008983146810300241,
      "loss": 5.2722,
      "step": 2952
    },
    {
      "epoch": 0.4044097507532183,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008983111370772875,
      "loss": 5.3331,
      "step": 2953
    },
    {
      "epoch": 0.40454669953437417,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008983075894092898,
      "loss": 5.1987,
      "step": 2954
    },
    {
      "epoch": 0.40468364831553,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008983040380260602,
      "loss": 5.2983,
      "step": 2955
    },
    {
      "epoch": 0.40482059709668583,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008983004829276283,
      "loss": 5.2325,
      "step": 2956
    },
    {
      "epoch": 0.40495754587784166,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008982969241140234,
      "loss": 5.3287,
      "step": 2957
    },
    {
      "epoch": 0.40509449465899755,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008982933615852751,
      "loss": 5.269,
      "step": 2958
    },
    {
      "epoch": 0.4052314434401534,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008982897953414129,
      "loss": 5.1981,
      "step": 2959
    },
    {
      "epoch": 0.4053683922213092,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008982862253824664,
      "loss": 5.2566,
      "step": 2960
    },
    {
      "epoch": 0.4055053410024651,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008982826517084649,
      "loss": 5.2819,
      "step": 2961
    },
    {
      "epoch": 0.40564228978362094,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008982790743194385,
      "loss": 5.2495,
      "step": 2962
    },
    {
      "epoch": 0.40577923856477677,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008982754932154165,
      "loss": 5.2733,
      "step": 2963
    },
    {
      "epoch": 0.4059161873459326,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008982719083964286,
      "loss": 5.2794,
      "step": 2964
    },
    {
      "epoch": 0.4060531361270885,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008982683198625047,
      "loss": 5.2844,
      "step": 2965
    },
    {
      "epoch": 0.4061900849082443,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008982647276136744,
      "loss": 5.2014,
      "step": 2966
    },
    {
      "epoch": 0.40632703368940015,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008982611316499673,
      "loss": 5.2452,
      "step": 2967
    },
    {
      "epoch": 0.406463982470556,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008982575319714137,
      "loss": 5.2758,
      "step": 2968
    },
    {
      "epoch": 0.4066009312517119,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008982539285780428,
      "loss": 5.3411,
      "step": 2969
    },
    {
      "epoch": 0.4067378800328677,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000898250321469885,
      "loss": 5.2428,
      "step": 2970
    },
    {
      "epoch": 0.40687482881402354,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00089824671064697,
      "loss": 5.2644,
      "step": 2971
    },
    {
      "epoch": 0.4070117775951794,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008982430961093276,
      "loss": 5.2837,
      "step": 2972
    },
    {
      "epoch": 0.40714872637633526,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008982394778569877,
      "loss": 5.267,
      "step": 2973
    },
    {
      "epoch": 0.4072856751574911,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008982358558899807,
      "loss": 5.2652,
      "step": 2974
    },
    {
      "epoch": 0.4074226239386469,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008982322302083362,
      "loss": 5.2612,
      "step": 2975
    },
    {
      "epoch": 0.4075595727198028,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008982286008120845,
      "loss": 5.284,
      "step": 2976
    },
    {
      "epoch": 0.40769652150095864,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008982249677012555,
      "loss": 5.2219,
      "step": 2977
    },
    {
      "epoch": 0.4078334702821145,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008982213308758794,
      "loss": 5.3007,
      "step": 2978
    },
    {
      "epoch": 0.40797041906327036,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008982176903359863,
      "loss": 5.254,
      "step": 2979
    },
    {
      "epoch": 0.4081073678444262,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008982140460816063,
      "loss": 5.2282,
      "step": 2980
    },
    {
      "epoch": 0.408244316625582,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008982103981127699,
      "loss": 5.2711,
      "step": 2981
    },
    {
      "epoch": 0.40838126540673786,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000898206746429507,
      "loss": 5.2577,
      "step": 2982
    },
    {
      "epoch": 0.40851821418789375,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008982030910318482,
      "loss": 5.2003,
      "step": 2983
    },
    {
      "epoch": 0.4086551629690496,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008981994319198234,
      "loss": 5.2097,
      "step": 2984
    },
    {
      "epoch": 0.4087921117502054,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000898195769093463,
      "loss": 5.3089,
      "step": 2985
    },
    {
      "epoch": 0.40892906053136124,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008981921025527977,
      "loss": 5.2076,
      "step": 2986
    },
    {
      "epoch": 0.40906600931251713,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008981884322978576,
      "loss": 5.2336,
      "step": 2987
    },
    {
      "epoch": 0.40920295809367296,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000898184758328673,
      "loss": 5.2839,
      "step": 2988
    },
    {
      "epoch": 0.4093399068748288,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008981810806452747,
      "loss": 5.1897,
      "step": 2989
    },
    {
      "epoch": 0.4094768556559847,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000898177399247693,
      "loss": 5.2326,
      "step": 2990
    },
    {
      "epoch": 0.4096138044371405,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008981737141359582,
      "loss": 5.2251,
      "step": 2991
    },
    {
      "epoch": 0.40975075321829635,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008981700253101012,
      "loss": 5.2328,
      "step": 2992
    },
    {
      "epoch": 0.4098877019994522,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008981663327701523,
      "loss": 5.2447,
      "step": 2993
    },
    {
      "epoch": 0.41002465078060807,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008981626365161423,
      "loss": 5.2893,
      "step": 2994
    },
    {
      "epoch": 0.4101615995617639,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008981589365481017,
      "loss": 5.258,
      "step": 2995
    },
    {
      "epoch": 0.41029854834291973,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008981552328660612,
      "loss": 5.2926,
      "step": 2996
    },
    {
      "epoch": 0.4104354971240756,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008981515254700514,
      "loss": 5.253,
      "step": 2997
    },
    {
      "epoch": 0.41057244590523145,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008981478143601033,
      "loss": 5.2434,
      "step": 2998
    },
    {
      "epoch": 0.4107093946863873,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008981440995362473,
      "loss": 5.3198,
      "step": 2999
    },
    {
      "epoch": 0.4108463434675431,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008981403809985145,
      "loss": 5.2376,
      "step": 3000
    },
    {
      "epoch": 0.410983292248699,
      "grad_norm": 0.5,
      "learning_rate": 0.0008981366587469356,
      "loss": 5.2458,
      "step": 3001
    },
    {
      "epoch": 0.41112024102985484,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008981329327815413,
      "loss": 5.2614,
      "step": 3002
    },
    {
      "epoch": 0.41125718981101067,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008981292031023626,
      "loss": 5.2791,
      "step": 3003
    },
    {
      "epoch": 0.4113941385921665,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008981254697094306,
      "loss": 5.2555,
      "step": 3004
    },
    {
      "epoch": 0.4115310873733224,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008981217326027758,
      "loss": 5.2173,
      "step": 3005
    },
    {
      "epoch": 0.4116680361544782,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008981179917824297,
      "loss": 5.274,
      "step": 3006
    },
    {
      "epoch": 0.41180498493563406,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008981142472484228,
      "loss": 5.2331,
      "step": 3007
    },
    {
      "epoch": 0.41194193371678994,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008981104990007863,
      "loss": 5.1756,
      "step": 3008
    },
    {
      "epoch": 0.4120788824979458,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008981067470395515,
      "loss": 5.2627,
      "step": 3009
    },
    {
      "epoch": 0.4122158312791016,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008981029913647492,
      "loss": 5.1993,
      "step": 3010
    },
    {
      "epoch": 0.41235278006025744,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008980992319764106,
      "loss": 5.2253,
      "step": 3011
    },
    {
      "epoch": 0.41248972884141333,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000898095468874567,
      "loss": 5.3443,
      "step": 3012
    },
    {
      "epoch": 0.41262667762256916,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008980917020592494,
      "loss": 5.2446,
      "step": 3013
    },
    {
      "epoch": 0.412763626403725,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008980879315304891,
      "loss": 5.2478,
      "step": 3014
    },
    {
      "epoch": 0.4129005751848809,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008980841572883174,
      "loss": 5.1479,
      "step": 3015
    },
    {
      "epoch": 0.4130375239660367,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008980803793327655,
      "loss": 5.2521,
      "step": 3016
    },
    {
      "epoch": 0.41317447274719254,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008980765976638647,
      "loss": 5.2034,
      "step": 3017
    },
    {
      "epoch": 0.4133114215283484,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008980728122816463,
      "loss": 5.2565,
      "step": 3018
    },
    {
      "epoch": 0.41344837030950427,
      "grad_norm": 0.5,
      "learning_rate": 0.000898069023186142,
      "loss": 5.2447,
      "step": 3019
    },
    {
      "epoch": 0.4135853190906601,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008980652303773826,
      "loss": 5.2287,
      "step": 3020
    },
    {
      "epoch": 0.41372226787181593,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008980614338554001,
      "loss": 5.2532,
      "step": 3021
    },
    {
      "epoch": 0.41385921665297176,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008980576336202255,
      "loss": 5.2291,
      "step": 3022
    },
    {
      "epoch": 0.41399616543412765,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008980538296718907,
      "loss": 5.2915,
      "step": 3023
    },
    {
      "epoch": 0.4141331142152835,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008980500220104271,
      "loss": 5.2167,
      "step": 3024
    },
    {
      "epoch": 0.4142700629964393,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008980462106358661,
      "loss": 5.2355,
      "step": 3025
    },
    {
      "epoch": 0.4144070117775952,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008980423955482393,
      "loss": 5.1505,
      "step": 3026
    },
    {
      "epoch": 0.41454396055875103,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008980385767475785,
      "loss": 5.2652,
      "step": 3027
    },
    {
      "epoch": 0.41468090933990687,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008980347542339153,
      "loss": 5.2808,
      "step": 3028
    },
    {
      "epoch": 0.4148178581210627,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008980309280072812,
      "loss": 5.3457,
      "step": 3029
    },
    {
      "epoch": 0.4149548069022186,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008980270980677082,
      "loss": 5.2132,
      "step": 3030
    },
    {
      "epoch": 0.4150917556833744,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008980232644152277,
      "loss": 5.3047,
      "step": 3031
    },
    {
      "epoch": 0.41522870446453025,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008980194270498718,
      "loss": 5.2509,
      "step": 3032
    },
    {
      "epoch": 0.41536565324568614,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008980155859716722,
      "loss": 5.3044,
      "step": 3033
    },
    {
      "epoch": 0.41550260202684197,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0008980117411806606,
      "loss": 5.3182,
      "step": 3034
    },
    {
      "epoch": 0.4156395508079978,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000898007892676869,
      "loss": 5.3057,
      "step": 3035
    },
    {
      "epoch": 0.41577649958915364,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000898004040460329,
      "loss": 5.2489,
      "step": 3036
    },
    {
      "epoch": 0.4159134483703095,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000898000184531073,
      "loss": 5.2552,
      "step": 3037
    },
    {
      "epoch": 0.41605039715146536,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008979963248891326,
      "loss": 5.1976,
      "step": 3038
    },
    {
      "epoch": 0.4161873459326212,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00089799246153454,
      "loss": 5.2256,
      "step": 3039
    },
    {
      "epoch": 0.416324294713777,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008979885944673271,
      "loss": 5.2005,
      "step": 3040
    },
    {
      "epoch": 0.4164612434949329,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008979847236875258,
      "loss": 5.2939,
      "step": 3041
    },
    {
      "epoch": 0.41659819227608874,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008979808491951687,
      "loss": 5.2501,
      "step": 3042
    },
    {
      "epoch": 0.4167351410572446,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008979769709902873,
      "loss": 5.2805,
      "step": 3043
    },
    {
      "epoch": 0.41687208983840046,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008979730890729139,
      "loss": 5.2414,
      "step": 3044
    },
    {
      "epoch": 0.4170090386195563,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000897969203443081,
      "loss": 5.3342,
      "step": 3045
    },
    {
      "epoch": 0.4171459874007121,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008979653141008205,
      "loss": 5.2671,
      "step": 3046
    },
    {
      "epoch": 0.41728293618186796,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008979614210461647,
      "loss": 5.1595,
      "step": 3047
    },
    {
      "epoch": 0.41741988496302385,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008979575242791458,
      "loss": 5.3486,
      "step": 3048
    },
    {
      "epoch": 0.4175568337441797,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008979536237997962,
      "loss": 5.1842,
      "step": 3049
    },
    {
      "epoch": 0.4176937825253355,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008979497196081482,
      "loss": 5.2647,
      "step": 3050
    },
    {
      "epoch": 0.4178307313064914,
      "grad_norm": 0.578125,
      "learning_rate": 0.000897945811704234,
      "loss": 5.2071,
      "step": 3051
    },
    {
      "epoch": 0.41796768008764723,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008979419000880863,
      "loss": 5.2661,
      "step": 3052
    },
    {
      "epoch": 0.41810462886880306,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008979379847597372,
      "loss": 5.3628,
      "step": 3053
    },
    {
      "epoch": 0.4182415776499589,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008979340657192193,
      "loss": 5.29,
      "step": 3054
    },
    {
      "epoch": 0.4183785264311148,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000897930142966565,
      "loss": 5.2212,
      "step": 3055
    },
    {
      "epoch": 0.4185154752122706,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008979262165018071,
      "loss": 5.2813,
      "step": 3056
    },
    {
      "epoch": 0.41865242399342645,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008979222863249777,
      "loss": 5.2381,
      "step": 3057
    },
    {
      "epoch": 0.4187893727745823,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008979183524361096,
      "loss": 5.2394,
      "step": 3058
    },
    {
      "epoch": 0.41892632155573817,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008979144148352355,
      "loss": 5.3235,
      "step": 3059
    },
    {
      "epoch": 0.419063270336894,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008979104735223879,
      "loss": 5.2315,
      "step": 3060
    },
    {
      "epoch": 0.41920021911804983,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008979065284975994,
      "loss": 5.2375,
      "step": 3061
    },
    {
      "epoch": 0.4193371678992057,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008979025797609027,
      "loss": 5.1488,
      "step": 3062
    },
    {
      "epoch": 0.41947411668036155,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008978986273123307,
      "loss": 5.3182,
      "step": 3063
    },
    {
      "epoch": 0.4196110654615174,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000897894671151916,
      "loss": 5.2443,
      "step": 3064
    },
    {
      "epoch": 0.4197480142426732,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008978907112796914,
      "loss": 5.2489,
      "step": 3065
    },
    {
      "epoch": 0.4198849630238291,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008978867476956898,
      "loss": 5.2914,
      "step": 3066
    },
    {
      "epoch": 0.42002191180498494,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008978827803999441,
      "loss": 5.2519,
      "step": 3067
    },
    {
      "epoch": 0.42015886058614077,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000897878809392487,
      "loss": 5.2257,
      "step": 3068
    },
    {
      "epoch": 0.42029580936729666,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008978748346733515,
      "loss": 5.2583,
      "step": 3069
    },
    {
      "epoch": 0.4204327581484525,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008978708562425705,
      "loss": 5.2553,
      "step": 3070
    },
    {
      "epoch": 0.4205697069296083,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008978668741001769,
      "loss": 5.3145,
      "step": 3071
    },
    {
      "epoch": 0.42070665571076415,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008978628882462038,
      "loss": 5.2266,
      "step": 3072
    },
    {
      "epoch": 0.42084360449192004,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008978588986806844,
      "loss": 5.2862,
      "step": 3073
    },
    {
      "epoch": 0.4209805532730759,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008978549054036515,
      "loss": 5.2395,
      "step": 3074
    },
    {
      "epoch": 0.4211175020542317,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008978509084151382,
      "loss": 5.238,
      "step": 3075
    },
    {
      "epoch": 0.42125445083538754,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008978469077151778,
      "loss": 5.247,
      "step": 3076
    },
    {
      "epoch": 0.4213913996165434,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008978429033038033,
      "loss": 5.2116,
      "step": 3077
    },
    {
      "epoch": 0.42152834839769926,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008978388951810479,
      "loss": 5.253,
      "step": 3078
    },
    {
      "epoch": 0.4216652971788551,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000897834883346945,
      "loss": 5.24,
      "step": 3079
    },
    {
      "epoch": 0.421802245960011,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008978308678015275,
      "loss": 5.2057,
      "step": 3080
    },
    {
      "epoch": 0.4219391947411668,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008978268485448291,
      "loss": 5.2327,
      "step": 3081
    },
    {
      "epoch": 0.42207614352232264,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000897822825576883,
      "loss": 5.251,
      "step": 3082
    },
    {
      "epoch": 0.4222130923034785,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008978187988977222,
      "loss": 5.2364,
      "step": 3083
    },
    {
      "epoch": 0.42235004108463436,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008978147685073803,
      "loss": 5.2858,
      "step": 3084
    },
    {
      "epoch": 0.4224869898657902,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008978107344058907,
      "loss": 5.2269,
      "step": 3085
    },
    {
      "epoch": 0.42262393864694603,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000897806696593287,
      "loss": 5.2504,
      "step": 3086
    },
    {
      "epoch": 0.4227608874281019,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008978026550696024,
      "loss": 5.22,
      "step": 3087
    },
    {
      "epoch": 0.42289783620925775,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008977986098348704,
      "loss": 5.2056,
      "step": 3088
    },
    {
      "epoch": 0.4230347849904136,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008977945608891248,
      "loss": 5.3539,
      "step": 3089
    },
    {
      "epoch": 0.4231717337715694,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008977905082323989,
      "loss": 5.2122,
      "step": 3090
    },
    {
      "epoch": 0.4233086825527253,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008977864518647263,
      "loss": 5.1894,
      "step": 3091
    },
    {
      "epoch": 0.42344563133388113,
      "grad_norm": 0.75,
      "learning_rate": 0.0008977823917861407,
      "loss": 5.2646,
      "step": 3092
    },
    {
      "epoch": 0.42358258011503697,
      "grad_norm": 0.75,
      "learning_rate": 0.0008977783279966756,
      "loss": 5.2352,
      "step": 3093
    },
    {
      "epoch": 0.4237195288961928,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008977742604963651,
      "loss": 5.2785,
      "step": 3094
    },
    {
      "epoch": 0.4238564776773487,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008977701892852423,
      "loss": 5.1531,
      "step": 3095
    },
    {
      "epoch": 0.4239934264585045,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008977661143633414,
      "loss": 5.2139,
      "step": 3096
    },
    {
      "epoch": 0.42413037523966035,
      "grad_norm": 0.515625,
      "learning_rate": 0.000897762035730696,
      "loss": 5.2651,
      "step": 3097
    },
    {
      "epoch": 0.42426732402081624,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008977579533873399,
      "loss": 5.2249,
      "step": 3098
    },
    {
      "epoch": 0.42440427280197207,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008977538673333069,
      "loss": 5.2225,
      "step": 3099
    },
    {
      "epoch": 0.4245412215831279,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008977497775686309,
      "loss": 5.315,
      "step": 3100
    },
    {
      "epoch": 0.42467817036428374,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008977456840933459,
      "loss": 5.1567,
      "step": 3101
    },
    {
      "epoch": 0.4248151191454396,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008977415869074857,
      "loss": 5.215,
      "step": 3102
    },
    {
      "epoch": 0.42495206792659546,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008977374860110843,
      "loss": 5.2695,
      "step": 3103
    },
    {
      "epoch": 0.4250890167077513,
      "grad_norm": 0.625,
      "learning_rate": 0.0008977333814041756,
      "loss": 5.2893,
      "step": 3104
    },
    {
      "epoch": 0.4252259654889072,
      "grad_norm": 0.625,
      "learning_rate": 0.0008977292730867936,
      "loss": 5.2356,
      "step": 3105
    },
    {
      "epoch": 0.425362914270063,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008977251610589726,
      "loss": 5.3022,
      "step": 3106
    },
    {
      "epoch": 0.42549986305121884,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008977210453207464,
      "loss": 5.2437,
      "step": 3107
    },
    {
      "epoch": 0.4256368118323747,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008977169258721491,
      "loss": 5.2686,
      "step": 3108
    },
    {
      "epoch": 0.42577376061353056,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008977128027132152,
      "loss": 5.2816,
      "step": 3109
    },
    {
      "epoch": 0.4259107093946864,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008977086758439785,
      "loss": 5.2778,
      "step": 3110
    },
    {
      "epoch": 0.4260476581758422,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008977045452644733,
      "loss": 5.2249,
      "step": 3111
    },
    {
      "epoch": 0.42618460695699806,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008977004109747339,
      "loss": 5.2375,
      "step": 3112
    },
    {
      "epoch": 0.42632155573815395,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008976962729747945,
      "loss": 5.3,
      "step": 3113
    },
    {
      "epoch": 0.4264585045193098,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008976921312646894,
      "loss": 5.2681,
      "step": 3114
    },
    {
      "epoch": 0.4265954533004656,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008976879858444529,
      "loss": 5.2344,
      "step": 3115
    },
    {
      "epoch": 0.4267324020816215,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0008976838367141194,
      "loss": 5.2419,
      "step": 3116
    },
    {
      "epoch": 0.42686935086277733,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008976796838737232,
      "loss": 5.2139,
      "step": 3117
    },
    {
      "epoch": 0.42700629964393316,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000897675527323299,
      "loss": 5.2951,
      "step": 3118
    },
    {
      "epoch": 0.427143248425089,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008976713670628808,
      "loss": 5.2798,
      "step": 3119
    },
    {
      "epoch": 0.4272801972062449,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008976672030925034,
      "loss": 5.2607,
      "step": 3120
    },
    {
      "epoch": 0.4274171459874007,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008976630354122012,
      "loss": 5.189,
      "step": 3121
    },
    {
      "epoch": 0.42755409476855655,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008976588640220087,
      "loss": 5.1788,
      "step": 3122
    },
    {
      "epoch": 0.42769104354971244,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008976546889219607,
      "loss": 5.1945,
      "step": 3123
    },
    {
      "epoch": 0.42782799233086827,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008976505101120915,
      "loss": 5.2246,
      "step": 3124
    },
    {
      "epoch": 0.4279649411120241,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008976463275924358,
      "loss": 5.2565,
      "step": 3125
    },
    {
      "epoch": 0.42810188989317993,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008976421413630284,
      "loss": 5.2104,
      "step": 3126
    },
    {
      "epoch": 0.4282388386743358,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008976379514239039,
      "loss": 5.1921,
      "step": 3127
    },
    {
      "epoch": 0.42837578745549165,
      "grad_norm": 0.625,
      "learning_rate": 0.0008976337577750969,
      "loss": 5.164,
      "step": 3128
    },
    {
      "epoch": 0.4285127362366475,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008976295604166425,
      "loss": 5.2088,
      "step": 3129
    },
    {
      "epoch": 0.4286496850178033,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008976253593485751,
      "loss": 5.18,
      "step": 3130
    },
    {
      "epoch": 0.4287866337989592,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008976211545709299,
      "loss": 5.2371,
      "step": 3131
    },
    {
      "epoch": 0.42892358258011504,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008976169460837412,
      "loss": 5.2598,
      "step": 3132
    },
    {
      "epoch": 0.42906053136127087,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008976127338870444,
      "loss": 5.2971,
      "step": 3133
    },
    {
      "epoch": 0.42919748014242676,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008976085179808743,
      "loss": 5.2806,
      "step": 3134
    },
    {
      "epoch": 0.4293344289235826,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008976042983652655,
      "loss": 5.2226,
      "step": 3135
    },
    {
      "epoch": 0.4294713777047384,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008976000750402534,
      "loss": 5.2399,
      "step": 3136
    },
    {
      "epoch": 0.42960832648589425,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008975958480058729,
      "loss": 5.2512,
      "step": 3137
    },
    {
      "epoch": 0.42974527526705014,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008975916172621588,
      "loss": 5.1259,
      "step": 3138
    },
    {
      "epoch": 0.429882224048206,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008975873828091463,
      "loss": 5.2327,
      "step": 3139
    },
    {
      "epoch": 0.4300191728293618,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008975831446468706,
      "loss": 5.2454,
      "step": 3140
    },
    {
      "epoch": 0.43015612161051764,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008975789027753667,
      "loss": 5.2361,
      "step": 3141
    },
    {
      "epoch": 0.4302930703916735,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008975746571946697,
      "loss": 5.2477,
      "step": 3142
    },
    {
      "epoch": 0.43043001917282936,
      "grad_norm": 0.40625,
      "learning_rate": 0.000897570407904815,
      "loss": 5.2225,
      "step": 3143
    },
    {
      "epoch": 0.4305669679539852,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008975661549058375,
      "loss": 5.1993,
      "step": 3144
    },
    {
      "epoch": 0.4307039167351411,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000897561898197773,
      "loss": 5.2998,
      "step": 3145
    },
    {
      "epoch": 0.4308408655162969,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008975576377806561,
      "loss": 5.216,
      "step": 3146
    },
    {
      "epoch": 0.43097781429745274,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008975533736545224,
      "loss": 5.2178,
      "step": 3147
    },
    {
      "epoch": 0.4311147630786086,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008975491058194073,
      "loss": 5.2165,
      "step": 3148
    },
    {
      "epoch": 0.43125171185976446,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008975448342753461,
      "loss": 5.1824,
      "step": 3149
    },
    {
      "epoch": 0.4313886606409203,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008975405590223742,
      "loss": 5.2346,
      "step": 3150
    },
    {
      "epoch": 0.43152560942207613,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000897536280060527,
      "loss": 5.2168,
      "step": 3151
    },
    {
      "epoch": 0.431662558203232,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008975319973898401,
      "loss": 5.2742,
      "step": 3152
    },
    {
      "epoch": 0.43179950698438785,
      "grad_norm": 0.46875,
      "learning_rate": 0.000897527711010349,
      "loss": 5.2619,
      "step": 3153
    },
    {
      "epoch": 0.4319364557655437,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000897523420922089,
      "loss": 5.2092,
      "step": 3154
    },
    {
      "epoch": 0.4320734045466995,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008975191271250957,
      "loss": 5.2208,
      "step": 3155
    },
    {
      "epoch": 0.4322103533278554,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008975148296194047,
      "loss": 5.2742,
      "step": 3156
    },
    {
      "epoch": 0.43234730210901123,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008975105284050518,
      "loss": 5.2573,
      "step": 3157
    },
    {
      "epoch": 0.43248425089016707,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008975062234820725,
      "loss": 5.2721,
      "step": 3158
    },
    {
      "epoch": 0.4326211996713229,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008975019148505025,
      "loss": 5.2066,
      "step": 3159
    },
    {
      "epoch": 0.4327581484524788,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008974976025103776,
      "loss": 5.2245,
      "step": 3160
    },
    {
      "epoch": 0.4328950972336346,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008974932864617333,
      "loss": 5.2764,
      "step": 3161
    },
    {
      "epoch": 0.43303204601479045,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008974889667046055,
      "loss": 5.2817,
      "step": 3162
    },
    {
      "epoch": 0.43316899479594634,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00089748464323903,
      "loss": 5.3186,
      "step": 3163
    },
    {
      "epoch": 0.43330594357710217,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008974803160650427,
      "loss": 5.2827,
      "step": 3164
    },
    {
      "epoch": 0.433442892358258,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008974759851826793,
      "loss": 5.27,
      "step": 3165
    },
    {
      "epoch": 0.43357984113941384,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008974716505919759,
      "loss": 5.2381,
      "step": 3166
    },
    {
      "epoch": 0.4337167899205697,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008974673122929683,
      "loss": 5.2725,
      "step": 3167
    },
    {
      "epoch": 0.43385373870172556,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008974629702856923,
      "loss": 5.3132,
      "step": 3168
    },
    {
      "epoch": 0.4339906874828814,
      "grad_norm": 0.453125,
      "learning_rate": 0.000897458624570184,
      "loss": 5.324,
      "step": 3169
    },
    {
      "epoch": 0.4341276362640373,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008974542751464797,
      "loss": 5.234,
      "step": 3170
    },
    {
      "epoch": 0.4342645850451931,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008974499220146151,
      "loss": 5.2253,
      "step": 3171
    },
    {
      "epoch": 0.43440153382634894,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008974455651746264,
      "loss": 5.2025,
      "step": 3172
    },
    {
      "epoch": 0.4345384826075048,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008974412046265497,
      "loss": 5.2489,
      "step": 3173
    },
    {
      "epoch": 0.43467543138866066,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000897436840370421,
      "loss": 5.1934,
      "step": 3174
    },
    {
      "epoch": 0.4348123801698165,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008974324724062766,
      "loss": 5.1975,
      "step": 3175
    },
    {
      "epoch": 0.4349493289509723,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008974281007341527,
      "loss": 5.2351,
      "step": 3176
    },
    {
      "epoch": 0.43508627773212816,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008974237253540857,
      "loss": 5.197,
      "step": 3177
    },
    {
      "epoch": 0.43522322651328404,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008974193462661115,
      "loss": 5.26,
      "step": 3178
    },
    {
      "epoch": 0.4353601752944399,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008974149634702666,
      "loss": 5.2469,
      "step": 3179
    },
    {
      "epoch": 0.4354971240755957,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008974105769665872,
      "loss": 5.2344,
      "step": 3180
    },
    {
      "epoch": 0.4356340728567516,
      "grad_norm": 0.75,
      "learning_rate": 0.0008974061867551099,
      "loss": 5.2541,
      "step": 3181
    },
    {
      "epoch": 0.43577102163790743,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008974017928358707,
      "loss": 5.1487,
      "step": 3182
    },
    {
      "epoch": 0.43590797041906326,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008973973952089064,
      "loss": 5.2784,
      "step": 3183
    },
    {
      "epoch": 0.4360449192002191,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008973929938742532,
      "loss": 5.2159,
      "step": 3184
    },
    {
      "epoch": 0.436181867981375,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008973885888319478,
      "loss": 5.247,
      "step": 3185
    },
    {
      "epoch": 0.4363188167625308,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008973841800820263,
      "loss": 5.2623,
      "step": 3186
    },
    {
      "epoch": 0.43645576554368665,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008973797676245257,
      "loss": 5.2119,
      "step": 3187
    },
    {
      "epoch": 0.43659271432484253,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008973753514594822,
      "loss": 5.1913,
      "step": 3188
    },
    {
      "epoch": 0.43672966310599837,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008973709315869327,
      "loss": 5.2579,
      "step": 3189
    },
    {
      "epoch": 0.4368666118871542,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008973665080069137,
      "loss": 5.2066,
      "step": 3190
    },
    {
      "epoch": 0.43700356066831003,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008973620807194618,
      "loss": 5.2106,
      "step": 3191
    },
    {
      "epoch": 0.4371405094494659,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008973576497246137,
      "loss": 5.2594,
      "step": 3192
    },
    {
      "epoch": 0.43727745823062175,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008973532150224062,
      "loss": 5.293,
      "step": 3193
    },
    {
      "epoch": 0.4374144070117776,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000897348776612876,
      "loss": 5.2618,
      "step": 3194
    },
    {
      "epoch": 0.4375513557929334,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008973443344960599,
      "loss": 5.2648,
      "step": 3195
    },
    {
      "epoch": 0.4376883045740893,
      "grad_norm": 0.625,
      "learning_rate": 0.0008973398886719948,
      "loss": 5.1681,
      "step": 3196
    },
    {
      "epoch": 0.43782525335524514,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008973354391407171,
      "loss": 5.2803,
      "step": 3197
    },
    {
      "epoch": 0.43796220213640097,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008973309859022643,
      "loss": 5.1682,
      "step": 3198
    },
    {
      "epoch": 0.43809915091755686,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008973265289566731,
      "loss": 5.2018,
      "step": 3199
    },
    {
      "epoch": 0.4382360996987127,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008973220683039802,
      "loss": 5.227,
      "step": 3200
    },
    {
      "epoch": 0.4383730484798685,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008973176039442228,
      "loss": 5.281,
      "step": 3201
    },
    {
      "epoch": 0.43850999726102435,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008973131358774377,
      "loss": 5.2116,
      "step": 3202
    },
    {
      "epoch": 0.43864694604218024,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008973086641036622,
      "loss": 5.1807,
      "step": 3203
    },
    {
      "epoch": 0.4387838948233361,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008973041886229332,
      "loss": 5.2358,
      "step": 3204
    },
    {
      "epoch": 0.4389208436044919,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008972997094352878,
      "loss": 5.2357,
      "step": 3205
    },
    {
      "epoch": 0.4390577923856478,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000897295226540763,
      "loss": 5.2913,
      "step": 3206
    },
    {
      "epoch": 0.4391947411668036,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008972907399393961,
      "loss": 5.303,
      "step": 3207
    },
    {
      "epoch": 0.43933168994795946,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008972862496312245,
      "loss": 5.2402,
      "step": 3208
    },
    {
      "epoch": 0.4394686387291153,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008972817556162849,
      "loss": 5.2319,
      "step": 3209
    },
    {
      "epoch": 0.4396055875102712,
      "grad_norm": 0.470703125,
      "learning_rate": 0.000897277257894615,
      "loss": 5.2197,
      "step": 3210
    },
    {
      "epoch": 0.439742536291427,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008972727564662517,
      "loss": 5.2649,
      "step": 3211
    },
    {
      "epoch": 0.43987948507258284,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008972682513312326,
      "loss": 5.1966,
      "step": 3212
    },
    {
      "epoch": 0.4400164338537387,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008972637424895948,
      "loss": 5.1986,
      "step": 3213
    },
    {
      "epoch": 0.44015338263489456,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008972592299413759,
      "loss": 5.1712,
      "step": 3214
    },
    {
      "epoch": 0.4402903314160504,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008972547136866132,
      "loss": 5.2354,
      "step": 3215
    },
    {
      "epoch": 0.44042728019720623,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008972501937253441,
      "loss": 5.1265,
      "step": 3216
    },
    {
      "epoch": 0.4405642289783621,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000897245670057606,
      "loss": 5.2609,
      "step": 3217
    },
    {
      "epoch": 0.44070117775951795,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008972411426834366,
      "loss": 5.1503,
      "step": 3218
    },
    {
      "epoch": 0.4408381265406738,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008972366116028732,
      "loss": 5.2459,
      "step": 3219
    },
    {
      "epoch": 0.4409750753218296,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008972320768159533,
      "loss": 5.2598,
      "step": 3220
    },
    {
      "epoch": 0.4411120241029855,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008972275383227148,
      "loss": 5.1851,
      "step": 3221
    },
    {
      "epoch": 0.44124897288414133,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000897222996123195,
      "loss": 5.257,
      "step": 3222
    },
    {
      "epoch": 0.44138592166529717,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008972184502174317,
      "loss": 5.2314,
      "step": 3223
    },
    {
      "epoch": 0.44152287044645305,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008972139006054625,
      "loss": 5.2059,
      "step": 3224
    },
    {
      "epoch": 0.4416598192276089,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008972093472873252,
      "loss": 5.2437,
      "step": 3225
    },
    {
      "epoch": 0.4417967680087647,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008972047902630575,
      "loss": 5.2418,
      "step": 3226
    },
    {
      "epoch": 0.44193371678992055,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008972002295326971,
      "loss": 5.2237,
      "step": 3227
    },
    {
      "epoch": 0.44207066557107644,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008971956650962818,
      "loss": 5.2337,
      "step": 3228
    },
    {
      "epoch": 0.44220761435223227,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008971910969538496,
      "loss": 5.1848,
      "step": 3229
    },
    {
      "epoch": 0.4423445631333881,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008971865251054381,
      "loss": 5.1928,
      "step": 3230
    },
    {
      "epoch": 0.44248151191454393,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008971819495510853,
      "loss": 5.2533,
      "step": 3231
    },
    {
      "epoch": 0.4426184606956998,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008971773702908291,
      "loss": 5.2132,
      "step": 3232
    },
    {
      "epoch": 0.44275540947685565,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008971727873247077,
      "loss": 5.23,
      "step": 3233
    },
    {
      "epoch": 0.4428923582580115,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008971682006527586,
      "loss": 5.2447,
      "step": 3234
    },
    {
      "epoch": 0.4430293070391674,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008971636102750201,
      "loss": 5.207,
      "step": 3235
    },
    {
      "epoch": 0.4431662558203232,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008971590161915302,
      "loss": 5.2854,
      "step": 3236
    },
    {
      "epoch": 0.44330320460147904,
      "grad_norm": 0.546875,
      "learning_rate": 0.000897154418402327,
      "loss": 5.2193,
      "step": 3237
    },
    {
      "epoch": 0.44344015338263487,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008971498169074488,
      "loss": 5.1916,
      "step": 3238
    },
    {
      "epoch": 0.44357710216379076,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008971452117069333,
      "loss": 5.2109,
      "step": 3239
    },
    {
      "epoch": 0.4437140509449466,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008971406028008189,
      "loss": 5.2347,
      "step": 3240
    },
    {
      "epoch": 0.4438509997261024,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008971359901891439,
      "loss": 5.2323,
      "step": 3241
    },
    {
      "epoch": 0.4439879485072583,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008971313738719462,
      "loss": 5.1612,
      "step": 3242
    },
    {
      "epoch": 0.44412489728841414,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008971267538492643,
      "loss": 5.2027,
      "step": 3243
    },
    {
      "epoch": 0.44426184606957,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008971221301211366,
      "loss": 5.2703,
      "step": 3244
    },
    {
      "epoch": 0.4443987948507258,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008971175026876012,
      "loss": 5.2213,
      "step": 3245
    },
    {
      "epoch": 0.4445357436318817,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008971128715486964,
      "loss": 5.3071,
      "step": 3246
    },
    {
      "epoch": 0.44467269241303753,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008971082367044608,
      "loss": 5.2046,
      "step": 3247
    },
    {
      "epoch": 0.44480964119419336,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008971035981549328,
      "loss": 5.2236,
      "step": 3248
    },
    {
      "epoch": 0.4449465899753492,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008970989559001506,
      "loss": 5.1989,
      "step": 3249
    },
    {
      "epoch": 0.4450835387565051,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008970943099401527,
      "loss": 5.2249,
      "step": 3250
    },
    {
      "epoch": 0.4452204875376609,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000897089660274978,
      "loss": 5.2337,
      "step": 3251
    },
    {
      "epoch": 0.44535743631881675,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008970850069046645,
      "loss": 5.2026,
      "step": 3252
    },
    {
      "epoch": 0.44549438509997263,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008970803498292511,
      "loss": 5.18,
      "step": 3253
    },
    {
      "epoch": 0.44563133388112847,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008970756890487765,
      "loss": 5.286,
      "step": 3254
    },
    {
      "epoch": 0.4457682826622843,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008970710245632788,
      "loss": 5.1818,
      "step": 3255
    },
    {
      "epoch": 0.44590523144344013,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008970663563727972,
      "loss": 5.226,
      "step": 3256
    },
    {
      "epoch": 0.446042180224596,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008970616844773701,
      "loss": 5.2261,
      "step": 3257
    },
    {
      "epoch": 0.44617912900575185,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008970570088770364,
      "loss": 5.2101,
      "step": 3258
    },
    {
      "epoch": 0.4463160777869077,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008970523295718345,
      "loss": 5.1914,
      "step": 3259
    },
    {
      "epoch": 0.44645302656806357,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008970476465618038,
      "loss": 5.249,
      "step": 3260
    },
    {
      "epoch": 0.4465899753492194,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008970429598469823,
      "loss": 5.2148,
      "step": 3261
    },
    {
      "epoch": 0.44672692413037524,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008970382694274095,
      "loss": 5.2669,
      "step": 3262
    },
    {
      "epoch": 0.44686387291153107,
      "grad_norm": 0.625,
      "learning_rate": 0.000897033575303124,
      "loss": 5.2279,
      "step": 3263
    },
    {
      "epoch": 0.44700082169268696,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008970288774741649,
      "loss": 5.2232,
      "step": 3264
    },
    {
      "epoch": 0.4471377704738428,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008970241759405707,
      "loss": 5.2396,
      "step": 3265
    },
    {
      "epoch": 0.4472747192549986,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008970194707023807,
      "loss": 5.2417,
      "step": 3266
    },
    {
      "epoch": 0.44741166803615445,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008970147617596339,
      "loss": 5.2649,
      "step": 3267
    },
    {
      "epoch": 0.44754861681731034,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008970100491123692,
      "loss": 5.2404,
      "step": 3268
    },
    {
      "epoch": 0.4476855655984662,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008970053327606258,
      "loss": 5.2465,
      "step": 3269
    },
    {
      "epoch": 0.447822514379622,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008970006127044426,
      "loss": 5.2076,
      "step": 3270
    },
    {
      "epoch": 0.4479594631607779,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008969958889438588,
      "loss": 5.1549,
      "step": 3271
    },
    {
      "epoch": 0.4480964119419337,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008969911614789136,
      "loss": 5.2825,
      "step": 3272
    },
    {
      "epoch": 0.44823336072308956,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008969864303096461,
      "loss": 5.2177,
      "step": 3273
    },
    {
      "epoch": 0.4483703095042454,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008969816954360956,
      "loss": 5.2276,
      "step": 3274
    },
    {
      "epoch": 0.4485072582854013,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008969769568583011,
      "loss": 5.2535,
      "step": 3275
    },
    {
      "epoch": 0.4486442070665571,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008969722145763023,
      "loss": 5.2172,
      "step": 3276
    },
    {
      "epoch": 0.44878115584771294,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008969674685901382,
      "loss": 5.1607,
      "step": 3277
    },
    {
      "epoch": 0.44891810462886883,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008969627188998481,
      "loss": 5.2277,
      "step": 3278
    },
    {
      "epoch": 0.44905505341002466,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0008969579655054713,
      "loss": 5.3053,
      "step": 3279
    },
    {
      "epoch": 0.4491920021911805,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008969532084070476,
      "loss": 5.2338,
      "step": 3280
    },
    {
      "epoch": 0.4493289509723363,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008969484476046161,
      "loss": 5.2277,
      "step": 3281
    },
    {
      "epoch": 0.4494658997534922,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0008969436830982163,
      "loss": 5.1333,
      "step": 3282
    },
    {
      "epoch": 0.44960284853464805,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008969389148878877,
      "loss": 5.1735,
      "step": 3283
    },
    {
      "epoch": 0.4497397973158039,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008969341429736697,
      "loss": 5.206,
      "step": 3284
    },
    {
      "epoch": 0.4498767460969597,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008969293673556021,
      "loss": 5.177,
      "step": 3285
    },
    {
      "epoch": 0.4500136948781156,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008969245880337243,
      "loss": 5.2682,
      "step": 3286
    },
    {
      "epoch": 0.45015064365927143,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008969198050080759,
      "loss": 5.2719,
      "step": 3287
    },
    {
      "epoch": 0.45028759244042726,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008969150182786965,
      "loss": 5.192,
      "step": 3288
    },
    {
      "epoch": 0.45042454122158315,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008969102278456259,
      "loss": 5.173,
      "step": 3289
    },
    {
      "epoch": 0.450561490002739,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008969054337089039,
      "loss": 5.2279,
      "step": 3290
    },
    {
      "epoch": 0.4506984387838948,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00089690063586857,
      "loss": 5.2151,
      "step": 3291
    },
    {
      "epoch": 0.45083538756505065,
      "grad_norm": 0.6015625,
      "learning_rate": 0.000896895834324664,
      "loss": 5.1879,
      "step": 3292
    },
    {
      "epoch": 0.45097233634620654,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008968910290772257,
      "loss": 5.2482,
      "step": 3293
    },
    {
      "epoch": 0.45110928512736237,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008968862201262951,
      "loss": 5.2899,
      "step": 3294
    },
    {
      "epoch": 0.4512462339085182,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008968814074719118,
      "loss": 5.223,
      "step": 3295
    },
    {
      "epoch": 0.4513831826896741,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008968765911141158,
      "loss": 5.2066,
      "step": 3296
    },
    {
      "epoch": 0.4515201314708299,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008968717710529469,
      "loss": 5.2005,
      "step": 3297
    },
    {
      "epoch": 0.45165708025198575,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008968669472884453,
      "loss": 5.1866,
      "step": 3298
    },
    {
      "epoch": 0.4517940290331416,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008968621198206508,
      "loss": 5.2393,
      "step": 3299
    },
    {
      "epoch": 0.4519309778142975,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008968572886496034,
      "loss": 5.1424,
      "step": 3300
    },
    {
      "epoch": 0.4520679265954533,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008968524537753432,
      "loss": 5.2602,
      "step": 3301
    },
    {
      "epoch": 0.45220487537660914,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008968476151979101,
      "loss": 5.2015,
      "step": 3302
    },
    {
      "epoch": 0.45234182415776497,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008968427729173446,
      "loss": 5.2503,
      "step": 3303
    },
    {
      "epoch": 0.45247877293892086,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008968379269336863,
      "loss": 5.2527,
      "step": 3304
    },
    {
      "epoch": 0.4526157217200767,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008968330772469758,
      "loss": 5.2453,
      "step": 3305
    },
    {
      "epoch": 0.4527526705012325,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008968282238572531,
      "loss": 5.1811,
      "step": 3306
    },
    {
      "epoch": 0.4528896192823884,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008968233667645582,
      "loss": 5.2151,
      "step": 3307
    },
    {
      "epoch": 0.45302656806354424,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008968185059689319,
      "loss": 5.122,
      "step": 3308
    },
    {
      "epoch": 0.4531635168447001,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008968136414704141,
      "loss": 5.1601,
      "step": 3309
    },
    {
      "epoch": 0.4533004656258559,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008968087732690451,
      "loss": 5.2109,
      "step": 3310
    },
    {
      "epoch": 0.4534374144070118,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008968039013648654,
      "loss": 5.2153,
      "step": 3311
    },
    {
      "epoch": 0.45357436318816763,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008967990257579152,
      "loss": 5.2482,
      "step": 3312
    },
    {
      "epoch": 0.45371131196932346,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008967941464482351,
      "loss": 5.2384,
      "step": 3313
    },
    {
      "epoch": 0.4538482607504793,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008967892634358653,
      "loss": 5.2649,
      "step": 3314
    },
    {
      "epoch": 0.4539852095316352,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008967843767208467,
      "loss": 5.1831,
      "step": 3315
    },
    {
      "epoch": 0.454122158312791,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008967794863032192,
      "loss": 5.2272,
      "step": 3316
    },
    {
      "epoch": 0.45425910709394685,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0008967745921830238,
      "loss": 5.2303,
      "step": 3317
    },
    {
      "epoch": 0.45439605587510273,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008967696943603009,
      "loss": 5.1754,
      "step": 3318
    },
    {
      "epoch": 0.45453300465625857,
      "grad_norm": 0.546875,
      "learning_rate": 0.000896764792835091,
      "loss": 5.2194,
      "step": 3319
    },
    {
      "epoch": 0.4546699534374144,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008967598876074348,
      "loss": 5.1982,
      "step": 3320
    },
    {
      "epoch": 0.45480690221857023,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000896754978677373,
      "loss": 5.2344,
      "step": 3321
    },
    {
      "epoch": 0.4549438509997261,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008967500660449462,
      "loss": 5.222,
      "step": 3322
    },
    {
      "epoch": 0.45508079978088195,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008967451497101952,
      "loss": 5.2955,
      "step": 3323
    },
    {
      "epoch": 0.4552177485620378,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008967402296731607,
      "loss": 5.1963,
      "step": 3324
    },
    {
      "epoch": 0.45535469734319367,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008967353059338834,
      "loss": 5.2376,
      "step": 3325
    },
    {
      "epoch": 0.4554916461243495,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008967303784924042,
      "loss": 5.1906,
      "step": 3326
    },
    {
      "epoch": 0.45562859490550534,
      "grad_norm": 0.5390625,
      "learning_rate": 0.000896725447348764,
      "loss": 5.1845,
      "step": 3327
    },
    {
      "epoch": 0.45576554368666117,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008967205125030033,
      "loss": 5.1122,
      "step": 3328
    },
    {
      "epoch": 0.45590249246781706,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008967155739551634,
      "loss": 5.1931,
      "step": 3329
    },
    {
      "epoch": 0.4560394412489729,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008967106317052853,
      "loss": 5.2304,
      "step": 3330
    },
    {
      "epoch": 0.4561763900301287,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008967056857534095,
      "loss": 5.2327,
      "step": 3331
    },
    {
      "epoch": 0.45631333881128455,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008967007360995773,
      "loss": 5.2997,
      "step": 3332
    },
    {
      "epoch": 0.45645028759244044,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008966957827438297,
      "loss": 5.2363,
      "step": 3333
    },
    {
      "epoch": 0.4565872363735963,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008966908256862077,
      "loss": 5.2196,
      "step": 3334
    },
    {
      "epoch": 0.4567241851547521,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008966858649267523,
      "loss": 5.1931,
      "step": 3335
    },
    {
      "epoch": 0.456861133935908,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008966809004655047,
      "loss": 5.2706,
      "step": 3336
    },
    {
      "epoch": 0.4569980827170638,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008966759323025061,
      "loss": 5.2257,
      "step": 3337
    },
    {
      "epoch": 0.45713503149821966,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008966709604377978,
      "loss": 5.2734,
      "step": 3338
    },
    {
      "epoch": 0.4572719802793755,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008966659848714206,
      "loss": 5.2069,
      "step": 3339
    },
    {
      "epoch": 0.4574089290605314,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000896661005603416,
      "loss": 5.1652,
      "step": 3340
    },
    {
      "epoch": 0.4575458778416872,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008966560226338252,
      "loss": 5.2145,
      "step": 3341
    },
    {
      "epoch": 0.45768282662284304,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008966510359626895,
      "loss": 5.2481,
      "step": 3342
    },
    {
      "epoch": 0.45781977540399893,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008966460455900503,
      "loss": 5.2421,
      "step": 3343
    },
    {
      "epoch": 0.45795672418515476,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008966410515159486,
      "loss": 5.1687,
      "step": 3344
    },
    {
      "epoch": 0.4580936729663106,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008966360537404264,
      "loss": 5.2199,
      "step": 3345
    },
    {
      "epoch": 0.4582306217474664,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008966310522635248,
      "loss": 5.1756,
      "step": 3346
    },
    {
      "epoch": 0.4583675705286223,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008966260470852851,
      "loss": 5.1592,
      "step": 3347
    },
    {
      "epoch": 0.45850451930977815,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000896621038205749,
      "loss": 5.2149,
      "step": 3348
    },
    {
      "epoch": 0.458641468090934,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008966160256249578,
      "loss": 5.2057,
      "step": 3349
    },
    {
      "epoch": 0.4587784168720898,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008966110093429533,
      "loss": 5.1942,
      "step": 3350
    },
    {
      "epoch": 0.4589153656532457,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008966059893597769,
      "loss": 5.2386,
      "step": 3351
    },
    {
      "epoch": 0.45905231443440153,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008966009656754703,
      "loss": 5.2528,
      "step": 3352
    },
    {
      "epoch": 0.45918926321555736,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008965959382900751,
      "loss": 5.2082,
      "step": 3353
    },
    {
      "epoch": 0.45932621199671325,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008965909072036328,
      "loss": 5.2254,
      "step": 3354
    },
    {
      "epoch": 0.4594631607778691,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008965858724161853,
      "loss": 5.191,
      "step": 3355
    },
    {
      "epoch": 0.4596001095590249,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008965808339277743,
      "loss": 5.1311,
      "step": 3356
    },
    {
      "epoch": 0.45973705834018075,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008965757917384416,
      "loss": 5.1987,
      "step": 3357
    },
    {
      "epoch": 0.45987400712133664,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008965707458482286,
      "loss": 5.2614,
      "step": 3358
    },
    {
      "epoch": 0.46001095590249247,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008965656962571778,
      "loss": 5.1927,
      "step": 3359
    },
    {
      "epoch": 0.4601479046836483,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008965606429653304,
      "loss": 5.234,
      "step": 3360
    },
    {
      "epoch": 0.4602848534648042,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008965555859727285,
      "loss": 5.2739,
      "step": 3361
    },
    {
      "epoch": 0.46042180224596,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008965505252794143,
      "loss": 5.2709,
      "step": 3362
    },
    {
      "epoch": 0.46055875102711585,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008965454608854293,
      "loss": 5.2104,
      "step": 3363
    },
    {
      "epoch": 0.4606956998082717,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008965403927908157,
      "loss": 5.2795,
      "step": 3364
    },
    {
      "epoch": 0.4608326485894276,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008965353209956156,
      "loss": 5.1605,
      "step": 3365
    },
    {
      "epoch": 0.4609695973705834,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008965302454998706,
      "loss": 5.2443,
      "step": 3366
    },
    {
      "epoch": 0.46110654615173924,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008965251663036232,
      "loss": 5.1831,
      "step": 3367
    },
    {
      "epoch": 0.46124349493289507,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008965200834069154,
      "loss": 5.2429,
      "step": 3368
    },
    {
      "epoch": 0.46138044371405096,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008965149968097893,
      "loss": 5.1771,
      "step": 3369
    },
    {
      "epoch": 0.4615173924952068,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008965099065122869,
      "loss": 5.0944,
      "step": 3370
    },
    {
      "epoch": 0.4616543412763626,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008965048125144506,
      "loss": 5.1912,
      "step": 3371
    },
    {
      "epoch": 0.4617912900575185,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008964997148163226,
      "loss": 5.2901,
      "step": 3372
    },
    {
      "epoch": 0.46192823883867434,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008964946134179449,
      "loss": 5.132,
      "step": 3373
    },
    {
      "epoch": 0.4620651876198302,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00089648950831936,
      "loss": 5.228,
      "step": 3374
    },
    {
      "epoch": 0.462202136400986,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008964843995206101,
      "loss": 5.2844,
      "step": 3375
    },
    {
      "epoch": 0.4623390851821419,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008964792870217377,
      "loss": 5.1969,
      "step": 3376
    },
    {
      "epoch": 0.46247603396329773,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008964741708227849,
      "loss": 5.2665,
      "step": 3377
    },
    {
      "epoch": 0.46261298274445356,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008964690509237944,
      "loss": 5.1926,
      "step": 3378
    },
    {
      "epoch": 0.46274993152560945,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008964639273248084,
      "loss": 5.1564,
      "step": 3379
    },
    {
      "epoch": 0.4628868803067653,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008964588000258695,
      "loss": 5.1747,
      "step": 3380
    },
    {
      "epoch": 0.4630238290879211,
      "grad_norm": 0.484375,
      "learning_rate": 0.00089645366902702,
      "loss": 5.1507,
      "step": 3381
    },
    {
      "epoch": 0.46316077786907695,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008964485343283028,
      "loss": 5.2215,
      "step": 3382
    },
    {
      "epoch": 0.46329772665023283,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008964433959297601,
      "loss": 5.2042,
      "step": 3383
    },
    {
      "epoch": 0.46343467543138867,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008964382538314347,
      "loss": 5.1506,
      "step": 3384
    },
    {
      "epoch": 0.4635716242125445,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008964331080333689,
      "loss": 5.2167,
      "step": 3385
    },
    {
      "epoch": 0.46370857299370033,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008964279585356057,
      "loss": 5.0436,
      "step": 3386
    },
    {
      "epoch": 0.4638455217748562,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008964228053381876,
      "loss": 5.2903,
      "step": 3387
    },
    {
      "epoch": 0.46398247055601205,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008964176484411574,
      "loss": 5.2358,
      "step": 3388
    },
    {
      "epoch": 0.4641194193371679,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008964124878445576,
      "loss": 5.2115,
      "step": 3389
    },
    {
      "epoch": 0.46425636811832377,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008964073235484313,
      "loss": 5.1951,
      "step": 3390
    },
    {
      "epoch": 0.4643933168994796,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000896402155552821,
      "loss": 5.1894,
      "step": 3391
    },
    {
      "epoch": 0.46453026568063543,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008963969838577698,
      "loss": 5.1626,
      "step": 3392
    },
    {
      "epoch": 0.46466721446179127,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008963918084633204,
      "loss": 5.2152,
      "step": 3393
    },
    {
      "epoch": 0.46480416324294715,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008963866293695156,
      "loss": 5.2161,
      "step": 3394
    },
    {
      "epoch": 0.464941112024103,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008963814465763985,
      "loss": 5.1988,
      "step": 3395
    },
    {
      "epoch": 0.4650780608052588,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008963762600840121,
      "loss": 5.2452,
      "step": 3396
    },
    {
      "epoch": 0.4652150095864147,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008963710698923991,
      "loss": 5.2451,
      "step": 3397
    },
    {
      "epoch": 0.46535195836757054,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008963658760016028,
      "loss": 5.2424,
      "step": 3398
    },
    {
      "epoch": 0.46548890714872637,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000896360678411666,
      "loss": 5.2868,
      "step": 3399
    },
    {
      "epoch": 0.4656258559298822,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000896355477122632,
      "loss": 5.1678,
      "step": 3400
    },
    {
      "epoch": 0.4657628047110381,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008963502721345437,
      "loss": 5.223,
      "step": 3401
    },
    {
      "epoch": 0.4658997534921939,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008963450634474446,
      "loss": 5.2892,
      "step": 3402
    },
    {
      "epoch": 0.46603670227334976,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008963398510613773,
      "loss": 5.2245,
      "step": 3403
    },
    {
      "epoch": 0.4661736510545056,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008963346349763854,
      "loss": 5.1931,
      "step": 3404
    },
    {
      "epoch": 0.4663105998356615,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008963294151925121,
      "loss": 5.2212,
      "step": 3405
    },
    {
      "epoch": 0.4664475486168173,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008963241917098004,
      "loss": 5.2345,
      "step": 3406
    },
    {
      "epoch": 0.46658449739797314,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008963189645282938,
      "loss": 5.2133,
      "step": 3407
    },
    {
      "epoch": 0.46672144617912903,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0008963137336480357,
      "loss": 5.1926,
      "step": 3408
    },
    {
      "epoch": 0.46685839496028486,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008963084990690693,
      "loss": 5.1747,
      "step": 3409
    },
    {
      "epoch": 0.4669953437414407,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008963032607914381,
      "loss": 5.2121,
      "step": 3410
    },
    {
      "epoch": 0.4671322925225965,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008962980188151853,
      "loss": 5.2188,
      "step": 3411
    },
    {
      "epoch": 0.4672692413037524,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008962927731403546,
      "loss": 5.231,
      "step": 3412
    },
    {
      "epoch": 0.46740619008490825,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008962875237669893,
      "loss": 5.2097,
      "step": 3413
    },
    {
      "epoch": 0.4675431388660641,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008962822706951328,
      "loss": 5.1964,
      "step": 3414
    },
    {
      "epoch": 0.46768008764721997,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008962770139248289,
      "loss": 5.2737,
      "step": 3415
    },
    {
      "epoch": 0.4678170364283758,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000896271753456121,
      "loss": 5.1942,
      "step": 3416
    },
    {
      "epoch": 0.46795398520953163,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008962664892890528,
      "loss": 5.2317,
      "step": 3417
    },
    {
      "epoch": 0.46809093399068746,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008962612214236679,
      "loss": 5.2039,
      "step": 3418
    },
    {
      "epoch": 0.46822788277184335,
      "grad_norm": 0.5,
      "learning_rate": 0.0008962559498600099,
      "loss": 5.2431,
      "step": 3419
    },
    {
      "epoch": 0.4683648315529992,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008962506745981224,
      "loss": 5.1569,
      "step": 3420
    },
    {
      "epoch": 0.468501780334155,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008962453956380493,
      "loss": 5.276,
      "step": 3421
    },
    {
      "epoch": 0.46863872911531085,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008962401129798342,
      "loss": 5.2042,
      "step": 3422
    },
    {
      "epoch": 0.46877567789646674,
      "grad_norm": 0.51171875,
      "learning_rate": 0.000896234826623521,
      "loss": 5.1769,
      "step": 3423
    },
    {
      "epoch": 0.46891262667762257,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008962295365691536,
      "loss": 5.2425,
      "step": 3424
    },
    {
      "epoch": 0.4690495754587784,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008962242428167755,
      "loss": 5.2356,
      "step": 3425
    },
    {
      "epoch": 0.4691865242399343,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008962189453664308,
      "loss": 5.1783,
      "step": 3426
    },
    {
      "epoch": 0.4693234730210901,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008962136442181634,
      "loss": 5.2298,
      "step": 3427
    },
    {
      "epoch": 0.46946042180224595,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008962083393720172,
      "loss": 5.2457,
      "step": 3428
    },
    {
      "epoch": 0.4695973705834018,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008962030308280362,
      "loss": 5.0978,
      "step": 3429
    },
    {
      "epoch": 0.4697343193645577,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008961977185862643,
      "loss": 5.1921,
      "step": 3430
    },
    {
      "epoch": 0.4698712681457135,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008961924026467457,
      "loss": 5.2163,
      "step": 3431
    },
    {
      "epoch": 0.47000821692686934,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008961870830095243,
      "loss": 5.2207,
      "step": 3432
    },
    {
      "epoch": 0.4701451657080252,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008961817596746442,
      "loss": 5.2709,
      "step": 3433
    },
    {
      "epoch": 0.47028211448918106,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008961764326421494,
      "loss": 5.1189,
      "step": 3434
    },
    {
      "epoch": 0.4704190632703369,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008961711019120843,
      "loss": 5.1932,
      "step": 3435
    },
    {
      "epoch": 0.4705560120514927,
      "grad_norm": 0.78125,
      "learning_rate": 0.000896165767484493,
      "loss": 5.1561,
      "step": 3436
    },
    {
      "epoch": 0.4706929608326486,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008961604293594197,
      "loss": 5.224,
      "step": 3437
    },
    {
      "epoch": 0.47082990961380444,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008961550875369085,
      "loss": 5.1183,
      "step": 3438
    },
    {
      "epoch": 0.4709668583949603,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008961497420170038,
      "loss": 5.1772,
      "step": 3439
    },
    {
      "epoch": 0.4711038071761161,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008961443927997498,
      "loss": 5.1669,
      "step": 3440
    },
    {
      "epoch": 0.471240755957272,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008961390398851911,
      "loss": 5.2166,
      "step": 3441
    },
    {
      "epoch": 0.4713777047384278,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008961336832733717,
      "loss": 5.2005,
      "step": 3442
    },
    {
      "epoch": 0.47151465351958366,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008961283229643363,
      "loss": 5.2349,
      "step": 3443
    },
    {
      "epoch": 0.47165160230073955,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000896122958958129,
      "loss": 5.2718,
      "step": 3444
    },
    {
      "epoch": 0.4717885510818954,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008961175912547947,
      "loss": 5.2017,
      "step": 3445
    },
    {
      "epoch": 0.4719254998630512,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008961122198543773,
      "loss": 5.1974,
      "step": 3446
    },
    {
      "epoch": 0.47206244864420704,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008961068447569219,
      "loss": 5.2281,
      "step": 3447
    },
    {
      "epoch": 0.47219939742536293,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008961014659624727,
      "loss": 5.1112,
      "step": 3448
    },
    {
      "epoch": 0.47233634620651876,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008960960834710742,
      "loss": 5.2473,
      "step": 3449
    },
    {
      "epoch": 0.4724732949876746,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008960906972827713,
      "loss": 5.2348,
      "step": 3450
    },
    {
      "epoch": 0.4726102437688305,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008960853073976085,
      "loss": 5.1961,
      "step": 3451
    },
    {
      "epoch": 0.4727471925499863,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008960799138156304,
      "loss": 5.2153,
      "step": 3452
    },
    {
      "epoch": 0.47288414133114215,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008960745165368819,
      "loss": 5.0949,
      "step": 3453
    },
    {
      "epoch": 0.473021090112298,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008960691155614074,
      "loss": 5.1696,
      "step": 3454
    },
    {
      "epoch": 0.47315803889345387,
      "grad_norm": 0.482421875,
      "learning_rate": 0.000896063710889252,
      "loss": 5.1717,
      "step": 3455
    },
    {
      "epoch": 0.4732949876746097,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008960583025204602,
      "loss": 5.1524,
      "step": 3456
    },
    {
      "epoch": 0.47343193645576553,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000896052890455077,
      "loss": 5.1444,
      "step": 3457
    },
    {
      "epoch": 0.47356888523692137,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008960474746931473,
      "loss": 5.1975,
      "step": 3458
    },
    {
      "epoch": 0.47370583401807725,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008960420552347158,
      "loss": 5.2106,
      "step": 3459
    },
    {
      "epoch": 0.4738427827992331,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008960366320798275,
      "loss": 5.199,
      "step": 3460
    },
    {
      "epoch": 0.4739797315803889,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008960312052285274,
      "loss": 5.1839,
      "step": 3461
    },
    {
      "epoch": 0.4741166803615448,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008960257746808604,
      "loss": 5.1357,
      "step": 3462
    },
    {
      "epoch": 0.47425362914270064,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008960203404368715,
      "loss": 5.182,
      "step": 3463
    },
    {
      "epoch": 0.47439057792385647,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008960149024966057,
      "loss": 5.1978,
      "step": 3464
    },
    {
      "epoch": 0.4745275267050123,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008960094608601082,
      "loss": 5.207,
      "step": 3465
    },
    {
      "epoch": 0.4746644754861682,
      "grad_norm": 0.49609375,
      "learning_rate": 0.000896004015527424,
      "loss": 5.2027,
      "step": 3466
    },
    {
      "epoch": 0.474801424267324,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008959985664985983,
      "loss": 5.2102,
      "step": 3467
    },
    {
      "epoch": 0.47493837304847986,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008959931137736762,
      "loss": 5.1849,
      "step": 3468
    },
    {
      "epoch": 0.47507532182963574,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008959876573527028,
      "loss": 5.2408,
      "step": 3469
    },
    {
      "epoch": 0.4752122706107916,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008959821972357233,
      "loss": 5.1919,
      "step": 3470
    },
    {
      "epoch": 0.4753492193919474,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008959767334227833,
      "loss": 5.2658,
      "step": 3471
    },
    {
      "epoch": 0.47548616817310324,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008959712659139279,
      "loss": 5.1643,
      "step": 3472
    },
    {
      "epoch": 0.47562311695425913,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008959657947092022,
      "loss": 5.2681,
      "step": 3473
    },
    {
      "epoch": 0.47576006573541496,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008959603198086517,
      "loss": 5.1585,
      "step": 3474
    },
    {
      "epoch": 0.4758970145165708,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008959548412123216,
      "loss": 5.2162,
      "step": 3475
    },
    {
      "epoch": 0.4760339632977266,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008959493589202577,
      "loss": 5.2222,
      "step": 3476
    },
    {
      "epoch": 0.4761709120788825,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008959438729325052,
      "loss": 5.1987,
      "step": 3477
    },
    {
      "epoch": 0.47630786086003835,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008959383832491094,
      "loss": 5.2537,
      "step": 3478
    },
    {
      "epoch": 0.4764448096411942,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008959328898701161,
      "loss": 5.1533,
      "step": 3479
    },
    {
      "epoch": 0.47658175842235007,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008959273927955706,
      "loss": 5.1409,
      "step": 3480
    },
    {
      "epoch": 0.4767187072035059,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008959218920255186,
      "loss": 5.2173,
      "step": 3481
    },
    {
      "epoch": 0.47685565598466173,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008959163875600055,
      "loss": 5.2597,
      "step": 3482
    },
    {
      "epoch": 0.47699260476581756,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008959108793990772,
      "loss": 5.1584,
      "step": 3483
    },
    {
      "epoch": 0.47712955354697345,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008959053675427791,
      "loss": 5.2508,
      "step": 3484
    },
    {
      "epoch": 0.4772665023281293,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000895899851991157,
      "loss": 5.276,
      "step": 3485
    },
    {
      "epoch": 0.4774034511092851,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008958943327442566,
      "loss": 5.1617,
      "step": 3486
    },
    {
      "epoch": 0.47754039989044095,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008958888098021235,
      "loss": 5.1178,
      "step": 3487
    },
    {
      "epoch": 0.47767734867159684,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008958832831648037,
      "loss": 5.1622,
      "step": 3488
    },
    {
      "epoch": 0.47781429745275267,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008958777528323428,
      "loss": 5.1762,
      "step": 3489
    },
    {
      "epoch": 0.4779512462339085,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008958722188047868,
      "loss": 5.2653,
      "step": 3490
    },
    {
      "epoch": 0.4780881950150644,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008958666810821815,
      "loss": 5.2161,
      "step": 3491
    },
    {
      "epoch": 0.4782251437962202,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008958611396645725,
      "loss": 5.2195,
      "step": 3492
    },
    {
      "epoch": 0.47836209257737605,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008958555945520063,
      "loss": 5.1353,
      "step": 3493
    },
    {
      "epoch": 0.4784990413585319,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008958500457445284,
      "loss": 5.159,
      "step": 3494
    },
    {
      "epoch": 0.4786359901396878,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008958444932421851,
      "loss": 5.247,
      "step": 3495
    },
    {
      "epoch": 0.4787729389208436,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000895838937045022,
      "loss": 5.197,
      "step": 3496
    },
    {
      "epoch": 0.47890988770199944,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008958333771530856,
      "loss": 5.2205,
      "step": 3497
    },
    {
      "epoch": 0.4790468364831553,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008958278135664217,
      "loss": 5.1995,
      "step": 3498
    },
    {
      "epoch": 0.47918378526431116,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008958222462850763,
      "loss": 5.2515,
      "step": 3499
    },
    {
      "epoch": 0.479320734045467,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000895816675309096,
      "loss": 5.1611,
      "step": 3500
    },
    {
      "epoch": 0.4794576828266228,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008958111006385264,
      "loss": 5.2478,
      "step": 3501
    },
    {
      "epoch": 0.4795946316077787,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008958055222734141,
      "loss": 5.2342,
      "step": 3502
    },
    {
      "epoch": 0.47973158038893454,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008957999402138052,
      "loss": 5.203,
      "step": 3503
    },
    {
      "epoch": 0.4798685291700904,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008957943544597462,
      "loss": 5.2114,
      "step": 3504
    },
    {
      "epoch": 0.4800054779512462,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008957887650112828,
      "loss": 5.2255,
      "step": 3505
    },
    {
      "epoch": 0.4801424267324021,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008957831718684619,
      "loss": 5.1954,
      "step": 3506
    },
    {
      "epoch": 0.4802793755135579,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008957775750313296,
      "loss": 5.1465,
      "step": 3507
    },
    {
      "epoch": 0.48041632429471376,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008957719744999322,
      "loss": 5.1916,
      "step": 3508
    },
    {
      "epoch": 0.48055327307586965,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008957663702743163,
      "loss": 5.1478,
      "step": 3509
    },
    {
      "epoch": 0.4806902218570255,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008957607623545282,
      "loss": 5.2415,
      "step": 3510
    },
    {
      "epoch": 0.4808271706381813,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008957551507406146,
      "loss": 5.1741,
      "step": 3511
    },
    {
      "epoch": 0.48096411941933714,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008957495354326218,
      "loss": 5.1784,
      "step": 3512
    },
    {
      "epoch": 0.48110106820049303,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008957439164305963,
      "loss": 5.1801,
      "step": 3513
    },
    {
      "epoch": 0.48123801698164886,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008957382937345848,
      "loss": 5.204,
      "step": 3514
    },
    {
      "epoch": 0.4813749657628047,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008957326673446338,
      "loss": 5.1959,
      "step": 3515
    },
    {
      "epoch": 0.4815119145439606,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00089572703726079,
      "loss": 5.1723,
      "step": 3516
    },
    {
      "epoch": 0.4816488633251164,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008957214034831,
      "loss": 5.1789,
      "step": 3517
    },
    {
      "epoch": 0.48178581210627225,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008957157660116106,
      "loss": 5.1803,
      "step": 3518
    },
    {
      "epoch": 0.4819227608874281,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008957101248463685,
      "loss": 5.2093,
      "step": 3519
    },
    {
      "epoch": 0.48205970966858397,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008957044799874202,
      "loss": 5.2663,
      "step": 3520
    },
    {
      "epoch": 0.4821966584497398,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008956988314348128,
      "loss": 5.1601,
      "step": 3521
    },
    {
      "epoch": 0.48233360723089563,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008956931791885929,
      "loss": 5.1757,
      "step": 3522
    },
    {
      "epoch": 0.48247055601205147,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008956875232488075,
      "loss": 5.1923,
      "step": 3523
    },
    {
      "epoch": 0.48260750479320735,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008956818636155034,
      "loss": 5.1913,
      "step": 3524
    },
    {
      "epoch": 0.4827444535743632,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008956762002887274,
      "loss": 5.1261,
      "step": 3525
    },
    {
      "epoch": 0.482881402355519,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008956705332685266,
      "loss": 5.2043,
      "step": 3526
    },
    {
      "epoch": 0.4830183511366749,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008956648625549478,
      "loss": 5.1199,
      "step": 3527
    },
    {
      "epoch": 0.48315529991783074,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008956591881480382,
      "loss": 5.0936,
      "step": 3528
    },
    {
      "epoch": 0.48329224869898657,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008956535100478446,
      "loss": 5.1962,
      "step": 3529
    },
    {
      "epoch": 0.4834291974801424,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008956478282544143,
      "loss": 5.104,
      "step": 3530
    },
    {
      "epoch": 0.4835661462612983,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008956421427677942,
      "loss": 5.151,
      "step": 3531
    },
    {
      "epoch": 0.4837030950424541,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008956364535880315,
      "loss": 5.0968,
      "step": 3532
    },
    {
      "epoch": 0.48384004382360996,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008956307607151732,
      "loss": 5.1254,
      "step": 3533
    },
    {
      "epoch": 0.48397699260476584,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008956250641492667,
      "loss": 5.2217,
      "step": 3534
    },
    {
      "epoch": 0.4841139413859217,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008956193638903591,
      "loss": 5.2058,
      "step": 3535
    },
    {
      "epoch": 0.4842508901670775,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008956136599384977,
      "loss": 5.1816,
      "step": 3536
    },
    {
      "epoch": 0.48438783894823334,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008956079522937296,
      "loss": 5.2271,
      "step": 3537
    },
    {
      "epoch": 0.48452478772938923,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008956022409561022,
      "loss": 5.2141,
      "step": 3538
    },
    {
      "epoch": 0.48466173651054506,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000895596525925663,
      "loss": 5.2516,
      "step": 3539
    },
    {
      "epoch": 0.4847986852917009,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008955908072024591,
      "loss": 5.1047,
      "step": 3540
    },
    {
      "epoch": 0.4849356340728567,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000895585084786538,
      "loss": 5.248,
      "step": 3541
    },
    {
      "epoch": 0.4850725828540126,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008955793586779471,
      "loss": 5.1755,
      "step": 3542
    },
    {
      "epoch": 0.48520953163516845,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000895573628876734,
      "loss": 5.1871,
      "step": 3543
    },
    {
      "epoch": 0.4853464804163243,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008955678953829461,
      "loss": 5.2281,
      "step": 3544
    },
    {
      "epoch": 0.48548342919748017,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008955621581966306,
      "loss": 5.1659,
      "step": 3545
    },
    {
      "epoch": 0.485620377978636,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008955564173178356,
      "loss": 5.1907,
      "step": 3546
    },
    {
      "epoch": 0.48575732675979183,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008955506727466084,
      "loss": 5.1537,
      "step": 3547
    },
    {
      "epoch": 0.48589427554094766,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008955449244829964,
      "loss": 5.1175,
      "step": 3548
    },
    {
      "epoch": 0.48603122432210355,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008955391725270475,
      "loss": 5.1125,
      "step": 3549
    },
    {
      "epoch": 0.4861681731032594,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008955334168788094,
      "loss": 5.2292,
      "step": 3550
    },
    {
      "epoch": 0.4863051218844152,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008955276575383297,
      "loss": 5.1171,
      "step": 3551
    },
    {
      "epoch": 0.4864420706655711,
      "grad_norm": 0.5,
      "learning_rate": 0.0008955218945056561,
      "loss": 5.1407,
      "step": 3552
    },
    {
      "epoch": 0.48657901944672693,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008955161277808364,
      "loss": 5.1503,
      "step": 3553
    },
    {
      "epoch": 0.48671596822788277,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008955103573639184,
      "loss": 5.2191,
      "step": 3554
    },
    {
      "epoch": 0.4868529170090386,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008955045832549499,
      "loss": 5.2039,
      "step": 3555
    },
    {
      "epoch": 0.4869898657901945,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008954988054539788,
      "loss": 5.3002,
      "step": 3556
    },
    {
      "epoch": 0.4871268145713503,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008954930239610529,
      "loss": 5.1272,
      "step": 3557
    },
    {
      "epoch": 0.48726376335250615,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008954872387762202,
      "loss": 5.1614,
      "step": 3558
    },
    {
      "epoch": 0.487400712133662,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008954814498995285,
      "loss": 5.1064,
      "step": 3559
    },
    {
      "epoch": 0.48753766091481787,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008954756573310259,
      "loss": 5.2367,
      "step": 3560
    },
    {
      "epoch": 0.4876746096959737,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008954698610707603,
      "loss": 5.2018,
      "step": 3561
    },
    {
      "epoch": 0.48781155847712954,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008954640611187801,
      "loss": 5.2009,
      "step": 3562
    },
    {
      "epoch": 0.4879485072582854,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008954582574751328,
      "loss": 5.1804,
      "step": 3563
    },
    {
      "epoch": 0.48808545603944126,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008954524501398669,
      "loss": 5.1696,
      "step": 3564
    },
    {
      "epoch": 0.4882224048205971,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008954466391130304,
      "loss": 5.2612,
      "step": 3565
    },
    {
      "epoch": 0.4883593536017529,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008954408243946714,
      "loss": 5.2054,
      "step": 3566
    },
    {
      "epoch": 0.4884963023829088,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008954350059848381,
      "loss": 5.1301,
      "step": 3567
    },
    {
      "epoch": 0.48863325116406464,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008954291838835789,
      "loss": 5.2027,
      "step": 3568
    },
    {
      "epoch": 0.4887701999452205,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008954233580909419,
      "loss": 5.1988,
      "step": 3569
    },
    {
      "epoch": 0.48890714872637636,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008954175286069753,
      "loss": 5.1853,
      "step": 3570
    },
    {
      "epoch": 0.4890440975075322,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008954116954317275,
      "loss": 5.1305,
      "step": 3571
    },
    {
      "epoch": 0.489181046288688,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008954058585652469,
      "loss": 5.1981,
      "step": 3572
    },
    {
      "epoch": 0.48931799506984386,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008954000180075818,
      "loss": 5.1854,
      "step": 3573
    },
    {
      "epoch": 0.48945494385099975,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008953941737587808,
      "loss": 5.1847,
      "step": 3574
    },
    {
      "epoch": 0.4895918926321556,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008953883258188918,
      "loss": 5.0894,
      "step": 3575
    },
    {
      "epoch": 0.4897288414133114,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008953824741879638,
      "loss": 5.1955,
      "step": 3576
    },
    {
      "epoch": 0.48986579019446724,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000895376618866045,
      "loss": 5.163,
      "step": 3577
    },
    {
      "epoch": 0.49000273897562313,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008953707598531842,
      "loss": 5.1961,
      "step": 3578
    },
    {
      "epoch": 0.49013968775677896,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008953648971494298,
      "loss": 5.201,
      "step": 3579
    },
    {
      "epoch": 0.4902766365379348,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008953590307548302,
      "loss": 5.1792,
      "step": 3580
    },
    {
      "epoch": 0.4904135853190907,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008953531606694342,
      "loss": 5.0731,
      "step": 3581
    },
    {
      "epoch": 0.4905505341002465,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008953472868932906,
      "loss": 5.1074,
      "step": 3582
    },
    {
      "epoch": 0.49068748288140235,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008953414094264478,
      "loss": 5.2293,
      "step": 3583
    },
    {
      "epoch": 0.4908244316625582,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008953355282689548,
      "loss": 5.1426,
      "step": 3584
    },
    {
      "epoch": 0.49096138044371407,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008953296434208599,
      "loss": 5.166,
      "step": 3585
    },
    {
      "epoch": 0.4910983292248699,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008953237548822122,
      "loss": 5.1833,
      "step": 3586
    },
    {
      "epoch": 0.49123527800602573,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008953178626530605,
      "loss": 5.1671,
      "step": 3587
    },
    {
      "epoch": 0.4913722267871816,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008953119667334535,
      "loss": 5.1634,
      "step": 3588
    },
    {
      "epoch": 0.49150917556833745,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008953060671234402,
      "loss": 5.1958,
      "step": 3589
    },
    {
      "epoch": 0.4916461243494933,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008953001638230693,
      "loss": 5.1676,
      "step": 3590
    },
    {
      "epoch": 0.4917830731306491,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0008952942568323899,
      "loss": 5.2256,
      "step": 3591
    },
    {
      "epoch": 0.491920021911805,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008952883461514509,
      "loss": 5.1966,
      "step": 3592
    },
    {
      "epoch": 0.49205697069296084,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008952824317803011,
      "loss": 5.094,
      "step": 3593
    },
    {
      "epoch": 0.49219391947411667,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00089527651371899,
      "loss": 5.1887,
      "step": 3594
    },
    {
      "epoch": 0.4923308682552725,
      "grad_norm": 0.75,
      "learning_rate": 0.0008952705919675661,
      "loss": 5.1977,
      "step": 3595
    },
    {
      "epoch": 0.4924678170364284,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008952646665260788,
      "loss": 5.1322,
      "step": 3596
    },
    {
      "epoch": 0.4926047658175842,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000895258737394577,
      "loss": 5.2354,
      "step": 3597
    },
    {
      "epoch": 0.49274171459874005,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00089525280457311,
      "loss": 5.1227,
      "step": 3598
    },
    {
      "epoch": 0.49287866337989594,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008952468680617269,
      "loss": 5.1799,
      "step": 3599
    },
    {
      "epoch": 0.4930156121610518,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000895240927860477,
      "loss": 5.2329,
      "step": 3600
    },
    {
      "epoch": 0.4931525609422076,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008952349839694093,
      "loss": 5.173,
      "step": 3601
    },
    {
      "epoch": 0.49328950972336344,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008952290363885732,
      "loss": 5.1273,
      "step": 3602
    },
    {
      "epoch": 0.4934264585045193,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008952230851180181,
      "loss": 5.1266,
      "step": 3603
    },
    {
      "epoch": 0.49356340728567516,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008952171301577932,
      "loss": 5.1638,
      "step": 3604
    },
    {
      "epoch": 0.493700356066831,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008952111715079477,
      "loss": 5.1388,
      "step": 3605
    },
    {
      "epoch": 0.4938373048479869,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008952052091685313,
      "loss": 5.2097,
      "step": 3606
    },
    {
      "epoch": 0.4939742536291427,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008951992431395932,
      "loss": 5.1553,
      "step": 3607
    },
    {
      "epoch": 0.49411120241029854,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008951932734211828,
      "loss": 5.2144,
      "step": 3608
    },
    {
      "epoch": 0.4942481511914544,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008951873000133496,
      "loss": 5.1466,
      "step": 3609
    },
    {
      "epoch": 0.49438509997261026,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008951813229161434,
      "loss": 5.1753,
      "step": 3610
    },
    {
      "epoch": 0.4945220487537661,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008951753421296134,
      "loss": 5.0765,
      "step": 3611
    },
    {
      "epoch": 0.49465899753492193,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008951693576538092,
      "loss": 5.17,
      "step": 3612
    },
    {
      "epoch": 0.49479594631607776,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008951633694887804,
      "loss": 5.1109,
      "step": 3613
    },
    {
      "epoch": 0.49493289509723365,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008951573776345769,
      "loss": 5.1318,
      "step": 3614
    },
    {
      "epoch": 0.4950698438783895,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008951513820912479,
      "loss": 5.0949,
      "step": 3615
    },
    {
      "epoch": 0.4952067926595453,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008951453828588434,
      "loss": 5.1347,
      "step": 3616
    },
    {
      "epoch": 0.4953437414407012,
      "grad_norm": 0.375,
      "learning_rate": 0.0008951393799374131,
      "loss": 5.1625,
      "step": 3617
    },
    {
      "epoch": 0.49548069022185703,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008951333733270067,
      "loss": 5.2205,
      "step": 3618
    },
    {
      "epoch": 0.49561763900301287,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008951273630276739,
      "loss": 5.2097,
      "step": 3619
    },
    {
      "epoch": 0.4957545877841687,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008951213490394646,
      "loss": 5.1512,
      "step": 3620
    },
    {
      "epoch": 0.4958915365653246,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008951153313624286,
      "loss": 5.1638,
      "step": 3621
    },
    {
      "epoch": 0.4960284853464804,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008951093099966159,
      "loss": 5.1964,
      "step": 3622
    },
    {
      "epoch": 0.49616543412763625,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008951032849420761,
      "loss": 5.1834,
      "step": 3623
    },
    {
      "epoch": 0.49630238290879214,
      "grad_norm": 0.375,
      "learning_rate": 0.0008950972561988594,
      "loss": 5.1277,
      "step": 3624
    },
    {
      "epoch": 0.49643933168994797,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008950912237670158,
      "loss": 5.2375,
      "step": 3625
    },
    {
      "epoch": 0.4965762804711038,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008950851876465949,
      "loss": 5.1859,
      "step": 3626
    },
    {
      "epoch": 0.49671322925225964,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008950791478376471,
      "loss": 5.2063,
      "step": 3627
    },
    {
      "epoch": 0.4968501780334155,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008950731043402225,
      "loss": 5.2164,
      "step": 3628
    },
    {
      "epoch": 0.49698712681457136,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008950670571543709,
      "loss": 5.1818,
      "step": 3629
    },
    {
      "epoch": 0.4971240755957272,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008950610062801424,
      "loss": 5.1044,
      "step": 3630
    },
    {
      "epoch": 0.497261024376883,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008950549517175875,
      "loss": 5.1382,
      "step": 3631
    },
    {
      "epoch": 0.4973979731580389,
      "grad_norm": 0.4609375,
      "learning_rate": 0.000895048893466756,
      "loss": 5.1428,
      "step": 3632
    },
    {
      "epoch": 0.49753492193919474,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008950428315276984,
      "loss": 5.2088,
      "step": 3633
    },
    {
      "epoch": 0.4976718707203506,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008950367659004649,
      "loss": 5.1955,
      "step": 3634
    },
    {
      "epoch": 0.49780881950150646,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008950306965851056,
      "loss": 5.1967,
      "step": 3635
    },
    {
      "epoch": 0.4979457682826623,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008950246235816708,
      "loss": 5.1527,
      "step": 3636
    },
    {
      "epoch": 0.4980827170638181,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008950185468902109,
      "loss": 5.1393,
      "step": 3637
    },
    {
      "epoch": 0.49821966584497396,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008950124665107762,
      "loss": 5.1246,
      "step": 3638
    },
    {
      "epoch": 0.49835661462612985,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008950063824434174,
      "loss": 5.2119,
      "step": 3639
    },
    {
      "epoch": 0.4984935634072857,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008950002946881844,
      "loss": 5.1715,
      "step": 3640
    },
    {
      "epoch": 0.4986305121884415,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008949942032451281,
      "loss": 5.1791,
      "step": 3641
    },
    {
      "epoch": 0.4987674609695974,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008949881081142986,
      "loss": 5.1832,
      "step": 3642
    },
    {
      "epoch": 0.49890440975075323,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008949820092957468,
      "loss": 5.1243,
      "step": 3643
    },
    {
      "epoch": 0.49904135853190906,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000894975906789523,
      "loss": 5.1941,
      "step": 3644
    },
    {
      "epoch": 0.4991783073130649,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008949698005956779,
      "loss": 5.2092,
      "step": 3645
    },
    {
      "epoch": 0.4993152560942208,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000894963690714262,
      "loss": 5.1338,
      "step": 3646
    },
    {
      "epoch": 0.4994522048753766,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008949575771453258,
      "loss": 5.1816,
      "step": 3647
    },
    {
      "epoch": 0.49958915365653245,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008949514598889202,
      "loss": 5.1096,
      "step": 3648
    },
    {
      "epoch": 0.4997261024376883,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008949453389450959,
      "loss": 5.2117,
      "step": 3649
    },
    {
      "epoch": 0.49986305121884417,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008949392143139036,
      "loss": 5.1467,
      "step": 3650
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000894933085995394,
      "loss": 5.1971,
      "step": 3651
    },
    {
      "epoch": 0.5001369487811559,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008949269539896178,
      "loss": 5.2127,
      "step": 3652
    },
    {
      "epoch": 0.5002738975623117,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008949208182966259,
      "loss": 5.2298,
      "step": 3653
    },
    {
      "epoch": 0.5004108463434676,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008949146789164692,
      "loss": 5.1102,
      "step": 3654
    },
    {
      "epoch": 0.5005477951246234,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008949085358491986,
      "loss": 5.2023,
      "step": 3655
    },
    {
      "epoch": 0.5006847439057792,
      "grad_norm": 0.5,
      "learning_rate": 0.0008949023890948649,
      "loss": 5.16,
      "step": 3656
    },
    {
      "epoch": 0.5008216926869351,
      "grad_norm": 0.625,
      "learning_rate": 0.0008948962386535191,
      "loss": 5.1743,
      "step": 3657
    },
    {
      "epoch": 0.5009586414680909,
      "grad_norm": 0.5,
      "learning_rate": 0.0008948900845252121,
      "loss": 5.2449,
      "step": 3658
    },
    {
      "epoch": 0.5010955902492468,
      "grad_norm": 0.4375,
      "learning_rate": 0.000894883926709995,
      "loss": 5.2078,
      "step": 3659
    },
    {
      "epoch": 0.5012325390304027,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008948777652079187,
      "loss": 5.2235,
      "step": 3660
    },
    {
      "epoch": 0.5013694878115584,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008948716000190344,
      "loss": 5.1471,
      "step": 3661
    },
    {
      "epoch": 0.5015064365927143,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008948654311433933,
      "loss": 5.2108,
      "step": 3662
    },
    {
      "epoch": 0.5016433853738702,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008948592585810462,
      "loss": 5.2411,
      "step": 3663
    },
    {
      "epoch": 0.501780334155026,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008948530823320444,
      "loss": 5.2511,
      "step": 3664
    },
    {
      "epoch": 0.5019172829361819,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008948469023964392,
      "loss": 5.1455,
      "step": 3665
    },
    {
      "epoch": 0.5020542317173378,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008948407187742817,
      "loss": 5.1789,
      "step": 3666
    },
    {
      "epoch": 0.5021911804984935,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008948345314656232,
      "loss": 5.206,
      "step": 3667
    },
    {
      "epoch": 0.5023281292796494,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008948283404705148,
      "loss": 5.1604,
      "step": 3668
    },
    {
      "epoch": 0.5024650780608052,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008948221457890081,
      "loss": 5.1667,
      "step": 3669
    },
    {
      "epoch": 0.5026020268419611,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008948159474211542,
      "loss": 5.1713,
      "step": 3670
    },
    {
      "epoch": 0.502738975623117,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008948097453670045,
      "loss": 5.2018,
      "step": 3671
    },
    {
      "epoch": 0.5028759244042728,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008948035396266105,
      "loss": 5.1681,
      "step": 3672
    },
    {
      "epoch": 0.5030128731854286,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008947973302000237,
      "loss": 5.1594,
      "step": 3673
    },
    {
      "epoch": 0.5031498219665845,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008947911170872953,
      "loss": 5.1347,
      "step": 3674
    },
    {
      "epoch": 0.5032867707477403,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008947849002884769,
      "loss": 5.2155,
      "step": 3675
    },
    {
      "epoch": 0.5034237195288962,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008947786798036201,
      "loss": 5.1932,
      "step": 3676
    },
    {
      "epoch": 0.5035606683100521,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008947724556327764,
      "loss": 5.1716,
      "step": 3677
    },
    {
      "epoch": 0.5036976170912079,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008947662277759975,
      "loss": 5.1457,
      "step": 3678
    },
    {
      "epoch": 0.5038345658723637,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008947599962333346,
      "loss": 5.2176,
      "step": 3679
    },
    {
      "epoch": 0.5039715146535196,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00089475376100484,
      "loss": 5.1543,
      "step": 3680
    },
    {
      "epoch": 0.5041084634346754,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008947475220905646,
      "loss": 5.1831,
      "step": 3681
    },
    {
      "epoch": 0.5042454122158313,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008947412794905608,
      "loss": 5.2188,
      "step": 3682
    },
    {
      "epoch": 0.5043823609969871,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008947350332048799,
      "loss": 5.2581,
      "step": 3683
    },
    {
      "epoch": 0.504519309778143,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008947287832335738,
      "loss": 5.1168,
      "step": 3684
    },
    {
      "epoch": 0.5046562585592989,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008947225295766944,
      "loss": 5.2021,
      "step": 3685
    },
    {
      "epoch": 0.5047932073404546,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008947162722342933,
      "loss": 5.1607,
      "step": 3686
    },
    {
      "epoch": 0.5049301561216105,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008947100112064226,
      "loss": 5.1433,
      "step": 3687
    },
    {
      "epoch": 0.5050671049027664,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000894703746493134,
      "loss": 5.1422,
      "step": 3688
    },
    {
      "epoch": 0.5052040536839222,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008946974780944795,
      "loss": 5.1988,
      "step": 3689
    },
    {
      "epoch": 0.5053410024650781,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000894691206010511,
      "loss": 5.1755,
      "step": 3690
    },
    {
      "epoch": 0.505477951246234,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008946849302412807,
      "loss": 5.1064,
      "step": 3691
    },
    {
      "epoch": 0.5056149000273897,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008946786507868402,
      "loss": 5.1613,
      "step": 3692
    },
    {
      "epoch": 0.5057518488085456,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008946723676472418,
      "loss": 5.1685,
      "step": 3693
    },
    {
      "epoch": 0.5058887975897014,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008946660808225376,
      "loss": 5.1552,
      "step": 3694
    },
    {
      "epoch": 0.5060257463708573,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008946597903127795,
      "loss": 5.1667,
      "step": 3695
    },
    {
      "epoch": 0.5061626951520132,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00089465349611802,
      "loss": 5.1158,
      "step": 3696
    },
    {
      "epoch": 0.506299643933169,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008946471982383108,
      "loss": 5.221,
      "step": 3697
    },
    {
      "epoch": 0.5064365927143248,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008946408966737043,
      "loss": 5.1398,
      "step": 3698
    },
    {
      "epoch": 0.5065735414954807,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008946345914242529,
      "loss": 5.2098,
      "step": 3699
    },
    {
      "epoch": 0.5067104902766365,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008946282824900087,
      "loss": 5.1871,
      "step": 3700
    },
    {
      "epoch": 0.5068474390577924,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008946219698710238,
      "loss": 5.106,
      "step": 3701
    },
    {
      "epoch": 0.5069843878389483,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008946156535673508,
      "loss": 5.1514,
      "step": 3702
    },
    {
      "epoch": 0.5071213366201041,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008946093335790419,
      "loss": 5.2908,
      "step": 3703
    },
    {
      "epoch": 0.50725828540126,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008946030099061495,
      "loss": 5.1309,
      "step": 3704
    },
    {
      "epoch": 0.5073952341824157,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008945966825487261,
      "loss": 5.1673,
      "step": 3705
    },
    {
      "epoch": 0.5075321829635716,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000894590351506824,
      "loss": 5.125,
      "step": 3706
    },
    {
      "epoch": 0.5076691317447275,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008945840167804956,
      "loss": 5.1828,
      "step": 3707
    },
    {
      "epoch": 0.5078060805258833,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008945776783697936,
      "loss": 5.2045,
      "step": 3708
    },
    {
      "epoch": 0.5079430293070392,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008945713362747705,
      "loss": 5.1723,
      "step": 3709
    },
    {
      "epoch": 0.508079978088195,
      "grad_norm": 0.5,
      "learning_rate": 0.0008945649904954788,
      "loss": 5.1467,
      "step": 3710
    },
    {
      "epoch": 0.5082169268693508,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000894558641031971,
      "loss": 5.1674,
      "step": 3711
    },
    {
      "epoch": 0.5083538756505067,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008945522878842999,
      "loss": 5.1356,
      "step": 3712
    },
    {
      "epoch": 0.5084908244316626,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000894545931052518,
      "loss": 5.1844,
      "step": 3713
    },
    {
      "epoch": 0.5086277732128184,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008945395705366781,
      "loss": 5.1816,
      "step": 3714
    },
    {
      "epoch": 0.5087647219939743,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008945332063368328,
      "loss": 5.1134,
      "step": 3715
    },
    {
      "epoch": 0.5089016707751302,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008945268384530348,
      "loss": 5.1892,
      "step": 3716
    },
    {
      "epoch": 0.5090386195562859,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000894520466885337,
      "loss": 5.1232,
      "step": 3717
    },
    {
      "epoch": 0.5091755683374418,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008945140916337924,
      "loss": 5.2452,
      "step": 3718
    },
    {
      "epoch": 0.5093125171185976,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008945077126984534,
      "loss": 5.1731,
      "step": 3719
    },
    {
      "epoch": 0.5094494658997535,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000894501330079373,
      "loss": 5.1613,
      "step": 3720
    },
    {
      "epoch": 0.5095864146809094,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008944949437766043,
      "loss": 5.1093,
      "step": 3721
    },
    {
      "epoch": 0.5097233634620651,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008944885537902001,
      "loss": 5.21,
      "step": 3722
    },
    {
      "epoch": 0.509860312243221,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008944821601202133,
      "loss": 5.1431,
      "step": 3723
    },
    {
      "epoch": 0.5099972610243769,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008944757627666969,
      "loss": 5.185,
      "step": 3724
    },
    {
      "epoch": 0.5101342098055327,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008944693617297039,
      "loss": 5.1312,
      "step": 3725
    },
    {
      "epoch": 0.5102711585866886,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008944629570092875,
      "loss": 5.1541,
      "step": 3726
    },
    {
      "epoch": 0.5104081073678445,
      "grad_norm": 0.5,
      "learning_rate": 0.0008944565486055005,
      "loss": 5.2162,
      "step": 3727
    },
    {
      "epoch": 0.5105450561490003,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008944501365183963,
      "loss": 5.1895,
      "step": 3728
    },
    {
      "epoch": 0.5106820049301561,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008944437207480277,
      "loss": 5.133,
      "step": 3729
    },
    {
      "epoch": 0.5108189537113119,
      "grad_norm": 0.5,
      "learning_rate": 0.0008944373012944485,
      "loss": 5.1412,
      "step": 3730
    },
    {
      "epoch": 0.5109559024924678,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008944308781577112,
      "loss": 5.1554,
      "step": 3731
    },
    {
      "epoch": 0.5110928512736237,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008944244513378693,
      "loss": 5.1333,
      "step": 3732
    },
    {
      "epoch": 0.5112298000547795,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008944180208349761,
      "loss": 5.1961,
      "step": 3733
    },
    {
      "epoch": 0.5113667488359354,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008944115866490849,
      "loss": 5.1285,
      "step": 3734
    },
    {
      "epoch": 0.5115036976170912,
      "grad_norm": 0.49609375,
      "learning_rate": 0.000894405148780249,
      "loss": 5.1159,
      "step": 3735
    },
    {
      "epoch": 0.511640646398247,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008943987072285218,
      "loss": 5.1897,
      "step": 3736
    },
    {
      "epoch": 0.5117775951794029,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008943922619939564,
      "loss": 5.2001,
      "step": 3737
    },
    {
      "epoch": 0.5119145439605588,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008943858130766066,
      "loss": 5.236,
      "step": 3738
    },
    {
      "epoch": 0.5120514927417146,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008943793604765256,
      "loss": 5.1518,
      "step": 3739
    },
    {
      "epoch": 0.5121884415228705,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000894372904193767,
      "loss": 5.125,
      "step": 3740
    },
    {
      "epoch": 0.5123253903040262,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008943664442283843,
      "loss": 5.207,
      "step": 3741
    },
    {
      "epoch": 0.5124623390851821,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008943599805804308,
      "loss": 5.1198,
      "step": 3742
    },
    {
      "epoch": 0.512599287866338,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008943535132499605,
      "loss": 5.119,
      "step": 3743
    },
    {
      "epoch": 0.5127362366474938,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008943470422370266,
      "loss": 5.209,
      "step": 3744
    },
    {
      "epoch": 0.5128731854286497,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008943405675416829,
      "loss": 5.2029,
      "step": 3745
    },
    {
      "epoch": 0.5130101342098056,
      "grad_norm": 0.46875,
      "learning_rate": 0.000894334089163983,
      "loss": 5.1574,
      "step": 3746
    },
    {
      "epoch": 0.5131470829909613,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008943276071039808,
      "loss": 5.1874,
      "step": 3747
    },
    {
      "epoch": 0.5132840317721172,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008943211213617296,
      "loss": 5.2245,
      "step": 3748
    },
    {
      "epoch": 0.5134209805532731,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008943146319372836,
      "loss": 5.2156,
      "step": 3749
    },
    {
      "epoch": 0.5135579293344289,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008943081388306962,
      "loss": 5.1708,
      "step": 3750
    },
    {
      "epoch": 0.5136948781155848,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008943016420420216,
      "loss": 5.1479,
      "step": 3751
    },
    {
      "epoch": 0.5138318268967407,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008942951415713133,
      "loss": 5.1961,
      "step": 3752
    },
    {
      "epoch": 0.5139687756778965,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008942886374186252,
      "loss": 5.2139,
      "step": 3753
    },
    {
      "epoch": 0.5141057244590523,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008942821295840114,
      "loss": 5.1715,
      "step": 3754
    },
    {
      "epoch": 0.5142426732402081,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008942756180675256,
      "loss": 5.1939,
      "step": 3755
    },
    {
      "epoch": 0.514379622021364,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008942691028692221,
      "loss": 5.1583,
      "step": 3756
    },
    {
      "epoch": 0.5145165708025199,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008942625839891545,
      "loss": 5.0969,
      "step": 3757
    },
    {
      "epoch": 0.5146535195836757,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008942560614273771,
      "loss": 5.191,
      "step": 3758
    },
    {
      "epoch": 0.5147904683648316,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008942495351839439,
      "loss": 5.1221,
      "step": 3759
    },
    {
      "epoch": 0.5149274171459874,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008942430052589087,
      "loss": 5.1416,
      "step": 3760
    },
    {
      "epoch": 0.5150643659271432,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008942364716523261,
      "loss": 5.1921,
      "step": 3761
    },
    {
      "epoch": 0.5152013147082991,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008942299343642499,
      "loss": 5.1301,
      "step": 3762
    },
    {
      "epoch": 0.515338263489455,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008942233933947344,
      "loss": 5.1939,
      "step": 3763
    },
    {
      "epoch": 0.5154752122706108,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008942168487438339,
      "loss": 5.2145,
      "step": 3764
    },
    {
      "epoch": 0.5156121610517667,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008942103004116023,
      "loss": 5.1356,
      "step": 3765
    },
    {
      "epoch": 0.5157491098329224,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008942037483980942,
      "loss": 5.1911,
      "step": 3766
    },
    {
      "epoch": 0.5158860586140783,
      "grad_norm": 0.5,
      "learning_rate": 0.0008941971927033638,
      "loss": 5.1414,
      "step": 3767
    },
    {
      "epoch": 0.5160230073952342,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008941906333274655,
      "loss": 5.0828,
      "step": 3768
    },
    {
      "epoch": 0.51615995617639,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008941840702704535,
      "loss": 5.1709,
      "step": 3769
    },
    {
      "epoch": 0.5162969049575459,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008941775035323821,
      "loss": 5.1176,
      "step": 3770
    },
    {
      "epoch": 0.5164338537387018,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008941709331133061,
      "loss": 5.0815,
      "step": 3771
    },
    {
      "epoch": 0.5165708025198575,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008941643590132797,
      "loss": 5.199,
      "step": 3772
    },
    {
      "epoch": 0.5167077513010134,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008941577812323575,
      "loss": 5.1506,
      "step": 3773
    },
    {
      "epoch": 0.5168447000821693,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008941511997705937,
      "loss": 5.1544,
      "step": 3774
    },
    {
      "epoch": 0.5169816488633251,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008941446146280433,
      "loss": 5.1206,
      "step": 3775
    },
    {
      "epoch": 0.517118597644481,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008941380258047605,
      "loss": 5.1483,
      "step": 3776
    },
    {
      "epoch": 0.5172555464256368,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008941314333008001,
      "loss": 5.138,
      "step": 3777
    },
    {
      "epoch": 0.5173924952067926,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008941248371162167,
      "loss": 5.1412,
      "step": 3778
    },
    {
      "epoch": 0.5175294439879485,
      "grad_norm": 0.625,
      "learning_rate": 0.0008941182372510649,
      "loss": 5.1828,
      "step": 3779
    },
    {
      "epoch": 0.5176663927691043,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008941116337053995,
      "loss": 5.2232,
      "step": 3780
    },
    {
      "epoch": 0.5178033415502602,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008941050264792751,
      "loss": 5.1862,
      "step": 3781
    },
    {
      "epoch": 0.5179402903314161,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008940984155727465,
      "loss": 5.1518,
      "step": 3782
    },
    {
      "epoch": 0.5180772391125719,
      "grad_norm": 0.375,
      "learning_rate": 0.0008940918009858686,
      "loss": 5.1284,
      "step": 3783
    },
    {
      "epoch": 0.5182141878937278,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008940851827186962,
      "loss": 5.2108,
      "step": 3784
    },
    {
      "epoch": 0.5183511366748836,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008940785607712839,
      "loss": 5.1572,
      "step": 3785
    },
    {
      "epoch": 0.5184880854560394,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008940719351436867,
      "loss": 5.1759,
      "step": 3786
    },
    {
      "epoch": 0.5186250342371953,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008940653058359597,
      "loss": 5.1179,
      "step": 3787
    },
    {
      "epoch": 0.5187619830183512,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008940586728481576,
      "loss": 5.1223,
      "step": 3788
    },
    {
      "epoch": 0.518898931799507,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008940520361803355,
      "loss": 5.2151,
      "step": 3789
    },
    {
      "epoch": 0.5190358805806629,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008940453958325485,
      "loss": 5.0978,
      "step": 3790
    },
    {
      "epoch": 0.5191728293618186,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008940387518048512,
      "loss": 5.132,
      "step": 3791
    },
    {
      "epoch": 0.5193097781429745,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008940321040972993,
      "loss": 5.1394,
      "step": 3792
    },
    {
      "epoch": 0.5194467269241304,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008940254527099473,
      "loss": 5.1852,
      "step": 3793
    },
    {
      "epoch": 0.5195836757052862,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008940187976428508,
      "loss": 5.1808,
      "step": 3794
    },
    {
      "epoch": 0.5197206244864421,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008940121388960645,
      "loss": 5.1785,
      "step": 3795
    },
    {
      "epoch": 0.519857573267598,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008940054764696438,
      "loss": 5.1399,
      "step": 3796
    },
    {
      "epoch": 0.5199945220487537,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000893998810363644,
      "loss": 5.1048,
      "step": 3797
    },
    {
      "epoch": 0.5201314708299096,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008939921405781202,
      "loss": 5.1039,
      "step": 3798
    },
    {
      "epoch": 0.5202684196110655,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008939854671131278,
      "loss": 5.1881,
      "step": 3799
    },
    {
      "epoch": 0.5204053683922213,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008939787899687221,
      "loss": 5.1751,
      "step": 3800
    },
    {
      "epoch": 0.5205423171733772,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008939721091449582,
      "loss": 5.1872,
      "step": 3801
    },
    {
      "epoch": 0.520679265954533,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008939654246418918,
      "loss": 5.1348,
      "step": 3802
    },
    {
      "epoch": 0.5208162147356888,
      "grad_norm": 0.4375,
      "learning_rate": 0.000893958736459578,
      "loss": 5.1398,
      "step": 3803
    },
    {
      "epoch": 0.5209531635168447,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008939520445980725,
      "loss": 5.1249,
      "step": 3804
    },
    {
      "epoch": 0.5210901122980005,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008939453490574305,
      "loss": 5.1402,
      "step": 3805
    },
    {
      "epoch": 0.5212270610791564,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008939386498377077,
      "loss": 5.1907,
      "step": 3806
    },
    {
      "epoch": 0.5213640098603123,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008939319469389595,
      "loss": 5.148,
      "step": 3807
    },
    {
      "epoch": 0.5215009586414681,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008939252403612414,
      "loss": 5.1382,
      "step": 3808
    },
    {
      "epoch": 0.521637907422624,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008939185301046091,
      "loss": 5.255,
      "step": 3809
    },
    {
      "epoch": 0.5217748562037798,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008939118161691181,
      "loss": 5.2355,
      "step": 3810
    },
    {
      "epoch": 0.5219118049849356,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008939050985548241,
      "loss": 5.1257,
      "step": 3811
    },
    {
      "epoch": 0.5220487537660915,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008938983772617829,
      "loss": 5.2317,
      "step": 3812
    },
    {
      "epoch": 0.5221857025472473,
      "grad_norm": 0.375,
      "learning_rate": 0.00089389165229005,
      "loss": 5.168,
      "step": 3813
    },
    {
      "epoch": 0.5223226513284032,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008938849236396812,
      "loss": 5.1501,
      "step": 3814
    },
    {
      "epoch": 0.522459600109559,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008938781913107323,
      "loss": 5.1279,
      "step": 3815
    },
    {
      "epoch": 0.5225965488907148,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000893871455303259,
      "loss": 5.2112,
      "step": 3816
    },
    {
      "epoch": 0.5227334976718707,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008938647156173171,
      "loss": 5.2558,
      "step": 3817
    },
    {
      "epoch": 0.5228704464530266,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008938579722529628,
      "loss": 5.2543,
      "step": 3818
    },
    {
      "epoch": 0.5230073952341824,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008938512252102515,
      "loss": 5.1693,
      "step": 3819
    },
    {
      "epoch": 0.5231443440153383,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008938444744892394,
      "loss": 5.1661,
      "step": 3820
    },
    {
      "epoch": 0.5232812927964942,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008938377200899824,
      "loss": 5.1628,
      "step": 3821
    },
    {
      "epoch": 0.5234182415776499,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008938309620125363,
      "loss": 5.254,
      "step": 3822
    },
    {
      "epoch": 0.5235551903588058,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008938242002569575,
      "loss": 5.1509,
      "step": 3823
    },
    {
      "epoch": 0.5236921391399617,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008938174348233016,
      "loss": 5.1495,
      "step": 3824
    },
    {
      "epoch": 0.5238290879211175,
      "grad_norm": 0.46875,
      "learning_rate": 0.000893810665711625,
      "loss": 5.1499,
      "step": 3825
    },
    {
      "epoch": 0.5239660367022734,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008938038929219836,
      "loss": 5.1876,
      "step": 3826
    },
    {
      "epoch": 0.5241029854834292,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008937971164544337,
      "loss": 5.1857,
      "step": 3827
    },
    {
      "epoch": 0.524239934264585,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008937903363090311,
      "loss": 5.1682,
      "step": 3828
    },
    {
      "epoch": 0.5243768830457409,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008937835524858324,
      "loss": 5.2298,
      "step": 3829
    },
    {
      "epoch": 0.5245138318268967,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008937767649848936,
      "loss": 5.2085,
      "step": 3830
    },
    {
      "epoch": 0.5246507806080526,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008937699738062711,
      "loss": 5.104,
      "step": 3831
    },
    {
      "epoch": 0.5247877293892085,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008937631789500209,
      "loss": 5.1477,
      "step": 3832
    },
    {
      "epoch": 0.5249246781703643,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008937563804161996,
      "loss": 5.1171,
      "step": 3833
    },
    {
      "epoch": 0.5250616269515201,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008937495782048634,
      "loss": 5.0861,
      "step": 3834
    },
    {
      "epoch": 0.525198575732676,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008937427723160688,
      "loss": 5.146,
      "step": 3835
    },
    {
      "epoch": 0.5253355245138318,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000893735962749872,
      "loss": 5.2079,
      "step": 3836
    },
    {
      "epoch": 0.5254724732949877,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008937291495063295,
      "loss": 5.1601,
      "step": 3837
    },
    {
      "epoch": 0.5256094220761435,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000893722332585498,
      "loss": 5.1651,
      "step": 3838
    },
    {
      "epoch": 0.5257463708572994,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008937155119874337,
      "loss": 5.1659,
      "step": 3839
    },
    {
      "epoch": 0.5258833196384552,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008937086877121932,
      "loss": 5.0939,
      "step": 3840
    },
    {
      "epoch": 0.526020268419611,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008937018597598331,
      "loss": 5.1052,
      "step": 3841
    },
    {
      "epoch": 0.5261572172007669,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008936950281304099,
      "loss": 5.1263,
      "step": 3842
    },
    {
      "epoch": 0.5262941659819228,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008936881928239803,
      "loss": 5.1175,
      "step": 3843
    },
    {
      "epoch": 0.5264311147630786,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008936813538406009,
      "loss": 5.1296,
      "step": 3844
    },
    {
      "epoch": 0.5265680635442345,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008936745111803285,
      "loss": 5.1354,
      "step": 3845
    },
    {
      "epoch": 0.5267050123253904,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008936676648432197,
      "loss": 5.1517,
      "step": 3846
    },
    {
      "epoch": 0.5268419611065461,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008936608148293311,
      "loss": 5.1085,
      "step": 3847
    },
    {
      "epoch": 0.526978909887702,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008936539611387197,
      "loss": 5.1517,
      "step": 3848
    },
    {
      "epoch": 0.5271158586688578,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008936471037714422,
      "loss": 5.1272,
      "step": 3849
    },
    {
      "epoch": 0.5272528074500137,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008936402427275554,
      "loss": 5.1064,
      "step": 3850
    },
    {
      "epoch": 0.5273897562311696,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008936333780071163,
      "loss": 5.1045,
      "step": 3851
    },
    {
      "epoch": 0.5275267050123253,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008936265096101815,
      "loss": 5.191,
      "step": 3852
    },
    {
      "epoch": 0.5276636537934812,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008936196375368083,
      "loss": 5.2173,
      "step": 3853
    },
    {
      "epoch": 0.5278006025746371,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008936127617870534,
      "loss": 5.076,
      "step": 3854
    },
    {
      "epoch": 0.5279375513557929,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008936058823609737,
      "loss": 5.1446,
      "step": 3855
    },
    {
      "epoch": 0.5280745001369488,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008935989992586265,
      "loss": 5.1659,
      "step": 3856
    },
    {
      "epoch": 0.5282114489181047,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008935921124800686,
      "loss": 5.1071,
      "step": 3857
    },
    {
      "epoch": 0.5283483976992605,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008935852220253572,
      "loss": 5.1692,
      "step": 3858
    },
    {
      "epoch": 0.5284853464804163,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008935783278945494,
      "loss": 5.0987,
      "step": 3859
    },
    {
      "epoch": 0.5286222952615721,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008935714300877022,
      "loss": 5.1658,
      "step": 3860
    },
    {
      "epoch": 0.528759244042728,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000893564528604873,
      "loss": 5.1472,
      "step": 3861
    },
    {
      "epoch": 0.5288961928238839,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008935576234461187,
      "loss": 5.1883,
      "step": 3862
    },
    {
      "epoch": 0.5290331416050397,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008935507146114968,
      "loss": 5.1412,
      "step": 3863
    },
    {
      "epoch": 0.5291700903861956,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008935438021010644,
      "loss": 5.0811,
      "step": 3864
    },
    {
      "epoch": 0.5293070391673514,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008935368859148789,
      "loss": 5.1287,
      "step": 3865
    },
    {
      "epoch": 0.5294439879485072,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008935299660529974,
      "loss": 5.2023,
      "step": 3866
    },
    {
      "epoch": 0.5295809367296631,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008935230425154775,
      "loss": 5.2156,
      "step": 3867
    },
    {
      "epoch": 0.529717885510819,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008935161153023763,
      "loss": 5.135,
      "step": 3868
    },
    {
      "epoch": 0.5298548342919748,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008935091844137515,
      "loss": 5.2183,
      "step": 3869
    },
    {
      "epoch": 0.5299917830731307,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008935022498496603,
      "loss": 5.14,
      "step": 3870
    },
    {
      "epoch": 0.5301287318542866,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008934953116101604,
      "loss": 5.1668,
      "step": 3871
    },
    {
      "epoch": 0.5302656806354423,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008934883696953091,
      "loss": 5.1812,
      "step": 3872
    },
    {
      "epoch": 0.5304026294165982,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000893481424105164,
      "loss": 5.1067,
      "step": 3873
    },
    {
      "epoch": 0.530539578197754,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008934744748397827,
      "loss": 5.1371,
      "step": 3874
    },
    {
      "epoch": 0.5306765269789099,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008934675218992227,
      "loss": 5.1175,
      "step": 3875
    },
    {
      "epoch": 0.5308134757600658,
      "grad_norm": 0.5,
      "learning_rate": 0.0008934605652835419,
      "loss": 5.1306,
      "step": 3876
    },
    {
      "epoch": 0.5309504245412215,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008934536049927975,
      "loss": 5.1409,
      "step": 3877
    },
    {
      "epoch": 0.5310873733223774,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008934466410270474,
      "loss": 5.195,
      "step": 3878
    },
    {
      "epoch": 0.5312243221035333,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008934396733863494,
      "loss": 5.0573,
      "step": 3879
    },
    {
      "epoch": 0.5313612708846891,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000893432702070761,
      "loss": 5.1965,
      "step": 3880
    },
    {
      "epoch": 0.531498219665845,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008934257270803404,
      "loss": 5.103,
      "step": 3881
    },
    {
      "epoch": 0.5316351684470009,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008934187484151448,
      "loss": 5.1827,
      "step": 3882
    },
    {
      "epoch": 0.5317721172281566,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008934117660752328,
      "loss": 5.1887,
      "step": 3883
    },
    {
      "epoch": 0.5319090660093125,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008934047800606615,
      "loss": 5.1635,
      "step": 3884
    },
    {
      "epoch": 0.5320460147904683,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008933977903714893,
      "loss": 5.2136,
      "step": 3885
    },
    {
      "epoch": 0.5321829635716242,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008933907970077738,
      "loss": 5.0641,
      "step": 3886
    },
    {
      "epoch": 0.5323199123527801,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008933837999695732,
      "loss": 5.0376,
      "step": 3887
    },
    {
      "epoch": 0.5324568611339359,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008933767992569453,
      "loss": 5.1748,
      "step": 3888
    },
    {
      "epoch": 0.5325938099150918,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008933697948699484,
      "loss": 5.1343,
      "step": 3889
    },
    {
      "epoch": 0.5327307586962476,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008933627868086402,
      "loss": 5.1281,
      "step": 3890
    },
    {
      "epoch": 0.5328677074774034,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008933557750730792,
      "loss": 5.2363,
      "step": 3891
    },
    {
      "epoch": 0.5330046562585593,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000893348759663323,
      "loss": 5.0643,
      "step": 3892
    },
    {
      "epoch": 0.5331416050397152,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00089334174057943,
      "loss": 5.1809,
      "step": 3893
    },
    {
      "epoch": 0.533278553820871,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008933347178214584,
      "loss": 5.1903,
      "step": 3894
    },
    {
      "epoch": 0.5334155026020269,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008933276913894663,
      "loss": 5.1139,
      "step": 3895
    },
    {
      "epoch": 0.5335524513831826,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008933206612835121,
      "loss": 5.1603,
      "step": 3896
    },
    {
      "epoch": 0.5336894001643385,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008933136275036538,
      "loss": 5.2062,
      "step": 3897
    },
    {
      "epoch": 0.5338263489454944,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008933065900499499,
      "loss": 5.1523,
      "step": 3898
    },
    {
      "epoch": 0.5339632977266502,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008932995489224588,
      "loss": 5.1346,
      "step": 3899
    },
    {
      "epoch": 0.5341002465078061,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008932925041212384,
      "loss": 5.0815,
      "step": 3900
    },
    {
      "epoch": 0.534237195288962,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008932854556463476,
      "loss": 5.1509,
      "step": 3901
    },
    {
      "epoch": 0.5343741440701177,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008932784034978446,
      "loss": 5.1751,
      "step": 3902
    },
    {
      "epoch": 0.5345110928512736,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008932713476757877,
      "loss": 5.2443,
      "step": 3903
    },
    {
      "epoch": 0.5346480416324295,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008932642881802357,
      "loss": 5.08,
      "step": 3904
    },
    {
      "epoch": 0.5347849904135853,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008932572250112468,
      "loss": 5.0438,
      "step": 3905
    },
    {
      "epoch": 0.5349219391947412,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008932501581688796,
      "loss": 5.1601,
      "step": 3906
    },
    {
      "epoch": 0.5350588879758971,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008932430876531928,
      "loss": 5.1291,
      "step": 3907
    },
    {
      "epoch": 0.5351958367570528,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000893236013464245,
      "loss": 5.128,
      "step": 3908
    },
    {
      "epoch": 0.5353327855382087,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008932289356020945,
      "loss": 5.1216,
      "step": 3909
    },
    {
      "epoch": 0.5354697343193645,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008932218540668004,
      "loss": 5.1932,
      "step": 3910
    },
    {
      "epoch": 0.5356066831005204,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008932147688584212,
      "loss": 5.1995,
      "step": 3911
    },
    {
      "epoch": 0.5357436318816763,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008932076799770155,
      "loss": 5.1078,
      "step": 3912
    },
    {
      "epoch": 0.5358805806628321,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008932005874226421,
      "loss": 5.1554,
      "step": 3913
    },
    {
      "epoch": 0.536017529443988,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008931934911953599,
      "loss": 5.0971,
      "step": 3914
    },
    {
      "epoch": 0.5361544782251438,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008931863912952277,
      "loss": 5.1405,
      "step": 3915
    },
    {
      "epoch": 0.5362914270062996,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008931792877223042,
      "loss": 5.1323,
      "step": 3916
    },
    {
      "epoch": 0.5364283757874555,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008931721804766483,
      "loss": 5.1481,
      "step": 3917
    },
    {
      "epoch": 0.5365653245686114,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000893165069558319,
      "loss": 5.2024,
      "step": 3918
    },
    {
      "epoch": 0.5367022733497672,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000893157954967375,
      "loss": 5.067,
      "step": 3919
    },
    {
      "epoch": 0.536839222130923,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008931508367038757,
      "loss": 5.1827,
      "step": 3920
    },
    {
      "epoch": 0.5369761709120788,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008931437147678798,
      "loss": 5.1608,
      "step": 3921
    },
    {
      "epoch": 0.5371131196932347,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000893136589159446,
      "loss": 5.1756,
      "step": 3922
    },
    {
      "epoch": 0.5372500684743906,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008931294598786341,
      "loss": 5.1087,
      "step": 3923
    },
    {
      "epoch": 0.5373870172555464,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008931223269255025,
      "loss": 5.162,
      "step": 3924
    },
    {
      "epoch": 0.5375239660367023,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008931151903001107,
      "loss": 5.076,
      "step": 3925
    },
    {
      "epoch": 0.5376609148178582,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008931080500025177,
      "loss": 5.1636,
      "step": 3926
    },
    {
      "epoch": 0.5377978635990139,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008931009060327827,
      "loss": 5.136,
      "step": 3927
    },
    {
      "epoch": 0.5379348123801698,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008930937583909649,
      "loss": 5.1362,
      "step": 3928
    },
    {
      "epoch": 0.5380717611613257,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008930866070771236,
      "loss": 5.129,
      "step": 3929
    },
    {
      "epoch": 0.5382087099424815,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000893079452091318,
      "loss": 5.1587,
      "step": 3930
    },
    {
      "epoch": 0.5383456587236374,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008930722934336073,
      "loss": 5.1085,
      "step": 3931
    },
    {
      "epoch": 0.5384826075047932,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000893065131104051,
      "loss": 5.0807,
      "step": 3932
    },
    {
      "epoch": 0.538619556285949,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008930579651027084,
      "loss": 5.1554,
      "step": 3933
    },
    {
      "epoch": 0.5387565050671049,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008930507954296389,
      "loss": 5.1013,
      "step": 3934
    },
    {
      "epoch": 0.5388934538482607,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008930436220849018,
      "loss": 5.1764,
      "step": 3935
    },
    {
      "epoch": 0.5390304026294166,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008930364450685566,
      "loss": 5.1992,
      "step": 3936
    },
    {
      "epoch": 0.5391673514105725,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008930292643806628,
      "loss": 5.164,
      "step": 3937
    },
    {
      "epoch": 0.5393043001917283,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00089302208002128,
      "loss": 5.2368,
      "step": 3938
    },
    {
      "epoch": 0.5394412489728841,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008930148919904676,
      "loss": 5.2202,
      "step": 3939
    },
    {
      "epoch": 0.53957819775404,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008930077002882854,
      "loss": 5.1308,
      "step": 3940
    },
    {
      "epoch": 0.5397151465351958,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008930005049147926,
      "loss": 5.0708,
      "step": 3941
    },
    {
      "epoch": 0.5398520953163517,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008929933058700493,
      "loss": 5.1712,
      "step": 3942
    },
    {
      "epoch": 0.5399890440975076,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008929861031541146,
      "loss": 5.1546,
      "step": 3943
    },
    {
      "epoch": 0.5401259928786634,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008929788967670487,
      "loss": 5.1692,
      "step": 3944
    },
    {
      "epoch": 0.5402629416598193,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008929716867089112,
      "loss": 5.1445,
      "step": 3945
    },
    {
      "epoch": 0.540399890440975,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008929644729797616,
      "loss": 5.1499,
      "step": 3946
    },
    {
      "epoch": 0.5405368392221309,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008929572555796601,
      "loss": 5.1494,
      "step": 3947
    },
    {
      "epoch": 0.5406737880032868,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000892950034508666,
      "loss": 5.1555,
      "step": 3948
    },
    {
      "epoch": 0.5408107367844426,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008929428097668395,
      "loss": 5.0987,
      "step": 3949
    },
    {
      "epoch": 0.5409476855655985,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008929355813542404,
      "loss": 5.1338,
      "step": 3950
    },
    {
      "epoch": 0.5410846343467544,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008929283492709286,
      "loss": 5.1499,
      "step": 3951
    },
    {
      "epoch": 0.5412215831279101,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000892921113516964,
      "loss": 5.0963,
      "step": 3952
    },
    {
      "epoch": 0.541358531909066,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008929138740924066,
      "loss": 5.1068,
      "step": 3953
    },
    {
      "epoch": 0.5414954806902219,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008929066309973164,
      "loss": 5.1509,
      "step": 3954
    },
    {
      "epoch": 0.5416324294713777,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008928993842317535,
      "loss": 5.1405,
      "step": 3955
    },
    {
      "epoch": 0.5417693782525336,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008928921337957777,
      "loss": 5.1831,
      "step": 3956
    },
    {
      "epoch": 0.5419063270336894,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008928848796894493,
      "loss": 5.1369,
      "step": 3957
    },
    {
      "epoch": 0.5420432758148452,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008928776219128285,
      "loss": 5.2009,
      "step": 3958
    },
    {
      "epoch": 0.5421802245960011,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008928703604659751,
      "loss": 5.1654,
      "step": 3959
    },
    {
      "epoch": 0.5423171733771569,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008928630953489496,
      "loss": 5.1798,
      "step": 3960
    },
    {
      "epoch": 0.5424541221583128,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008928558265618122,
      "loss": 5.1561,
      "step": 3961
    },
    {
      "epoch": 0.5425910709394687,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008928485541046229,
      "loss": 5.2106,
      "step": 3962
    },
    {
      "epoch": 0.5427280197206245,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008928412779774421,
      "loss": 5.1655,
      "step": 3963
    },
    {
      "epoch": 0.5428649685017803,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008928339981803302,
      "loss": 5.1704,
      "step": 3964
    },
    {
      "epoch": 0.5430019172829362,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008928267147133473,
      "loss": 5.0474,
      "step": 3965
    },
    {
      "epoch": 0.543138866064092,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000892819427576554,
      "loss": 5.1844,
      "step": 3966
    },
    {
      "epoch": 0.5432758148452479,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008928121367700106,
      "loss": 5.1139,
      "step": 3967
    },
    {
      "epoch": 0.5434127636264037,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008928048422937776,
      "loss": 5.1975,
      "step": 3968
    },
    {
      "epoch": 0.5435497124075596,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008927975441479151,
      "loss": 5.1274,
      "step": 3969
    },
    {
      "epoch": 0.5436866611887154,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000892790242332484,
      "loss": 5.119,
      "step": 3970
    },
    {
      "epoch": 0.5438236099698712,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008927829368475446,
      "loss": 5.1996,
      "step": 3971
    },
    {
      "epoch": 0.5439605587510271,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008927756276931576,
      "loss": 5.0869,
      "step": 3972
    },
    {
      "epoch": 0.544097507532183,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008927683148693833,
      "loss": 5.1152,
      "step": 3973
    },
    {
      "epoch": 0.5442344563133388,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008927609983762825,
      "loss": 5.1453,
      "step": 3974
    },
    {
      "epoch": 0.5443714050944947,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008927536782139158,
      "loss": 5.1198,
      "step": 3975
    },
    {
      "epoch": 0.5445083538756506,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000892746354382344,
      "loss": 5.0823,
      "step": 3976
    },
    {
      "epoch": 0.5446453026568063,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008927390268816277,
      "loss": 5.1358,
      "step": 3977
    },
    {
      "epoch": 0.5447822514379622,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008927316957118274,
      "loss": 5.1195,
      "step": 3978
    },
    {
      "epoch": 0.5449192002191181,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008927243608730041,
      "loss": 5.1486,
      "step": 3979
    },
    {
      "epoch": 0.5450561490002739,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008927170223652185,
      "loss": 5.1357,
      "step": 3980
    },
    {
      "epoch": 0.5451930977814298,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008927096801885315,
      "loss": 5.0945,
      "step": 3981
    },
    {
      "epoch": 0.5453300465625855,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008927023343430038,
      "loss": 5.1407,
      "step": 3982
    },
    {
      "epoch": 0.5454669953437414,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008926949848286964,
      "loss": 5.116,
      "step": 3983
    },
    {
      "epoch": 0.5456039441248973,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008926876316456703,
      "loss": 5.2612,
      "step": 3984
    },
    {
      "epoch": 0.5457408929060531,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008926802747939861,
      "loss": 5.2142,
      "step": 3985
    },
    {
      "epoch": 0.545877841687209,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008926729142737052,
      "loss": 5.0138,
      "step": 3986
    },
    {
      "epoch": 0.5460147904683649,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008926655500848882,
      "loss": 5.1424,
      "step": 3987
    },
    {
      "epoch": 0.5461517392495207,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008926581822275965,
      "loss": 5.1526,
      "step": 3988
    },
    {
      "epoch": 0.5462886880306765,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008926508107018908,
      "loss": 5.1263,
      "step": 3989
    },
    {
      "epoch": 0.5464256368118324,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008926434355078323,
      "loss": 5.1912,
      "step": 3990
    },
    {
      "epoch": 0.5465625855929882,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008926360566454823,
      "loss": 5.0703,
      "step": 3991
    },
    {
      "epoch": 0.5466995343741441,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008926286741149017,
      "loss": 5.1473,
      "step": 3992
    },
    {
      "epoch": 0.5468364831552999,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008926212879161519,
      "loss": 5.1219,
      "step": 3993
    },
    {
      "epoch": 0.5469734319364558,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008926138980492941,
      "loss": 5.139,
      "step": 3994
    },
    {
      "epoch": 0.5471103807176116,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008926065045143894,
      "loss": 5.1223,
      "step": 3995
    },
    {
      "epoch": 0.5472473294987674,
      "grad_norm": 0.46875,
      "learning_rate": 0.000892599107311499,
      "loss": 5.0561,
      "step": 3996
    },
    {
      "epoch": 0.5473842782799233,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008925917064406845,
      "loss": 5.1092,
      "step": 3997
    },
    {
      "epoch": 0.5475212270610792,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000892584301902007,
      "loss": 5.1851,
      "step": 3998
    },
    {
      "epoch": 0.547658175842235,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008925768936955279,
      "loss": 5.197,
      "step": 3999
    },
    {
      "epoch": 0.5477951246233909,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008925694818213085,
      "loss": 5.1503,
      "step": 4000
    },
    {
      "epoch": 0.5479320734045467,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008925620662794105,
      "loss": 5.1129,
      "step": 4001
    },
    {
      "epoch": 0.5480690221857025,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008925546470698952,
      "loss": 5.0975,
      "step": 4002
    },
    {
      "epoch": 0.5482059709668584,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008925472241928241,
      "loss": 5.142,
      "step": 4003
    },
    {
      "epoch": 0.5483429197480142,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008925397976482587,
      "loss": 5.1973,
      "step": 4004
    },
    {
      "epoch": 0.5484798685291701,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008925323674362605,
      "loss": 5.116,
      "step": 4005
    },
    {
      "epoch": 0.548616817310326,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008925249335568912,
      "loss": 5.1783,
      "step": 4006
    },
    {
      "epoch": 0.5487537660914817,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008925174960102122,
      "loss": 5.1365,
      "step": 4007
    },
    {
      "epoch": 0.5488907148726376,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008925100547962854,
      "loss": 5.0796,
      "step": 4008
    },
    {
      "epoch": 0.5490276636537935,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008925026099151722,
      "loss": 5.2006,
      "step": 4009
    },
    {
      "epoch": 0.5491646124349493,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008924951613669343,
      "loss": 5.1415,
      "step": 4010
    },
    {
      "epoch": 0.5493015612161052,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000892487709151634,
      "loss": 5.0848,
      "step": 4011
    },
    {
      "epoch": 0.5494385099972611,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008924802532693322,
      "loss": 5.1028,
      "step": 4012
    },
    {
      "epoch": 0.5495754587784168,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008924727937200912,
      "loss": 5.1665,
      "step": 4013
    },
    {
      "epoch": 0.5497124075595727,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000892465330503973,
      "loss": 5.1552,
      "step": 4014
    },
    {
      "epoch": 0.5498493563407285,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008924578636210388,
      "loss": 5.0178,
      "step": 4015
    },
    {
      "epoch": 0.5499863051218844,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008924503930713511,
      "loss": 5.1136,
      "step": 4016
    },
    {
      "epoch": 0.5501232539030403,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008924429188549714,
      "loss": 5.1712,
      "step": 4017
    },
    {
      "epoch": 0.5502602026841961,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008924354409719619,
      "loss": 5.1919,
      "step": 4018
    },
    {
      "epoch": 0.550397151465352,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008924279594223844,
      "loss": 5.1331,
      "step": 4019
    },
    {
      "epoch": 0.5505341002465078,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008924204742063011,
      "loss": 5.1893,
      "step": 4020
    },
    {
      "epoch": 0.5506710490276636,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008924129853237738,
      "loss": 5.1361,
      "step": 4021
    },
    {
      "epoch": 0.5508079978088195,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008924054927748647,
      "loss": 5.1419,
      "step": 4022
    },
    {
      "epoch": 0.5509449465899754,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000892397996559636,
      "loss": 5.147,
      "step": 4023
    },
    {
      "epoch": 0.5510818953711312,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008923904966781495,
      "loss": 5.115,
      "step": 4024
    },
    {
      "epoch": 0.5512188441522871,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008923829931304675,
      "loss": 5.1753,
      "step": 4025
    },
    {
      "epoch": 0.551355792933443,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008923754859166523,
      "loss": 5.0966,
      "step": 4026
    },
    {
      "epoch": 0.5514927417145987,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008923679750367661,
      "loss": 5.1309,
      "step": 4027
    },
    {
      "epoch": 0.5516296904957546,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000892360460490871,
      "loss": 5.1064,
      "step": 4028
    },
    {
      "epoch": 0.5517666392769104,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008923529422790294,
      "loss": 5.1307,
      "step": 4029
    },
    {
      "epoch": 0.5519035880580663,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008923454204013035,
      "loss": 5.1861,
      "step": 4030
    },
    {
      "epoch": 0.5520405368392222,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008923378948577558,
      "loss": 5.1639,
      "step": 4031
    },
    {
      "epoch": 0.5521774856203779,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008923303656484484,
      "loss": 5.1376,
      "step": 4032
    },
    {
      "epoch": 0.5523144344015338,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000892322832773444,
      "loss": 5.151,
      "step": 4033
    },
    {
      "epoch": 0.5524513831826897,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008923152962328049,
      "loss": 5.1343,
      "step": 4034
    },
    {
      "epoch": 0.5525883319638455,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008923077560265935,
      "loss": 5.1224,
      "step": 4035
    },
    {
      "epoch": 0.5527252807450014,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008923002121548723,
      "loss": 5.1718,
      "step": 4036
    },
    {
      "epoch": 0.5528622295261573,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008922926646177038,
      "loss": 5.1635,
      "step": 4037
    },
    {
      "epoch": 0.552999178307313,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008922851134151507,
      "loss": 5.1447,
      "step": 4038
    },
    {
      "epoch": 0.5531361270884689,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008922775585472754,
      "loss": 5.033,
      "step": 4039
    },
    {
      "epoch": 0.5532730758696247,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008922700000141406,
      "loss": 5.0671,
      "step": 4040
    },
    {
      "epoch": 0.5534100246507806,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000892262437815809,
      "loss": 5.1541,
      "step": 4041
    },
    {
      "epoch": 0.5535469734319365,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008922548719523431,
      "loss": 5.1987,
      "step": 4042
    },
    {
      "epoch": 0.5536839222130923,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008922473024238057,
      "loss": 5.1203,
      "step": 4043
    },
    {
      "epoch": 0.5538208709942481,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008922397292302596,
      "loss": 5.0388,
      "step": 4044
    },
    {
      "epoch": 0.553957819775404,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008922321523717674,
      "loss": 5.1238,
      "step": 4045
    },
    {
      "epoch": 0.5540947685565598,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000892224571848392,
      "loss": 5.147,
      "step": 4046
    },
    {
      "epoch": 0.5542317173377157,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008922169876601962,
      "loss": 5.1783,
      "step": 4047
    },
    {
      "epoch": 0.5543686661188716,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008922093998072428,
      "loss": 5.1249,
      "step": 4048
    },
    {
      "epoch": 0.5545056149000274,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008922018082895949,
      "loss": 5.1294,
      "step": 4049
    },
    {
      "epoch": 0.5546425636811833,
      "grad_norm": 0.609375,
      "learning_rate": 0.000892194213107315,
      "loss": 5.1561,
      "step": 4050
    },
    {
      "epoch": 0.554779512462339,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008921866142604664,
      "loss": 5.1181,
      "step": 4051
    },
    {
      "epoch": 0.5549164612434949,
      "grad_norm": 0.6640625,
      "learning_rate": 0.000892179011749112,
      "loss": 5.2164,
      "step": 4052
    },
    {
      "epoch": 0.5550534100246508,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008921714055733148,
      "loss": 5.23,
      "step": 4053
    },
    {
      "epoch": 0.5551903588058066,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008921637957331378,
      "loss": 5.091,
      "step": 4054
    },
    {
      "epoch": 0.5553273075869625,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008921561822286439,
      "loss": 5.1214,
      "step": 4055
    },
    {
      "epoch": 0.5554642563681184,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008921485650598965,
      "loss": 5.0258,
      "step": 4056
    },
    {
      "epoch": 0.5556012051492741,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008921409442269585,
      "loss": 5.1315,
      "step": 4057
    },
    {
      "epoch": 0.55573815393043,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008921333197298932,
      "loss": 5.1354,
      "step": 4058
    },
    {
      "epoch": 0.5558751027115859,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008921256915687636,
      "loss": 5.1046,
      "step": 4059
    },
    {
      "epoch": 0.5560120514927417,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008921180597436331,
      "loss": 5.1482,
      "step": 4060
    },
    {
      "epoch": 0.5561490002738976,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008921104242545651,
      "loss": 5.0755,
      "step": 4061
    },
    {
      "epoch": 0.5562859490550535,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008921027851016225,
      "loss": 5.1087,
      "step": 4062
    },
    {
      "epoch": 0.5564228978362092,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008920951422848687,
      "loss": 5.1638,
      "step": 4063
    },
    {
      "epoch": 0.5565598466173651,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008920874958043672,
      "loss": 5.1196,
      "step": 4064
    },
    {
      "epoch": 0.5566967953985209,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008920798456601813,
      "loss": 5.0617,
      "step": 4065
    },
    {
      "epoch": 0.5568337441796768,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008920721918523743,
      "loss": 5.1748,
      "step": 4066
    },
    {
      "epoch": 0.5569706929608327,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008920645343810097,
      "loss": 5.1117,
      "step": 4067
    },
    {
      "epoch": 0.5571076417419885,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008920568732461511,
      "loss": 5.1657,
      "step": 4068
    },
    {
      "epoch": 0.5572445905231443,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008920492084478616,
      "loss": 5.1339,
      "step": 4069
    },
    {
      "epoch": 0.5573815393043002,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008920415399862052,
      "loss": 5.1806,
      "step": 4070
    },
    {
      "epoch": 0.557518488085456,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008920338678612452,
      "loss": 5.1481,
      "step": 4071
    },
    {
      "epoch": 0.5576554368666119,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008920261920730452,
      "loss": 5.1125,
      "step": 4072
    },
    {
      "epoch": 0.5577923856477678,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008920185126216687,
      "loss": 5.1228,
      "step": 4073
    },
    {
      "epoch": 0.5579293344289236,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008920108295071796,
      "loss": 5.0935,
      "step": 4074
    },
    {
      "epoch": 0.5580662832100795,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008920031427296413,
      "loss": 5.101,
      "step": 4075
    },
    {
      "epoch": 0.5582032319912352,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008919954522891178,
      "loss": 5.1003,
      "step": 4076
    },
    {
      "epoch": 0.5583401807723911,
      "grad_norm": 0.625,
      "learning_rate": 0.0008919877581856725,
      "loss": 5.0916,
      "step": 4077
    },
    {
      "epoch": 0.558477129553547,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008919800604193694,
      "loss": 5.143,
      "step": 4078
    },
    {
      "epoch": 0.5586140783347028,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008919723589902723,
      "loss": 5.2265,
      "step": 4079
    },
    {
      "epoch": 0.5587510271158587,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008919646538984448,
      "loss": 5.112,
      "step": 4080
    },
    {
      "epoch": 0.5588879758970146,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008919569451439509,
      "loss": 5.1394,
      "step": 4081
    },
    {
      "epoch": 0.5590249246781703,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008919492327268546,
      "loss": 5.1194,
      "step": 4082
    },
    {
      "epoch": 0.5591618734593262,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008919415166472196,
      "loss": 5.1223,
      "step": 4083
    },
    {
      "epoch": 0.5592988222404821,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00089193379690511,
      "loss": 5.1964,
      "step": 4084
    },
    {
      "epoch": 0.5594357710216379,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008919260735005898,
      "loss": 5.1792,
      "step": 4085
    },
    {
      "epoch": 0.5595727198027938,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008919183464337226,
      "loss": 5.0597,
      "step": 4086
    },
    {
      "epoch": 0.5597096685839495,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008919106157045731,
      "loss": 5.1506,
      "step": 4087
    },
    {
      "epoch": 0.5598466173651054,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008919028813132049,
      "loss": 5.1293,
      "step": 4088
    },
    {
      "epoch": 0.5599835661462613,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008918951432596823,
      "loss": 5.1732,
      "step": 4089
    },
    {
      "epoch": 0.5601205149274171,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008918874015440693,
      "loss": 5.064,
      "step": 4090
    },
    {
      "epoch": 0.560257463708573,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008918796561664302,
      "loss": 5.1976,
      "step": 4091
    },
    {
      "epoch": 0.5603944124897289,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000891871907126829,
      "loss": 5.1563,
      "step": 4092
    },
    {
      "epoch": 0.5605313612708847,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008918641544253299,
      "loss": 5.1408,
      "step": 4093
    },
    {
      "epoch": 0.5606683100520405,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008918563980619973,
      "loss": 5.0794,
      "step": 4094
    },
    {
      "epoch": 0.5608052588331964,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008918486380368956,
      "loss": 5.1044,
      "step": 4095
    },
    {
      "epoch": 0.5609422076143522,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008918408743500888,
      "loss": 5.1047,
      "step": 4096
    },
    {
      "epoch": 0.5610791563955081,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008918331070016414,
      "loss": 5.1447,
      "step": 4097
    },
    {
      "epoch": 0.561216105176664,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008918253359916179,
      "loss": 5.1216,
      "step": 4098
    },
    {
      "epoch": 0.5613530539578198,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008918175613200824,
      "loss": 5.0671,
      "step": 4099
    },
    {
      "epoch": 0.5614900027389756,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008918097829870997,
      "loss": 5.1414,
      "step": 4100
    },
    {
      "epoch": 0.5616269515201314,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008918020009927337,
      "loss": 5.0828,
      "step": 4101
    },
    {
      "epoch": 0.5617639003012873,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008917942153370495,
      "loss": 5.2285,
      "step": 4102
    },
    {
      "epoch": 0.5619008490824432,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008917864260201113,
      "loss": 5.0271,
      "step": 4103
    },
    {
      "epoch": 0.562037797863599,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008917786330419837,
      "loss": 5.1265,
      "step": 4104
    },
    {
      "epoch": 0.5621747466447549,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008917708364027313,
      "loss": 5.0508,
      "step": 4105
    },
    {
      "epoch": 0.5623116954259108,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008917630361024189,
      "loss": 5.1533,
      "step": 4106
    },
    {
      "epoch": 0.5624486442070665,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008917552321411106,
      "loss": 5.0652,
      "step": 4107
    },
    {
      "epoch": 0.5625855929882224,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008917474245188717,
      "loss": 5.1621,
      "step": 4108
    },
    {
      "epoch": 0.5627225417693783,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008917396132357665,
      "loss": 5.1506,
      "step": 4109
    },
    {
      "epoch": 0.5628594905505341,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00089173179829186,
      "loss": 5.1233,
      "step": 4110
    },
    {
      "epoch": 0.56299643933169,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008917239796872166,
      "loss": 5.0943,
      "step": 4111
    },
    {
      "epoch": 0.5631333881128457,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008917161574219015,
      "loss": 5.1521,
      "step": 4112
    },
    {
      "epoch": 0.5632703368940016,
      "grad_norm": 0.5,
      "learning_rate": 0.0008917083314959793,
      "loss": 5.096,
      "step": 4113
    },
    {
      "epoch": 0.5634072856751575,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008917005019095149,
      "loss": 5.0576,
      "step": 4114
    },
    {
      "epoch": 0.5635442344563133,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008916926686625733,
      "loss": 5.1042,
      "step": 4115
    },
    {
      "epoch": 0.5636811832374692,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008916848317552192,
      "loss": 5.1531,
      "step": 4116
    },
    {
      "epoch": 0.5638181320186251,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008916769911875177,
      "loss": 5.124,
      "step": 4117
    },
    {
      "epoch": 0.5639550807997809,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008916691469595336,
      "loss": 5.0494,
      "step": 4118
    },
    {
      "epoch": 0.5640920295809367,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008916612990713322,
      "loss": 5.1365,
      "step": 4119
    },
    {
      "epoch": 0.5642289783620926,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008916534475229783,
      "loss": 5.1555,
      "step": 4120
    },
    {
      "epoch": 0.5643659271432484,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000891645592314537,
      "loss": 5.1066,
      "step": 4121
    },
    {
      "epoch": 0.5645028759244043,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008916377334460734,
      "loss": 5.1121,
      "step": 4122
    },
    {
      "epoch": 0.5646398247055601,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008916298709176527,
      "loss": 5.1192,
      "step": 4123
    },
    {
      "epoch": 0.564776773486716,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008916220047293402,
      "loss": 5.0899,
      "step": 4124
    },
    {
      "epoch": 0.5649137222678718,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008916141348812007,
      "loss": 5.1306,
      "step": 4125
    },
    {
      "epoch": 0.5650506710490276,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008916062613732997,
      "loss": 5.1009,
      "step": 4126
    },
    {
      "epoch": 0.5651876198301835,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008915983842057024,
      "loss": 5.1604,
      "step": 4127
    },
    {
      "epoch": 0.5653245686113394,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000891590503378474,
      "loss": 5.1697,
      "step": 4128
    },
    {
      "epoch": 0.5654615173924952,
      "grad_norm": 0.515625,
      "learning_rate": 0.00089158261889168,
      "loss": 5.0808,
      "step": 4129
    },
    {
      "epoch": 0.5655984661736511,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008915747307453855,
      "loss": 5.0703,
      "step": 4130
    },
    {
      "epoch": 0.565735414954807,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000891566838939656,
      "loss": 5.1581,
      "step": 4131
    },
    {
      "epoch": 0.5658723637359627,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008915589434745569,
      "loss": 5.1014,
      "step": 4132
    },
    {
      "epoch": 0.5660093125171186,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008915510443501537,
      "loss": 5.1388,
      "step": 4133
    },
    {
      "epoch": 0.5661462612982745,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008915431415665116,
      "loss": 5.1046,
      "step": 4134
    },
    {
      "epoch": 0.5662832100794303,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008915352351236965,
      "loss": 5.2011,
      "step": 4135
    },
    {
      "epoch": 0.5664201588605862,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008915273250217735,
      "loss": 5.1758,
      "step": 4136
    },
    {
      "epoch": 0.5665571076417419,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008915194112608085,
      "loss": 5.0752,
      "step": 4137
    },
    {
      "epoch": 0.5666940564228978,
      "grad_norm": 0.48828125,
      "learning_rate": 0.000891511493840867,
      "loss": 5.1381,
      "step": 4138
    },
    {
      "epoch": 0.5668310052040537,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008915035727620143,
      "loss": 5.1813,
      "step": 4139
    },
    {
      "epoch": 0.5669679539852095,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008914956480243164,
      "loss": 5.1174,
      "step": 4140
    },
    {
      "epoch": 0.5671049027663654,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008914877196278388,
      "loss": 5.0671,
      "step": 4141
    },
    {
      "epoch": 0.5672418515475213,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008914797875726473,
      "loss": 5.1838,
      "step": 4142
    },
    {
      "epoch": 0.567378800328677,
      "grad_norm": 0.625,
      "learning_rate": 0.0008914718518588076,
      "loss": 5.1102,
      "step": 4143
    },
    {
      "epoch": 0.5675157491098329,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008914639124863855,
      "loss": 5.1755,
      "step": 4144
    },
    {
      "epoch": 0.5676526978909888,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008914559694554467,
      "loss": 5.1042,
      "step": 4145
    },
    {
      "epoch": 0.5677896466721446,
      "grad_norm": 0.5,
      "learning_rate": 0.0008914480227660571,
      "loss": 5.1511,
      "step": 4146
    },
    {
      "epoch": 0.5679265954533005,
      "grad_norm": 0.5,
      "learning_rate": 0.0008914400724182826,
      "loss": 5.0378,
      "step": 4147
    },
    {
      "epoch": 0.5680635442344563,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008914321184121891,
      "loss": 5.0554,
      "step": 4148
    },
    {
      "epoch": 0.5682004930156122,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008914241607478424,
      "loss": 5.0048,
      "step": 4149
    },
    {
      "epoch": 0.568337441796768,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008914161994253083,
      "loss": 5.0781,
      "step": 4150
    },
    {
      "epoch": 0.5684743905779238,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008914082344446531,
      "loss": 5.1827,
      "step": 4151
    },
    {
      "epoch": 0.5686113393590797,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008914002658059429,
      "loss": 5.1774,
      "step": 4152
    },
    {
      "epoch": 0.5687482881402356,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008913922935092432,
      "loss": 5.0931,
      "step": 4153
    },
    {
      "epoch": 0.5688852369213914,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008913843175546205,
      "loss": 5.1503,
      "step": 4154
    },
    {
      "epoch": 0.5690221857025473,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008913763379421408,
      "loss": 5.1957,
      "step": 4155
    },
    {
      "epoch": 0.5691591344837031,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008913683546718702,
      "loss": 5.1116,
      "step": 4156
    },
    {
      "epoch": 0.5692960832648589,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008913603677438749,
      "loss": 5.2032,
      "step": 4157
    },
    {
      "epoch": 0.5694330320460148,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000891352377158221,
      "loss": 5.056,
      "step": 4158
    },
    {
      "epoch": 0.5695699808271706,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008913443829149749,
      "loss": 5.1266,
      "step": 4159
    },
    {
      "epoch": 0.5697069296083265,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008913363850142025,
      "loss": 5.1054,
      "step": 4160
    },
    {
      "epoch": 0.5698438783894824,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008913283834559705,
      "loss": 5.0678,
      "step": 4161
    },
    {
      "epoch": 0.5699808271706381,
      "grad_norm": 0.515625,
      "learning_rate": 0.000891320378240345,
      "loss": 5.1056,
      "step": 4162
    },
    {
      "epoch": 0.570117775951794,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008913123693673924,
      "loss": 5.0874,
      "step": 4163
    },
    {
      "epoch": 0.5702547247329499,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008913043568371789,
      "loss": 5.116,
      "step": 4164
    },
    {
      "epoch": 0.5703916735141057,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008912963406497712,
      "loss": 5.1312,
      "step": 4165
    },
    {
      "epoch": 0.5705286222952616,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008912883208052354,
      "loss": 5.0529,
      "step": 4166
    },
    {
      "epoch": 0.5706655710764175,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008912802973036384,
      "loss": 5.1446,
      "step": 4167
    },
    {
      "epoch": 0.5708025198575732,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008912722701450461,
      "loss": 5.1497,
      "step": 4168
    },
    {
      "epoch": 0.5709394686387291,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008912642393295255,
      "loss": 5.0662,
      "step": 4169
    },
    {
      "epoch": 0.571076417419885,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008912562048571431,
      "loss": 5.1495,
      "step": 4170
    },
    {
      "epoch": 0.5712133662010408,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008912481667279652,
      "loss": 5.1111,
      "step": 4171
    },
    {
      "epoch": 0.5713503149821967,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008912401249420587,
      "loss": 5.0578,
      "step": 4172
    },
    {
      "epoch": 0.5714872637633525,
      "grad_norm": 0.5625,
      "learning_rate": 0.00089123207949949,
      "loss": 5.1225,
      "step": 4173
    },
    {
      "epoch": 0.5716242125445083,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008912240304003262,
      "loss": 5.1485,
      "step": 4174
    },
    {
      "epoch": 0.5717611613256642,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008912159776446336,
      "loss": 5.1831,
      "step": 4175
    },
    {
      "epoch": 0.57189811010682,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008912079212324791,
      "loss": 5.1154,
      "step": 4176
    },
    {
      "epoch": 0.5720350588879759,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008911998611639292,
      "loss": 5.1197,
      "step": 4177
    },
    {
      "epoch": 0.5721720076691318,
      "grad_norm": 0.494140625,
      "learning_rate": 0.000891191797439051,
      "loss": 5.1727,
      "step": 4178
    },
    {
      "epoch": 0.5723089564502876,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008911837300579114,
      "loss": 5.0921,
      "step": 4179
    },
    {
      "epoch": 0.5724459052314435,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008911756590205771,
      "loss": 5.1205,
      "step": 4180
    },
    {
      "epoch": 0.5725828540125993,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008911675843271148,
      "loss": 5.0989,
      "step": 4181
    },
    {
      "epoch": 0.5727198027937551,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008911595059775916,
      "loss": 5.0652,
      "step": 4182
    },
    {
      "epoch": 0.572856751574911,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008911514239720746,
      "loss": 5.1416,
      "step": 4183
    },
    {
      "epoch": 0.5729937003560668,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008911433383106307,
      "loss": 5.1588,
      "step": 4184
    },
    {
      "epoch": 0.5731306491372227,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008911352489933266,
      "loss": 5.084,
      "step": 4185
    },
    {
      "epoch": 0.5732675979183786,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008911271560202297,
      "loss": 5.1093,
      "step": 4186
    },
    {
      "epoch": 0.5734045466995343,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008911190593914069,
      "loss": 5.1213,
      "step": 4187
    },
    {
      "epoch": 0.5735414954806902,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008911109591069255,
      "loss": 5.0766,
      "step": 4188
    },
    {
      "epoch": 0.5736784442618461,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008911028551668523,
      "loss": 5.0568,
      "step": 4189
    },
    {
      "epoch": 0.5738153930430019,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008910947475712547,
      "loss": 5.1128,
      "step": 4190
    },
    {
      "epoch": 0.5739523418241578,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008910866363201998,
      "loss": 5.0513,
      "step": 4191
    },
    {
      "epoch": 0.5740892906053137,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008910785214137548,
      "loss": 5.0916,
      "step": 4192
    },
    {
      "epoch": 0.5742262393864694,
      "grad_norm": 0.421875,
      "learning_rate": 0.000891070402851987,
      "loss": 5.1297,
      "step": 4193
    },
    {
      "epoch": 0.5743631881676253,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008910622806349637,
      "loss": 5.1314,
      "step": 4194
    },
    {
      "epoch": 0.5745001369487811,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008910541547627523,
      "loss": 5.0982,
      "step": 4195
    },
    {
      "epoch": 0.574637085729937,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008910460252354198,
      "loss": 5.1205,
      "step": 4196
    },
    {
      "epoch": 0.5747740345110929,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008910378920530339,
      "loss": 5.1738,
      "step": 4197
    },
    {
      "epoch": 0.5749109832922487,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008910297552156619,
      "loss": 5.1262,
      "step": 4198
    },
    {
      "epoch": 0.5750479320734045,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008910216147233713,
      "loss": 5.1035,
      "step": 4199
    },
    {
      "epoch": 0.5751848808545604,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008910134705762293,
      "loss": 5.0641,
      "step": 4200
    },
    {
      "epoch": 0.5753218296357162,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008910053227743036,
      "loss": 5.0646,
      "step": 4201
    },
    {
      "epoch": 0.5754587784168721,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008909971713176619,
      "loss": 5.249,
      "step": 4202
    },
    {
      "epoch": 0.575595727198028,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008909890162063714,
      "loss": 5.1421,
      "step": 4203
    },
    {
      "epoch": 0.5757326759791838,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008909808574405,
      "loss": 5.1071,
      "step": 4204
    },
    {
      "epoch": 0.5758696247603396,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000890972695020115,
      "loss": 5.1328,
      "step": 4205
    },
    {
      "epoch": 0.5760065735414954,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008909645289452842,
      "loss": 5.1043,
      "step": 4206
    },
    {
      "epoch": 0.5761435223226513,
      "grad_norm": 0.375,
      "learning_rate": 0.0008909563592160753,
      "loss": 5.1886,
      "step": 4207
    },
    {
      "epoch": 0.5762804711038072,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000890948185832556,
      "loss": 5.1172,
      "step": 4208
    },
    {
      "epoch": 0.576417419884963,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008909400087947941,
      "loss": 5.1724,
      "step": 4209
    },
    {
      "epoch": 0.5765543686661189,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008909318281028571,
      "loss": 5.0842,
      "step": 4210
    },
    {
      "epoch": 0.5766913174472748,
      "grad_norm": 0.515625,
      "learning_rate": 0.000890923643756813,
      "loss": 5.105,
      "step": 4211
    },
    {
      "epoch": 0.5768282662284305,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008909154557567297,
      "loss": 5.133,
      "step": 4212
    },
    {
      "epoch": 0.5769652150095864,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008909072641026749,
      "loss": 5.1237,
      "step": 4213
    },
    {
      "epoch": 0.5771021637907423,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008908990687947166,
      "loss": 5.1157,
      "step": 4214
    },
    {
      "epoch": 0.5772391125718981,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008908908698329226,
      "loss": 5.1005,
      "step": 4215
    },
    {
      "epoch": 0.577376061353054,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008908826672173609,
      "loss": 5.1412,
      "step": 4216
    },
    {
      "epoch": 0.5775130101342099,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008908744609480994,
      "loss": 5.0738,
      "step": 4217
    },
    {
      "epoch": 0.5776499589153656,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008908662510252063,
      "loss": 5.1058,
      "step": 4218
    },
    {
      "epoch": 0.5777869076965215,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008908580374487495,
      "loss": 5.1237,
      "step": 4219
    },
    {
      "epoch": 0.5779238564776773,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000890849820218797,
      "loss": 5.1538,
      "step": 4220
    },
    {
      "epoch": 0.5780608052588332,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008908415993354172,
      "loss": 5.1044,
      "step": 4221
    },
    {
      "epoch": 0.5781977540399891,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008908333747986778,
      "loss": 5.1383,
      "step": 4222
    },
    {
      "epoch": 0.5783347028211449,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008908251466086474,
      "loss": 5.1131,
      "step": 4223
    },
    {
      "epoch": 0.5784716516023007,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008908169147653938,
      "loss": 5.1127,
      "step": 4224
    },
    {
      "epoch": 0.5786086003834566,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008908086792689853,
      "loss": 5.107,
      "step": 4225
    },
    {
      "epoch": 0.5787455491646124,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008908004401194905,
      "loss": 5.1549,
      "step": 4226
    },
    {
      "epoch": 0.5788824979457683,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008907921973169773,
      "loss": 5.1029,
      "step": 4227
    },
    {
      "epoch": 0.5790194467269242,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008907839508615141,
      "loss": 5.1106,
      "step": 4228
    },
    {
      "epoch": 0.57915639550808,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008907757007531692,
      "loss": 5.1378,
      "step": 4229
    },
    {
      "epoch": 0.5792933442892358,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008907674469920111,
      "loss": 5.0726,
      "step": 4230
    },
    {
      "epoch": 0.5794302930703916,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008907591895781082,
      "loss": 5.1437,
      "step": 4231
    },
    {
      "epoch": 0.5795672418515475,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008907509285115288,
      "loss": 5.152,
      "step": 4232
    },
    {
      "epoch": 0.5797041906327034,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008907426637923414,
      "loss": 5.1045,
      "step": 4233
    },
    {
      "epoch": 0.5798411394138592,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008907343954206147,
      "loss": 5.0786,
      "step": 4234
    },
    {
      "epoch": 0.5799780881950151,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008907261233964169,
      "loss": 5.1154,
      "step": 4235
    },
    {
      "epoch": 0.580115036976171,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008907178477198166,
      "loss": 5.0464,
      "step": 4236
    },
    {
      "epoch": 0.5802519857573267,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008907095683908826,
      "loss": 5.1251,
      "step": 4237
    },
    {
      "epoch": 0.5803889345384826,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008907012854096834,
      "loss": 5.0826,
      "step": 4238
    },
    {
      "epoch": 0.5805258833196385,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008906929987762876,
      "loss": 5.1409,
      "step": 4239
    },
    {
      "epoch": 0.5806628321007943,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000890684708490764,
      "loss": 5.0882,
      "step": 4240
    },
    {
      "epoch": 0.5807997808819502,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000890676414553181,
      "loss": 5.1375,
      "step": 4241
    },
    {
      "epoch": 0.5809367296631059,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008906681169636078,
      "loss": 5.1409,
      "step": 4242
    },
    {
      "epoch": 0.5810736784442618,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008906598157221128,
      "loss": 5.1333,
      "step": 4243
    },
    {
      "epoch": 0.5812106272254177,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008906515108287648,
      "loss": 5.1556,
      "step": 4244
    },
    {
      "epoch": 0.5813475760065735,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008906432022836329,
      "loss": 5.1783,
      "step": 4245
    },
    {
      "epoch": 0.5814845247877294,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008906348900867856,
      "loss": 5.1293,
      "step": 4246
    },
    {
      "epoch": 0.5816214735688853,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008906265742382922,
      "loss": 5.1318,
      "step": 4247
    },
    {
      "epoch": 0.581758422350041,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008906182547382212,
      "loss": 5.1027,
      "step": 4248
    },
    {
      "epoch": 0.5818953711311969,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008906099315866417,
      "loss": 5.1532,
      "step": 4249
    },
    {
      "epoch": 0.5820323199123528,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008906016047836228,
      "loss": 5.0991,
      "step": 4250
    },
    {
      "epoch": 0.5821692686935086,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008905932743292334,
      "loss": 5.0671,
      "step": 4251
    },
    {
      "epoch": 0.5823062174746645,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008905849402235426,
      "loss": 5.1333,
      "step": 4252
    },
    {
      "epoch": 0.5824431662558204,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008905766024666194,
      "loss": 5.1389,
      "step": 4253
    },
    {
      "epoch": 0.5825801150369762,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008905682610585329,
      "loss": 5.0736,
      "step": 4254
    },
    {
      "epoch": 0.582717063818132,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008905599159993522,
      "loss": 5.0964,
      "step": 4255
    },
    {
      "epoch": 0.5828540125992878,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008905515672891465,
      "loss": 5.1175,
      "step": 4256
    },
    {
      "epoch": 0.5829909613804437,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000890543214927985,
      "loss": 5.0704,
      "step": 4257
    },
    {
      "epoch": 0.5831279101615996,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008905348589159369,
      "loss": 5.065,
      "step": 4258
    },
    {
      "epoch": 0.5832648589427554,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008905264992530714,
      "loss": 5.1425,
      "step": 4259
    },
    {
      "epoch": 0.5834018077239113,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008905181359394579,
      "loss": 5.0735,
      "step": 4260
    },
    {
      "epoch": 0.5835387565050671,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008905097689751655,
      "loss": 5.0973,
      "step": 4261
    },
    {
      "epoch": 0.5836757052862229,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008905013983602639,
      "loss": 5.0989,
      "step": 4262
    },
    {
      "epoch": 0.5838126540673788,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008904930240948221,
      "loss": 5.1674,
      "step": 4263
    },
    {
      "epoch": 0.5839496028485347,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008904846461789096,
      "loss": 5.1171,
      "step": 4264
    },
    {
      "epoch": 0.5840865516296905,
      "grad_norm": 0.486328125,
      "learning_rate": 0.000890476264612596,
      "loss": 5.1063,
      "step": 4265
    },
    {
      "epoch": 0.5842235004108464,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008904678793959506,
      "loss": 5.0911,
      "step": 4266
    },
    {
      "epoch": 0.5843604491920021,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008904594905290429,
      "loss": 5.0845,
      "step": 4267
    },
    {
      "epoch": 0.584497397973158,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008904510980119425,
      "loss": 5.1025,
      "step": 4268
    },
    {
      "epoch": 0.5846343467543139,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008904427018447188,
      "loss": 5.1069,
      "step": 4269
    },
    {
      "epoch": 0.5847712955354697,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008904343020274416,
      "loss": 5.0366,
      "step": 4270
    },
    {
      "epoch": 0.5849082443166256,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008904258985601803,
      "loss": 5.0658,
      "step": 4271
    },
    {
      "epoch": 0.5850451930977815,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008904174914430047,
      "loss": 5.0458,
      "step": 4272
    },
    {
      "epoch": 0.5851821418789372,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008904090806759844,
      "loss": 5.1084,
      "step": 4273
    },
    {
      "epoch": 0.5853190906600931,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008904006662591891,
      "loss": 5.0867,
      "step": 4274
    },
    {
      "epoch": 0.585456039441249,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008903922481926883,
      "loss": 5.087,
      "step": 4275
    },
    {
      "epoch": 0.5855929882224048,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008903838264765524,
      "loss": 5.093,
      "step": 4276
    },
    {
      "epoch": 0.5857299370035607,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008903754011108505,
      "loss": 5.2042,
      "step": 4277
    },
    {
      "epoch": 0.5858668857847165,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008903669720956527,
      "loss": 5.0581,
      "step": 4278
    },
    {
      "epoch": 0.5860038345658724,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008903585394310291,
      "loss": 5.1113,
      "step": 4279
    },
    {
      "epoch": 0.5861407833470282,
      "grad_norm": 0.5,
      "learning_rate": 0.0008903501031170492,
      "loss": 5.1637,
      "step": 4280
    },
    {
      "epoch": 0.586277732128184,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008903416631537831,
      "loss": 5.0506,
      "step": 4281
    },
    {
      "epoch": 0.5864146809093399,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008903332195413007,
      "loss": 5.0899,
      "step": 4282
    },
    {
      "epoch": 0.5865516296904958,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008903247722796719,
      "loss": 5.1009,
      "step": 4283
    },
    {
      "epoch": 0.5866885784716516,
      "grad_norm": 0.421875,
      "learning_rate": 0.000890316321368967,
      "loss": 5.0916,
      "step": 4284
    },
    {
      "epoch": 0.5868255272528075,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008903078668092557,
      "loss": 5.0471,
      "step": 4285
    },
    {
      "epoch": 0.5869624760339633,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008902994086006081,
      "loss": 5.204,
      "step": 4286
    },
    {
      "epoch": 0.5870994248151191,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008902909467430946,
      "loss": 5.1281,
      "step": 4287
    },
    {
      "epoch": 0.587236373596275,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008902824812367848,
      "loss": 5.1179,
      "step": 4288
    },
    {
      "epoch": 0.5873733223774309,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008902740120817495,
      "loss": 5.163,
      "step": 4289
    },
    {
      "epoch": 0.5875102711585867,
      "grad_norm": 0.5,
      "learning_rate": 0.0008902655392780584,
      "loss": 5.0955,
      "step": 4290
    },
    {
      "epoch": 0.5876472199397426,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008902570628257819,
      "loss": 5.1057,
      "step": 4291
    },
    {
      "epoch": 0.5877841687208983,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008902485827249901,
      "loss": 5.1185,
      "step": 4292
    },
    {
      "epoch": 0.5879211175020542,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008902400989757535,
      "loss": 5.0998,
      "step": 4293
    },
    {
      "epoch": 0.5880580662832101,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008902316115781423,
      "loss": 5.0913,
      "step": 4294
    },
    {
      "epoch": 0.5881950150643659,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008902231205322267,
      "loss": 5.1271,
      "step": 4295
    },
    {
      "epoch": 0.5883319638455218,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008902146258380774,
      "loss": 5.1563,
      "step": 4296
    },
    {
      "epoch": 0.5884689126266777,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008902061274957644,
      "loss": 5.1643,
      "step": 4297
    },
    {
      "epoch": 0.5886058614078334,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008901976255053585,
      "loss": 5.0764,
      "step": 4298
    },
    {
      "epoch": 0.5887428101889893,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008901891198669299,
      "loss": 5.1459,
      "step": 4299
    },
    {
      "epoch": 0.5888797589701452,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008901806105805491,
      "loss": 5.1174,
      "step": 4300
    },
    {
      "epoch": 0.589016707751301,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008901720976462868,
      "loss": 5.123,
      "step": 4301
    },
    {
      "epoch": 0.5891536565324569,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008901635810642135,
      "loss": 5.1183,
      "step": 4302
    },
    {
      "epoch": 0.5892906053136127,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008901550608343996,
      "loss": 5.0257,
      "step": 4303
    },
    {
      "epoch": 0.5894275540947685,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008901465369569158,
      "loss": 5.0552,
      "step": 4304
    },
    {
      "epoch": 0.5895645028759244,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008901380094318329,
      "loss": 5.0982,
      "step": 4305
    },
    {
      "epoch": 0.5897014516570802,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008901294782592213,
      "loss": 5.0806,
      "step": 4306
    },
    {
      "epoch": 0.5898384004382361,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008901209434391518,
      "loss": 5.0854,
      "step": 4307
    },
    {
      "epoch": 0.589975349219392,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008901124049716954,
      "loss": 5.0684,
      "step": 4308
    },
    {
      "epoch": 0.5901122980005478,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008901038628569224,
      "loss": 5.1203,
      "step": 4309
    },
    {
      "epoch": 0.5902492467817037,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008900953170949039,
      "loss": 5.094,
      "step": 4310
    },
    {
      "epoch": 0.5903861955628595,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008900867676857106,
      "loss": 5.106,
      "step": 4311
    },
    {
      "epoch": 0.5905231443440153,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008900782146294135,
      "loss": 5.0624,
      "step": 4312
    },
    {
      "epoch": 0.5906600931251712,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008900696579260832,
      "loss": 5.0397,
      "step": 4313
    },
    {
      "epoch": 0.590797041906327,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008900610975757909,
      "loss": 5.1392,
      "step": 4314
    },
    {
      "epoch": 0.5909339906874829,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008900525335786074,
      "loss": 5.0802,
      "step": 4315
    },
    {
      "epoch": 0.5910709394686388,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008900439659346038,
      "loss": 5.1532,
      "step": 4316
    },
    {
      "epoch": 0.5912078882497945,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008900353946438509,
      "loss": 5.1309,
      "step": 4317
    },
    {
      "epoch": 0.5913448370309504,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008900268197064197,
      "loss": 5.1156,
      "step": 4318
    },
    {
      "epoch": 0.5914817858121063,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008900182411223816,
      "loss": 5.0816,
      "step": 4319
    },
    {
      "epoch": 0.5916187345932621,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008900096588918074,
      "loss": 5.1318,
      "step": 4320
    },
    {
      "epoch": 0.591755683374418,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008900010730147683,
      "loss": 5.1614,
      "step": 4321
    },
    {
      "epoch": 0.5918926321555739,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008899924834913355,
      "loss": 5.0674,
      "step": 4322
    },
    {
      "epoch": 0.5920295809367296,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008899838903215801,
      "loss": 5.0834,
      "step": 4323
    },
    {
      "epoch": 0.5921665297178855,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008899752935055734,
      "loss": 5.159,
      "step": 4324
    },
    {
      "epoch": 0.5923034784990414,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008899666930433865,
      "loss": 5.0967,
      "step": 4325
    },
    {
      "epoch": 0.5924404272801972,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008899580889350909,
      "loss": 5.1309,
      "step": 4326
    },
    {
      "epoch": 0.5925773760613531,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008899494811807577,
      "loss": 5.1834,
      "step": 4327
    },
    {
      "epoch": 0.5927143248425089,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008899408697804585,
      "loss": 5.1752,
      "step": 4328
    },
    {
      "epoch": 0.5928512736236647,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008899322547342642,
      "loss": 5.0728,
      "step": 4329
    },
    {
      "epoch": 0.5929882224048206,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008899236360422467,
      "loss": 5.1597,
      "step": 4330
    },
    {
      "epoch": 0.5931251711859764,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008899150137044771,
      "loss": 5.1125,
      "step": 4331
    },
    {
      "epoch": 0.5932621199671323,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008899063877210269,
      "loss": 5.1236,
      "step": 4332
    },
    {
      "epoch": 0.5933990687482882,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008898977580919677,
      "loss": 5.1496,
      "step": 4333
    },
    {
      "epoch": 0.593536017529444,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000889889124817371,
      "loss": 5.1193,
      "step": 4334
    },
    {
      "epoch": 0.5936729663105998,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008898804878973083,
      "loss": 5.1563,
      "step": 4335
    },
    {
      "epoch": 0.5938099150917557,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008898718473318511,
      "loss": 5.1184,
      "step": 4336
    },
    {
      "epoch": 0.5939468638729115,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008898632031210711,
      "loss": 5.0976,
      "step": 4337
    },
    {
      "epoch": 0.5940838126540674,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00088985455526504,
      "loss": 5.0606,
      "step": 4338
    },
    {
      "epoch": 0.5942207614352232,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008898459037638294,
      "loss": 5.0051,
      "step": 4339
    },
    {
      "epoch": 0.5943577102163791,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008898372486175109,
      "loss": 5.0761,
      "step": 4340
    },
    {
      "epoch": 0.594494658997535,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008898285898261564,
      "loss": 5.1327,
      "step": 4341
    },
    {
      "epoch": 0.5946316077786907,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008898199273898375,
      "loss": 5.0844,
      "step": 4342
    },
    {
      "epoch": 0.5947685565598466,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000889811261308626,
      "loss": 5.162,
      "step": 4343
    },
    {
      "epoch": 0.5949055053410025,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008898025915825939,
      "loss": 5.1473,
      "step": 4344
    },
    {
      "epoch": 0.5950424541221583,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008897939182118128,
      "loss": 5.0688,
      "step": 4345
    },
    {
      "epoch": 0.5951794029033142,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008897852411963548,
      "loss": 5.0581,
      "step": 4346
    },
    {
      "epoch": 0.5953163516844701,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008897765605362918,
      "loss": 5.1204,
      "step": 4347
    },
    {
      "epoch": 0.5954533004656258,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008897678762316953,
      "loss": 5.0876,
      "step": 4348
    },
    {
      "epoch": 0.5955902492467817,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008897591882826379,
      "loss": 5.1425,
      "step": 4349
    },
    {
      "epoch": 0.5957271980279375,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008897504966891913,
      "loss": 5.1712,
      "step": 4350
    },
    {
      "epoch": 0.5958641468090934,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008897418014514276,
      "loss": 5.0708,
      "step": 4351
    },
    {
      "epoch": 0.5960010955902493,
      "grad_norm": 0.5,
      "learning_rate": 0.0008897331025694186,
      "loss": 5.088,
      "step": 4352
    },
    {
      "epoch": 0.596138044371405,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008897244000432368,
      "loss": 5.065,
      "step": 4353
    },
    {
      "epoch": 0.5962749931525609,
      "grad_norm": 0.49609375,
      "learning_rate": 0.000889715693872954,
      "loss": 5.0852,
      "step": 4354
    },
    {
      "epoch": 0.5964119419337168,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008897069840586425,
      "loss": 5.0276,
      "step": 4355
    },
    {
      "epoch": 0.5965488907148726,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008896982706003745,
      "loss": 5.1043,
      "step": 4356
    },
    {
      "epoch": 0.5966858394960285,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000889689553498222,
      "loss": 5.0607,
      "step": 4357
    },
    {
      "epoch": 0.5968227882771844,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008896808327522576,
      "loss": 5.0705,
      "step": 4358
    },
    {
      "epoch": 0.5969597370583402,
      "grad_norm": 0.625,
      "learning_rate": 0.0008896721083625532,
      "loss": 5.1598,
      "step": 4359
    },
    {
      "epoch": 0.597096685839496,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008896633803291813,
      "loss": 5.1124,
      "step": 4360
    },
    {
      "epoch": 0.5972336346206518,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008896546486522143,
      "loss": 5.0109,
      "step": 4361
    },
    {
      "epoch": 0.5973705834018077,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008896459133317245,
      "loss": 5.1175,
      "step": 4362
    },
    {
      "epoch": 0.5975075321829636,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008896371743677841,
      "loss": 5.0883,
      "step": 4363
    },
    {
      "epoch": 0.5976444809641194,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008896284317604658,
      "loss": 4.9982,
      "step": 4364
    },
    {
      "epoch": 0.5977814297452753,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008896196855098418,
      "loss": 5.0661,
      "step": 4365
    },
    {
      "epoch": 0.5979183785264311,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008896109356159847,
      "loss": 5.1102,
      "step": 4366
    },
    {
      "epoch": 0.5980553273075869,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008896021820789671,
      "loss": 5.1349,
      "step": 4367
    },
    {
      "epoch": 0.5981922760887428,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008895934248988616,
      "loss": 5.1495,
      "step": 4368
    },
    {
      "epoch": 0.5983292248698987,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008895846640757405,
      "loss": 5.0784,
      "step": 4369
    },
    {
      "epoch": 0.5984661736510545,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008895758996096764,
      "loss": 5.1499,
      "step": 4370
    },
    {
      "epoch": 0.5986031224322104,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008895671315007422,
      "loss": 5.0872,
      "step": 4371
    },
    {
      "epoch": 0.5987400712133663,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008895583597490105,
      "loss": 5.1322,
      "step": 4372
    },
    {
      "epoch": 0.598877019994522,
      "grad_norm": 0.4609375,
      "learning_rate": 0.000889549584354554,
      "loss": 5.1195,
      "step": 4373
    },
    {
      "epoch": 0.5990139687756779,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008895408053174451,
      "loss": 5.0694,
      "step": 4374
    },
    {
      "epoch": 0.5991509175568337,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000889532022637757,
      "loss": 5.1746,
      "step": 4375
    },
    {
      "epoch": 0.5992878663379896,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008895232363155623,
      "loss": 5.0707,
      "step": 4376
    },
    {
      "epoch": 0.5994248151191455,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008895144463509337,
      "loss": 5.0547,
      "step": 4377
    },
    {
      "epoch": 0.5995617639003012,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008895056527439443,
      "loss": 5.0567,
      "step": 4378
    },
    {
      "epoch": 0.5996987126814571,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008894968554946666,
      "loss": 5.0349,
      "step": 4379
    },
    {
      "epoch": 0.599835661462613,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008894880546031739,
      "loss": 5.1255,
      "step": 4380
    },
    {
      "epoch": 0.5999726102437688,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000889479250069539,
      "loss": 5.1479,
      "step": 4381
    },
    {
      "epoch": 0.6001095590249247,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008894704418938347,
      "loss": 5.1308,
      "step": 4382
    },
    {
      "epoch": 0.6002465078060806,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008894616300761342,
      "loss": 5.0918,
      "step": 4383
    },
    {
      "epoch": 0.6003834565872364,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008894528146165104,
      "loss": 5.0683,
      "step": 4384
    },
    {
      "epoch": 0.6005204053683922,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008894439955150366,
      "loss": 5.1075,
      "step": 4385
    },
    {
      "epoch": 0.600657354149548,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008894351727717855,
      "loss": 5.0654,
      "step": 4386
    },
    {
      "epoch": 0.6007943029307039,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008894263463868305,
      "loss": 5.1037,
      "step": 4387
    },
    {
      "epoch": 0.6009312517118598,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008894175163602446,
      "loss": 5.0746,
      "step": 4388
    },
    {
      "epoch": 0.6010682004930156,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008894086826921009,
      "loss": 5.1179,
      "step": 4389
    },
    {
      "epoch": 0.6012051492741715,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008893998453824731,
      "loss": 5.1509,
      "step": 4390
    },
    {
      "epoch": 0.6013420980553273,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008893910044314339,
      "loss": 5.1092,
      "step": 4391
    },
    {
      "epoch": 0.6014790468364831,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008893821598390567,
      "loss": 5.1515,
      "step": 4392
    },
    {
      "epoch": 0.601615995617639,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008893733116054148,
      "loss": 5.1399,
      "step": 4393
    },
    {
      "epoch": 0.6017529443987949,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008893644597305816,
      "loss": 5.1027,
      "step": 4394
    },
    {
      "epoch": 0.6018898931799507,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008893556042146305,
      "loss": 5.1018,
      "step": 4395
    },
    {
      "epoch": 0.6020268419611066,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008893467450576347,
      "loss": 5.1241,
      "step": 4396
    },
    {
      "epoch": 0.6021637907422623,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008893378822596679,
      "loss": 5.1569,
      "step": 4397
    },
    {
      "epoch": 0.6023007395234182,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008893290158208032,
      "loss": 5.1491,
      "step": 4398
    },
    {
      "epoch": 0.6024376883045741,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008893201457411145,
      "loss": 5.162,
      "step": 4399
    },
    {
      "epoch": 0.6025746370857299,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008893112720206749,
      "loss": 5.1372,
      "step": 4400
    },
    {
      "epoch": 0.6027115858668858,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008893023946595581,
      "loss": 5.0932,
      "step": 4401
    },
    {
      "epoch": 0.6028485346480417,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008892935136578376,
      "loss": 5.1039,
      "step": 4402
    },
    {
      "epoch": 0.6029854834291974,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008892846290155872,
      "loss": 5.0512,
      "step": 4403
    },
    {
      "epoch": 0.6031224322103533,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008892757407328804,
      "loss": 5.0744,
      "step": 4404
    },
    {
      "epoch": 0.6032593809915092,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008892668488097909,
      "loss": 5.0425,
      "step": 4405
    },
    {
      "epoch": 0.603396329772665,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008892579532463922,
      "loss": 5.08,
      "step": 4406
    },
    {
      "epoch": 0.6035332785538209,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008892490540427584,
      "loss": 5.0695,
      "step": 4407
    },
    {
      "epoch": 0.6036702273349768,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008892401511989628,
      "loss": 5.1283,
      "step": 4408
    },
    {
      "epoch": 0.6038071761161325,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008892312447150795,
      "loss": 5.0693,
      "step": 4409
    },
    {
      "epoch": 0.6039441248972884,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008892223345911822,
      "loss": 5.1139,
      "step": 4410
    },
    {
      "epoch": 0.6040810736784442,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008892134208273447,
      "loss": 5.0993,
      "step": 4411
    },
    {
      "epoch": 0.6042180224596001,
      "grad_norm": 0.390625,
      "learning_rate": 0.000889204503423641,
      "loss": 5.1326,
      "step": 4412
    },
    {
      "epoch": 0.604354971240756,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008891955823801448,
      "loss": 5.1188,
      "step": 4413
    },
    {
      "epoch": 0.6044919200219118,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008891866576969304,
      "loss": 5.0048,
      "step": 4414
    },
    {
      "epoch": 0.6046288688030677,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008891777293740713,
      "loss": 5.1699,
      "step": 4415
    },
    {
      "epoch": 0.6047658175842235,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008891687974116418,
      "loss": 5.1028,
      "step": 4416
    },
    {
      "epoch": 0.6049027663653793,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008891598618097158,
      "loss": 5.128,
      "step": 4417
    },
    {
      "epoch": 0.6050397151465352,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008891509225683673,
      "loss": 5.1223,
      "step": 4418
    },
    {
      "epoch": 0.6051766639276911,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008891419796876706,
      "loss": 5.0579,
      "step": 4419
    },
    {
      "epoch": 0.6053136127088469,
      "grad_norm": 0.375,
      "learning_rate": 0.0008891330331676997,
      "loss": 5.0328,
      "step": 4420
    },
    {
      "epoch": 0.6054505614900028,
      "grad_norm": 0.375,
      "learning_rate": 0.0008891240830085285,
      "loss": 5.0893,
      "step": 4421
    },
    {
      "epoch": 0.6055875102711585,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008891151292102315,
      "loss": 5.0872,
      "step": 4422
    },
    {
      "epoch": 0.6057244590523144,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008891061717728829,
      "loss": 5.0212,
      "step": 4423
    },
    {
      "epoch": 0.6058614078334703,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008890972106965567,
      "loss": 5.1019,
      "step": 4424
    },
    {
      "epoch": 0.6059983566146261,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008890882459813273,
      "loss": 5.1716,
      "step": 4425
    },
    {
      "epoch": 0.606135305395782,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008890792776272689,
      "loss": 5.1195,
      "step": 4426
    },
    {
      "epoch": 0.6062722541769379,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008890703056344561,
      "loss": 5.1572,
      "step": 4427
    },
    {
      "epoch": 0.6064092029580936,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008890613300029629,
      "loss": 5.065,
      "step": 4428
    },
    {
      "epoch": 0.6065461517392495,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000889052350732864,
      "loss": 5.1056,
      "step": 4429
    },
    {
      "epoch": 0.6066831005204054,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008890433678242334,
      "loss": 5.0726,
      "step": 4430
    },
    {
      "epoch": 0.6068200493015612,
      "grad_norm": 0.578125,
      "learning_rate": 0.000889034381277146,
      "loss": 5.178,
      "step": 4431
    },
    {
      "epoch": 0.6069569980827171,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008890253910916759,
      "loss": 5.0745,
      "step": 4432
    },
    {
      "epoch": 0.6070939468638729,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008890163972678978,
      "loss": 5.1397,
      "step": 4433
    },
    {
      "epoch": 0.6072308956450287,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008890073998058863,
      "loss": 5.1427,
      "step": 4434
    },
    {
      "epoch": 0.6073678444261846,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008889983987057158,
      "loss": 5.0513,
      "step": 4435
    },
    {
      "epoch": 0.6075047932073404,
      "grad_norm": 0.5,
      "learning_rate": 0.000888989393967461,
      "loss": 5.0891,
      "step": 4436
    },
    {
      "epoch": 0.6076417419884963,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008889803855911965,
      "loss": 5.0994,
      "step": 4437
    },
    {
      "epoch": 0.6077786907696522,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008889713735769968,
      "loss": 5.1571,
      "step": 4438
    },
    {
      "epoch": 0.607915639550808,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008889623579249368,
      "loss": 5.0829,
      "step": 4439
    },
    {
      "epoch": 0.6080525883319639,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008889533386350913,
      "loss": 5.1546,
      "step": 4440
    },
    {
      "epoch": 0.6081895371131197,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008889443157075348,
      "loss": 5.1472,
      "step": 4441
    },
    {
      "epoch": 0.6083264858942755,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008889352891423421,
      "loss": 5.143,
      "step": 4442
    },
    {
      "epoch": 0.6084634346754314,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008889262589395881,
      "loss": 5.0756,
      "step": 4443
    },
    {
      "epoch": 0.6086003834565873,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008889172250993476,
      "loss": 5.0483,
      "step": 4444
    },
    {
      "epoch": 0.6087373322377431,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008889081876216954,
      "loss": 5.0386,
      "step": 4445
    },
    {
      "epoch": 0.608874281018899,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008888991465067066,
      "loss": 5.0614,
      "step": 4446
    },
    {
      "epoch": 0.6090112298000547,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008888901017544558,
      "loss": 5.1369,
      "step": 4447
    },
    {
      "epoch": 0.6091481785812106,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008888810533650182,
      "loss": 5.061,
      "step": 4448
    },
    {
      "epoch": 0.6092851273623665,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008888720013384689,
      "loss": 5.0208,
      "step": 4449
    },
    {
      "epoch": 0.6094220761435223,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008888629456748826,
      "loss": 5.1276,
      "step": 4450
    },
    {
      "epoch": 0.6095590249246782,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008888538863743346,
      "loss": 5.0927,
      "step": 4451
    },
    {
      "epoch": 0.6096959737058341,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008888448234368998,
      "loss": 5.0413,
      "step": 4452
    },
    {
      "epoch": 0.6098329224869898,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008888357568626533,
      "loss": 5.0617,
      "step": 4453
    },
    {
      "epoch": 0.6099698712681457,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008888266866516704,
      "loss": 5.0756,
      "step": 4454
    },
    {
      "epoch": 0.6101068200493016,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008888176128040261,
      "loss": 5.0834,
      "step": 4455
    },
    {
      "epoch": 0.6102437688304574,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008888085353197959,
      "loss": 5.1058,
      "step": 4456
    },
    {
      "epoch": 0.6103807176116133,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008887994541990547,
      "loss": 5.1178,
      "step": 4457
    },
    {
      "epoch": 0.610517666392769,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008887903694418778,
      "loss": 5.0992,
      "step": 4458
    },
    {
      "epoch": 0.6106546151739249,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008887812810483405,
      "loss": 5.1096,
      "step": 4459
    },
    {
      "epoch": 0.6107915639550808,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008887721890185182,
      "loss": 5.0924,
      "step": 4460
    },
    {
      "epoch": 0.6109285127362366,
      "grad_norm": 0.5,
      "learning_rate": 0.0008887630933524862,
      "loss": 5.1362,
      "step": 4461
    },
    {
      "epoch": 0.6110654615173925,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008887539940503201,
      "loss": 5.1398,
      "step": 4462
    },
    {
      "epoch": 0.6112024102985484,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008887448911120949,
      "loss": 5.145,
      "step": 4463
    },
    {
      "epoch": 0.6113393590797042,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008887357845378862,
      "loss": 5.0765,
      "step": 4464
    },
    {
      "epoch": 0.61147630786086,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008887266743277697,
      "loss": 5.0931,
      "step": 4465
    },
    {
      "epoch": 0.6116132566420159,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008887175604818206,
      "loss": 5.106,
      "step": 4466
    },
    {
      "epoch": 0.6117502054231717,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008887084430001146,
      "loss": 5.0424,
      "step": 4467
    },
    {
      "epoch": 0.6118871542043276,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000888699321882727,
      "loss": 5.1166,
      "step": 4468
    },
    {
      "epoch": 0.6120241029854834,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008886901971297338,
      "loss": 5.1042,
      "step": 4469
    },
    {
      "epoch": 0.6121610517666393,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008886810687412103,
      "loss": 5.1483,
      "step": 4470
    },
    {
      "epoch": 0.6122980005477952,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008886719367172323,
      "loss": 5.0971,
      "step": 4471
    },
    {
      "epoch": 0.6124349493289509,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008886628010578753,
      "loss": 5.0811,
      "step": 4472
    },
    {
      "epoch": 0.6125718981101068,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008886536617632153,
      "loss": 5.1405,
      "step": 4473
    },
    {
      "epoch": 0.6127088468912627,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008886445188333278,
      "loss": 5.0256,
      "step": 4474
    },
    {
      "epoch": 0.6128457956724185,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008886353722682887,
      "loss": 5.0972,
      "step": 4475
    },
    {
      "epoch": 0.6129827444535744,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008886262220681737,
      "loss": 5.1083,
      "step": 4476
    },
    {
      "epoch": 0.6131196932347303,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008886170682330587,
      "loss": 5.1318,
      "step": 4477
    },
    {
      "epoch": 0.613256642015886,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008886079107630196,
      "loss": 5.1016,
      "step": 4478
    },
    {
      "epoch": 0.6133935907970419,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008885987496581322,
      "loss": 5.0435,
      "step": 4479
    },
    {
      "epoch": 0.6135305395781978,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008885895849184724,
      "loss": 5.0657,
      "step": 4480
    },
    {
      "epoch": 0.6136674883593536,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008885804165441161,
      "loss": 5.1281,
      "step": 4481
    },
    {
      "epoch": 0.6138044371405095,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008885712445351395,
      "loss": 5.0774,
      "step": 4482
    },
    {
      "epoch": 0.6139413859216653,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008885620688916186,
      "loss": 5.048,
      "step": 4483
    },
    {
      "epoch": 0.6140783347028211,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008885528896136292,
      "loss": 5.0758,
      "step": 4484
    },
    {
      "epoch": 0.614215283483977,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008885437067012474,
      "loss": 5.0933,
      "step": 4485
    },
    {
      "epoch": 0.6143522322651328,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008885345201545496,
      "loss": 5.0674,
      "step": 4486
    },
    {
      "epoch": 0.6144891810462887,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008885253299736116,
      "loss": 5.0701,
      "step": 4487
    },
    {
      "epoch": 0.6146261298274446,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008885161361585098,
      "loss": 5.0245,
      "step": 4488
    },
    {
      "epoch": 0.6147630786086004,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008885069387093203,
      "loss": 5.1166,
      "step": 4489
    },
    {
      "epoch": 0.6149000273897562,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008884977376261192,
      "loss": 4.988,
      "step": 4490
    },
    {
      "epoch": 0.6150369761709121,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008884885329089829,
      "loss": 5.1074,
      "step": 4491
    },
    {
      "epoch": 0.6151739249520679,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008884793245579878,
      "loss": 5.0804,
      "step": 4492
    },
    {
      "epoch": 0.6153108737332238,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008884701125732097,
      "loss": 5.0101,
      "step": 4493
    },
    {
      "epoch": 0.6154478225143796,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008884608969547257,
      "loss": 5.0611,
      "step": 4494
    },
    {
      "epoch": 0.6155847712955355,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008884516777026114,
      "loss": 5.1349,
      "step": 4495
    },
    {
      "epoch": 0.6157217200766913,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008884424548169437,
      "loss": 5.0861,
      "step": 4496
    },
    {
      "epoch": 0.6158586688578471,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000888433228297799,
      "loss": 5.0818,
      "step": 4497
    },
    {
      "epoch": 0.615995617639003,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008884239981452536,
      "loss": 5.1397,
      "step": 4498
    },
    {
      "epoch": 0.6161325664201589,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000888414764359384,
      "loss": 5.0955,
      "step": 4499
    },
    {
      "epoch": 0.6162695152013147,
      "grad_norm": 0.482421875,
      "learning_rate": 0.000888405526940267,
      "loss": 5.1161,
      "step": 4500
    },
    {
      "epoch": 0.6164064639824706,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008883962858879787,
      "loss": 5.0783,
      "step": 4501
    },
    {
      "epoch": 0.6165434127636265,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008883870412025959,
      "loss": 5.1971,
      "step": 4502
    },
    {
      "epoch": 0.6166803615447822,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008883777928841953,
      "loss": 5.1695,
      "step": 4503
    },
    {
      "epoch": 0.6168173103259381,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008883685409328536,
      "loss": 5.0627,
      "step": 4504
    },
    {
      "epoch": 0.6169542591070939,
      "grad_norm": 0.375,
      "learning_rate": 0.0008883592853486472,
      "loss": 5.0119,
      "step": 4505
    },
    {
      "epoch": 0.6170912078882498,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008883500261316529,
      "loss": 5.0865,
      "step": 4506
    },
    {
      "epoch": 0.6172281566694057,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008883407632819476,
      "loss": 5.1135,
      "step": 4507
    },
    {
      "epoch": 0.6173651054505614,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008883314967996079,
      "loss": 5.0954,
      "step": 4508
    },
    {
      "epoch": 0.6175020542317173,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008883222266847106,
      "loss": 5.1555,
      "step": 4509
    },
    {
      "epoch": 0.6176390030128732,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008883129529373326,
      "loss": 5.0011,
      "step": 4510
    },
    {
      "epoch": 0.617775951794029,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008883036755575508,
      "loss": 4.9475,
      "step": 4511
    },
    {
      "epoch": 0.6179129005751849,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008882943945454419,
      "loss": 5.0786,
      "step": 4512
    },
    {
      "epoch": 0.6180498493563408,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008882851099010828,
      "loss": 5.1721,
      "step": 4513
    },
    {
      "epoch": 0.6181867981374966,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008882758216245507,
      "loss": 5.1932,
      "step": 4514
    },
    {
      "epoch": 0.6183237469186524,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008882665297159225,
      "loss": 5.0303,
      "step": 4515
    },
    {
      "epoch": 0.6184606956998083,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008882572341752749,
      "loss": 5.079,
      "step": 4516
    },
    {
      "epoch": 0.6185976444809641,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008882479350026854,
      "loss": 5.0879,
      "step": 4517
    },
    {
      "epoch": 0.61873459326212,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008882386321982308,
      "loss": 5.1797,
      "step": 4518
    },
    {
      "epoch": 0.6188715420432758,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008882293257619882,
      "loss": 5.0874,
      "step": 4519
    },
    {
      "epoch": 0.6190084908244317,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008882200156940346,
      "loss": 5.1549,
      "step": 4520
    },
    {
      "epoch": 0.6191454396055875,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008882107019944474,
      "loss": 5.0238,
      "step": 4521
    },
    {
      "epoch": 0.6192823883867433,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008882013846633038,
      "loss": 5.1124,
      "step": 4522
    },
    {
      "epoch": 0.6194193371678992,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008881920637006809,
      "loss": 4.9867,
      "step": 4523
    },
    {
      "epoch": 0.6195562859490551,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008881827391066558,
      "loss": 5.0869,
      "step": 4524
    },
    {
      "epoch": 0.6196932347302109,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000888173410881306,
      "loss": 5.0731,
      "step": 4525
    },
    {
      "epoch": 0.6198301835113668,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008881640790247087,
      "loss": 5.0833,
      "step": 4526
    },
    {
      "epoch": 0.6199671322925226,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008881547435369414,
      "loss": 5.1405,
      "step": 4527
    },
    {
      "epoch": 0.6201040810736784,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008881454044180813,
      "loss": 5.1456,
      "step": 4528
    },
    {
      "epoch": 0.6202410298548343,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008881360616682057,
      "loss": 5.0085,
      "step": 4529
    },
    {
      "epoch": 0.6203779786359901,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008881267152873923,
      "loss": 5.037,
      "step": 4530
    },
    {
      "epoch": 0.620514927417146,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008881173652757183,
      "loss": 5.0193,
      "step": 4531
    },
    {
      "epoch": 0.6206518761983019,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008881080116332614,
      "loss": 5.0376,
      "step": 4532
    },
    {
      "epoch": 0.6207888249794576,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008880986543600989,
      "loss": 5.028,
      "step": 4533
    },
    {
      "epoch": 0.6209257737606135,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008880892934563085,
      "loss": 5.0847,
      "step": 4534
    },
    {
      "epoch": 0.6210627225417694,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008880799289219677,
      "loss": 5.1254,
      "step": 4535
    },
    {
      "epoch": 0.6211996713229252,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008880705607571543,
      "loss": 5.1582,
      "step": 4536
    },
    {
      "epoch": 0.6213366201040811,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008880611889619456,
      "loss": 5.1103,
      "step": 4537
    },
    {
      "epoch": 0.621473568885237,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008880518135364196,
      "loss": 5.0836,
      "step": 4538
    },
    {
      "epoch": 0.6216105176663927,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008880424344806537,
      "loss": 5.0732,
      "step": 4539
    },
    {
      "epoch": 0.6217474664475486,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008880330517947258,
      "loss": 5.093,
      "step": 4540
    },
    {
      "epoch": 0.6218844152287044,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008880236654787136,
      "loss": 5.1179,
      "step": 4541
    },
    {
      "epoch": 0.6220213640098603,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008880142755326951,
      "loss": 5.1527,
      "step": 4542
    },
    {
      "epoch": 0.6221583127910162,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008880048819567476,
      "loss": 5.1245,
      "step": 4543
    },
    {
      "epoch": 0.622295261572172,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008879954847509495,
      "loss": 5.1476,
      "step": 4544
    },
    {
      "epoch": 0.6224322103533279,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008879860839153784,
      "loss": 5.0654,
      "step": 4545
    },
    {
      "epoch": 0.6225691591344837,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008879766794501122,
      "loss": 5.113,
      "step": 4546
    },
    {
      "epoch": 0.6227061079156395,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000887967271355229,
      "loss": 5.1142,
      "step": 4547
    },
    {
      "epoch": 0.6228430566967954,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008879578596308065,
      "loss": 5.0584,
      "step": 4548
    },
    {
      "epoch": 0.6229800054779513,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000887948444276923,
      "loss": 5.0661,
      "step": 4549
    },
    {
      "epoch": 0.6231169542591071,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008879390252936564,
      "loss": 5.1507,
      "step": 4550
    },
    {
      "epoch": 0.623253903040263,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008879296026810847,
      "loss": 5.0936,
      "step": 4551
    },
    {
      "epoch": 0.6233908518214187,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000887920176439286,
      "loss": 5.177,
      "step": 4552
    },
    {
      "epoch": 0.6235278006025746,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008879107465683385,
      "loss": 5.1334,
      "step": 4553
    },
    {
      "epoch": 0.6236647493837305,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008879013130683202,
      "loss": 5.1073,
      "step": 4554
    },
    {
      "epoch": 0.6238016981648863,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008878918759393095,
      "loss": 5.1219,
      "step": 4555
    },
    {
      "epoch": 0.6239386469460422,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008878824351813843,
      "loss": 5.1191,
      "step": 4556
    },
    {
      "epoch": 0.6240755957271981,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008878729907946232,
      "loss": 5.1356,
      "step": 4557
    },
    {
      "epoch": 0.6242125445083538,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008878635427791041,
      "loss": 5.0879,
      "step": 4558
    },
    {
      "epoch": 0.6243494932895097,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008878540911349057,
      "loss": 5.0933,
      "step": 4559
    },
    {
      "epoch": 0.6244864420706656,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000887844635862106,
      "loss": 5.12,
      "step": 4560
    },
    {
      "epoch": 0.6246233908518214,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008878351769607835,
      "loss": 5.1148,
      "step": 4561
    },
    {
      "epoch": 0.6247603396329773,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008878257144310165,
      "loss": 5.067,
      "step": 4562
    },
    {
      "epoch": 0.6248972884141332,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008878162482728835,
      "loss": 5.1767,
      "step": 4563
    },
    {
      "epoch": 0.6250342371952889,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008878067784864629,
      "loss": 5.1299,
      "step": 4564
    },
    {
      "epoch": 0.6251711859764448,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008877973050718333,
      "loss": 5.1224,
      "step": 4565
    },
    {
      "epoch": 0.6253081347576006,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008877878280290731,
      "loss": 5.1413,
      "step": 4566
    },
    {
      "epoch": 0.6254450835387565,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008877783473582607,
      "loss": 5.1166,
      "step": 4567
    },
    {
      "epoch": 0.6255820323199124,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008877688630594748,
      "loss": 5.0993,
      "step": 4568
    },
    {
      "epoch": 0.6257189811010682,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008877593751327942,
      "loss": 5.0942,
      "step": 4569
    },
    {
      "epoch": 0.625855929882224,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008877498835782974,
      "loss": 5.1614,
      "step": 4570
    },
    {
      "epoch": 0.6259928786633799,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008877403883960629,
      "loss": 5.1522,
      "step": 4571
    },
    {
      "epoch": 0.6261298274445357,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008877308895861695,
      "loss": 5.1183,
      "step": 4572
    },
    {
      "epoch": 0.6262667762256916,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008877213871486958,
      "loss": 5.0375,
      "step": 4573
    },
    {
      "epoch": 0.6264037250068475,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008877118810837208,
      "loss": 5.0657,
      "step": 4574
    },
    {
      "epoch": 0.6265406737880033,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008877023713913231,
      "loss": 5.1291,
      "step": 4575
    },
    {
      "epoch": 0.6266776225691592,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008876928580715815,
      "loss": 5.1007,
      "step": 4576
    },
    {
      "epoch": 0.6268145713503149,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008876833411245749,
      "loss": 5.1239,
      "step": 4577
    },
    {
      "epoch": 0.6269515201314708,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008876738205503822,
      "loss": 5.0852,
      "step": 4578
    },
    {
      "epoch": 0.6270884689126267,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008876642963490822,
      "loss": 5.0495,
      "step": 4579
    },
    {
      "epoch": 0.6272254176937825,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008876547685207541,
      "loss": 4.9992,
      "step": 4580
    },
    {
      "epoch": 0.6273623664749384,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008876452370654765,
      "loss": 5.0737,
      "step": 4581
    },
    {
      "epoch": 0.6274993152560943,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008876357019833284,
      "loss": 5.0325,
      "step": 4582
    },
    {
      "epoch": 0.62763626403725,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008876261632743891,
      "loss": 5.0221,
      "step": 4583
    },
    {
      "epoch": 0.6277732128184059,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008876166209387374,
      "loss": 5.1352,
      "step": 4584
    },
    {
      "epoch": 0.6279101615995618,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008876070749764526,
      "loss": 5.0329,
      "step": 4585
    },
    {
      "epoch": 0.6280471103807176,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008875975253876137,
      "loss": 5.0815,
      "step": 4586
    },
    {
      "epoch": 0.6281840591618735,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008875879721722997,
      "loss": 5.1341,
      "step": 4587
    },
    {
      "epoch": 0.6283210079430293,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008875784153305901,
      "loss": 5.0717,
      "step": 4588
    },
    {
      "epoch": 0.6284579567241851,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008875688548625637,
      "loss": 5.1079,
      "step": 4589
    },
    {
      "epoch": 0.628594905505341,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008875592907683001,
      "loss": 5.0701,
      "step": 4590
    },
    {
      "epoch": 0.6287318542864968,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008875497230478784,
      "loss": 5.0866,
      "step": 4591
    },
    {
      "epoch": 0.6288688030676527,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008875401517013776,
      "loss": 5.1351,
      "step": 4592
    },
    {
      "epoch": 0.6290057518488086,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008875305767288774,
      "loss": 5.0019,
      "step": 4593
    },
    {
      "epoch": 0.6291427006299644,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008875209981304572,
      "loss": 5.0949,
      "step": 4594
    },
    {
      "epoch": 0.6292796494111202,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008875114159061963,
      "loss": 5.0866,
      "step": 4595
    },
    {
      "epoch": 0.6294165981922761,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008875018300561739,
      "loss": 5.1521,
      "step": 4596
    },
    {
      "epoch": 0.6295535469734319,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008874922405804696,
      "loss": 5.0576,
      "step": 4597
    },
    {
      "epoch": 0.6296904957545878,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008874826474791629,
      "loss": 5.061,
      "step": 4598
    },
    {
      "epoch": 0.6298274445357437,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008874730507523332,
      "loss": 5.0983,
      "step": 4599
    },
    {
      "epoch": 0.6299643933168995,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008874634504000601,
      "loss": 5.0063,
      "step": 4600
    },
    {
      "epoch": 0.6301013420980554,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008874538464224232,
      "loss": 5.1318,
      "step": 4601
    },
    {
      "epoch": 0.6302382908792111,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008874442388195021,
      "loss": 5.0337,
      "step": 4602
    },
    {
      "epoch": 0.630375239660367,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008874346275913761,
      "loss": 5.0993,
      "step": 4603
    },
    {
      "epoch": 0.6305121884415229,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008874250127381253,
      "loss": 5.125,
      "step": 4604
    },
    {
      "epoch": 0.6306491372226787,
      "grad_norm": 0.498046875,
      "learning_rate": 0.000887415394259829,
      "loss": 5.0937,
      "step": 4605
    },
    {
      "epoch": 0.6307860860038346,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008874057721565673,
      "loss": 5.0163,
      "step": 4606
    },
    {
      "epoch": 0.6309230347849905,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008873961464284198,
      "loss": 5.019,
      "step": 4607
    },
    {
      "epoch": 0.6310599835661462,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008873865170754659,
      "loss": 5.0911,
      "step": 4608
    },
    {
      "epoch": 0.6311969323473021,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000887376884097786,
      "loss": 5.1036,
      "step": 4609
    },
    {
      "epoch": 0.631333881128458,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008873672474954595,
      "loss": 5.0181,
      "step": 4610
    },
    {
      "epoch": 0.6314708299096138,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008873576072685665,
      "loss": 5.0426,
      "step": 4611
    },
    {
      "epoch": 0.6316077786907697,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008873479634171867,
      "loss": 5.1217,
      "step": 4612
    },
    {
      "epoch": 0.6317447274719254,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008873383159414,
      "loss": 5.0711,
      "step": 4613
    },
    {
      "epoch": 0.6318816762530813,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008873286648412867,
      "loss": 5.0248,
      "step": 4614
    },
    {
      "epoch": 0.6320186250342372,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008873190101169263,
      "loss": 5.0276,
      "step": 4615
    },
    {
      "epoch": 0.632155573815393,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008873093517683993,
      "loss": 5.0919,
      "step": 4616
    },
    {
      "epoch": 0.6322925225965489,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008872996897957854,
      "loss": 5.0755,
      "step": 4617
    },
    {
      "epoch": 0.6324294713777048,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008872900241991647,
      "loss": 5.0802,
      "step": 4618
    },
    {
      "epoch": 0.6325664201588606,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008872803549786175,
      "loss": 5.1619,
      "step": 4619
    },
    {
      "epoch": 0.6327033689400164,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008872706821342237,
      "loss": 5.1064,
      "step": 4620
    },
    {
      "epoch": 0.6328403177211723,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008872610056660636,
      "loss": 5.067,
      "step": 4621
    },
    {
      "epoch": 0.6329772665023281,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008872513255742174,
      "loss": 5.0807,
      "step": 4622
    },
    {
      "epoch": 0.633114215283484,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008872416418587651,
      "loss": 5.0663,
      "step": 4623
    },
    {
      "epoch": 0.6332511640646398,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008872319545197873,
      "loss": 5.1215,
      "step": 4624
    },
    {
      "epoch": 0.6333881128457957,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008872222635573641,
      "loss": 5.0355,
      "step": 4625
    },
    {
      "epoch": 0.6335250616269515,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008872125689715758,
      "loss": 5.1098,
      "step": 4626
    },
    {
      "epoch": 0.6336620104081073,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008872028707625027,
      "loss": 5.1125,
      "step": 4627
    },
    {
      "epoch": 0.6337989591892632,
      "grad_norm": 0.375,
      "learning_rate": 0.0008871931689302253,
      "loss": 5.0837,
      "step": 4628
    },
    {
      "epoch": 0.6339359079704191,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008871834634748239,
      "loss": 5.0638,
      "step": 4629
    },
    {
      "epoch": 0.6340728567515749,
      "grad_norm": 0.5,
      "learning_rate": 0.0008871737543963788,
      "loss": 5.058,
      "step": 4630
    },
    {
      "epoch": 0.6342098055327308,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008871640416949709,
      "loss": 5.0747,
      "step": 4631
    },
    {
      "epoch": 0.6343467543138867,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008871543253706802,
      "loss": 5.0804,
      "step": 4632
    },
    {
      "epoch": 0.6344837030950424,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008871446054235875,
      "loss": 5.0722,
      "step": 4633
    },
    {
      "epoch": 0.6346206518761983,
      "grad_norm": 0.5,
      "learning_rate": 0.0008871348818537734,
      "loss": 5.0273,
      "step": 4634
    },
    {
      "epoch": 0.6347576006573542,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008871251546613184,
      "loss": 5.0378,
      "step": 4635
    },
    {
      "epoch": 0.63489454943851,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008871154238463027,
      "loss": 5.0717,
      "step": 4636
    },
    {
      "epoch": 0.6350314982196659,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008871056894088077,
      "loss": 5.0583,
      "step": 4637
    },
    {
      "epoch": 0.6351684470008216,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008870959513489136,
      "loss": 5.1405,
      "step": 4638
    },
    {
      "epoch": 0.6353053957819775,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008870862096667012,
      "loss": 5.0884,
      "step": 4639
    },
    {
      "epoch": 0.6354423445631334,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008870764643622512,
      "loss": 5.0428,
      "step": 4640
    },
    {
      "epoch": 0.6355792933442892,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008870667154356444,
      "loss": 5.0824,
      "step": 4641
    },
    {
      "epoch": 0.6357162421254451,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008870569628869616,
      "loss": 5.1161,
      "step": 4642
    },
    {
      "epoch": 0.635853190906601,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008870472067162835,
      "loss": 5.1574,
      "step": 4643
    },
    {
      "epoch": 0.6359901396877568,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008870374469236912,
      "loss": 5.0868,
      "step": 4644
    },
    {
      "epoch": 0.6361270884689126,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008870276835092654,
      "loss": 5.1914,
      "step": 4645
    },
    {
      "epoch": 0.6362640372500685,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008870179164730871,
      "loss": 5.0778,
      "step": 4646
    },
    {
      "epoch": 0.6364009860312243,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008870081458152371,
      "loss": 5.0433,
      "step": 4647
    },
    {
      "epoch": 0.6365379348123802,
      "grad_norm": 0.375,
      "learning_rate": 0.0008869983715357964,
      "loss": 5.1142,
      "step": 4648
    },
    {
      "epoch": 0.636674883593536,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008869885936348462,
      "loss": 5.0418,
      "step": 4649
    },
    {
      "epoch": 0.6368118323746919,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008869788121124673,
      "loss": 5.1128,
      "step": 4650
    },
    {
      "epoch": 0.6369487811558477,
      "grad_norm": 0.47265625,
      "learning_rate": 0.000886969026968741,
      "loss": 5.0523,
      "step": 4651
    },
    {
      "epoch": 0.6370857299370035,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008869592382037481,
      "loss": 5.1625,
      "step": 4652
    },
    {
      "epoch": 0.6372226787181594,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008869494458175699,
      "loss": 5.0951,
      "step": 4653
    },
    {
      "epoch": 0.6373596274993153,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008869396498102875,
      "loss": 4.9751,
      "step": 4654
    },
    {
      "epoch": 0.6374965762804711,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008869298501819822,
      "loss": 5.106,
      "step": 4655
    },
    {
      "epoch": 0.637633525061627,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008869200469327351,
      "loss": 5.0165,
      "step": 4656
    },
    {
      "epoch": 0.6377704738427828,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008869102400626274,
      "loss": 5.0143,
      "step": 4657
    },
    {
      "epoch": 0.6379074226239386,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008869004295717404,
      "loss": 5.0656,
      "step": 4658
    },
    {
      "epoch": 0.6380443714050945,
      "grad_norm": 0.5,
      "learning_rate": 0.0008868906154601556,
      "loss": 5.054,
      "step": 4659
    },
    {
      "epoch": 0.6381813201862503,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000886880797727954,
      "loss": 5.1065,
      "step": 4660
    },
    {
      "epoch": 0.6383182689674062,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008868709763752172,
      "loss": 5.1108,
      "step": 4661
    },
    {
      "epoch": 0.6384552177485621,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008868611514020265,
      "loss": 5.0786,
      "step": 4662
    },
    {
      "epoch": 0.6385921665297178,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008868513228084634,
      "loss": 5.0536,
      "step": 4663
    },
    {
      "epoch": 0.6387291153108737,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008868414905946092,
      "loss": 5.1545,
      "step": 4664
    },
    {
      "epoch": 0.6388660640920296,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008868316547605454,
      "loss": 5.0512,
      "step": 4665
    },
    {
      "epoch": 0.6390030128731854,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008868218153063538,
      "loss": 5.0662,
      "step": 4666
    },
    {
      "epoch": 0.6391399616543413,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008868119722321155,
      "loss": 5.0369,
      "step": 4667
    },
    {
      "epoch": 0.6392769104354972,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008868021255379125,
      "loss": 5.0245,
      "step": 4668
    },
    {
      "epoch": 0.639413859216653,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008867922752238261,
      "loss": 5.0738,
      "step": 4669
    },
    {
      "epoch": 0.6395508079978088,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008867824212899381,
      "loss": 5.1061,
      "step": 4670
    },
    {
      "epoch": 0.6396877567789647,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008867725637363299,
      "loss": 5.1359,
      "step": 4671
    },
    {
      "epoch": 0.6398247055601205,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008867627025630836,
      "loss": 5.1273,
      "step": 4672
    },
    {
      "epoch": 0.6399616543412764,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008867528377702806,
      "loss": 5.0165,
      "step": 4673
    },
    {
      "epoch": 0.6400986031224322,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008867429693580029,
      "loss": 4.9962,
      "step": 4674
    },
    {
      "epoch": 0.640235551903588,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000886733097326332,
      "loss": 5.0742,
      "step": 4675
    },
    {
      "epoch": 0.6403725006847439,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008867232216753499,
      "loss": 5.112,
      "step": 4676
    },
    {
      "epoch": 0.6405094494658997,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008867133424051384,
      "loss": 5.0686,
      "step": 4677
    },
    {
      "epoch": 0.6406463982470556,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008867034595157795,
      "loss": 5.051,
      "step": 4678
    },
    {
      "epoch": 0.6407833470282115,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008866935730073547,
      "loss": 5.0702,
      "step": 4679
    },
    {
      "epoch": 0.6409202958093673,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008866836828799464,
      "loss": 5.0868,
      "step": 4680
    },
    {
      "epoch": 0.6410572445905232,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008866737891336362,
      "loss": 5.1307,
      "step": 4681
    },
    {
      "epoch": 0.641194193371679,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008866638917685064,
      "loss": 5.1217,
      "step": 4682
    },
    {
      "epoch": 0.6413311421528348,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008866539907846388,
      "loss": 5.1023,
      "step": 4683
    },
    {
      "epoch": 0.6414680909339907,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008866440861821156,
      "loss": 5.1345,
      "step": 4684
    },
    {
      "epoch": 0.6416050397151465,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008866341779610187,
      "loss": 5.0762,
      "step": 4685
    },
    {
      "epoch": 0.6417419884963024,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008866242661214305,
      "loss": 5.0576,
      "step": 4686
    },
    {
      "epoch": 0.6418789372774583,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008866143506634328,
      "loss": 5.0778,
      "step": 4687
    },
    {
      "epoch": 0.642015886058614,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000886604431587108,
      "loss": 5.0933,
      "step": 4688
    },
    {
      "epoch": 0.6421528348397699,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008865945088925382,
      "loss": 5.0696,
      "step": 4689
    },
    {
      "epoch": 0.6422897836209258,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008865845825798058,
      "loss": 5.0463,
      "step": 4690
    },
    {
      "epoch": 0.6424267324020816,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008865746526489928,
      "loss": 5.0793,
      "step": 4691
    },
    {
      "epoch": 0.6425636811832375,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008865647191001817,
      "loss": 5.0388,
      "step": 4692
    },
    {
      "epoch": 0.6427006299643934,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008865547819334547,
      "loss": 5.05,
      "step": 4693
    },
    {
      "epoch": 0.6428375787455491,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008865448411488942,
      "loss": 5.085,
      "step": 4694
    },
    {
      "epoch": 0.642974527526705,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008865348967465826,
      "loss": 5.1014,
      "step": 4695
    },
    {
      "epoch": 0.6431114763078608,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008865249487266023,
      "loss": 5.0874,
      "step": 4696
    },
    {
      "epoch": 0.6432484250890167,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008865149970890357,
      "loss": 5.0456,
      "step": 4697
    },
    {
      "epoch": 0.6433853738701726,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008865050418339652,
      "loss": 5.0946,
      "step": 4698
    },
    {
      "epoch": 0.6435223226513284,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008864950829614735,
      "loss": 5.0549,
      "step": 4699
    },
    {
      "epoch": 0.6436592714324842,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008864851204716431,
      "loss": 5.0651,
      "step": 4700
    },
    {
      "epoch": 0.6437962202136401,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008864751543645565,
      "loss": 5.0849,
      "step": 4701
    },
    {
      "epoch": 0.6439331689947959,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008864651846402962,
      "loss": 5.1249,
      "step": 4702
    },
    {
      "epoch": 0.6440701177759518,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008864552112989449,
      "loss": 5.1385,
      "step": 4703
    },
    {
      "epoch": 0.6442070665571077,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008864452343405854,
      "loss": 5.055,
      "step": 4704
    },
    {
      "epoch": 0.6443440153382635,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008864352537653001,
      "loss": 5.1175,
      "step": 4705
    },
    {
      "epoch": 0.6444809641194194,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000886425269573172,
      "loss": 5.0682,
      "step": 4706
    },
    {
      "epoch": 0.6446179129005751,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008864152817642836,
      "loss": 4.9832,
      "step": 4707
    },
    {
      "epoch": 0.644754861681731,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008864052903387177,
      "loss": 5.034,
      "step": 4708
    },
    {
      "epoch": 0.6448918104628869,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008863952952965572,
      "loss": 5.099,
      "step": 4709
    },
    {
      "epoch": 0.6450287592440427,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000886385296637885,
      "loss": 5.1039,
      "step": 4710
    },
    {
      "epoch": 0.6451657080251986,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008863752943627838,
      "loss": 5.1814,
      "step": 4711
    },
    {
      "epoch": 0.6453026568063545,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008863652884713366,
      "loss": 5.0744,
      "step": 4712
    },
    {
      "epoch": 0.6454396055875102,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008863552789636261,
      "loss": 5.0846,
      "step": 4713
    },
    {
      "epoch": 0.6455765543686661,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008863452658397357,
      "loss": 4.9827,
      "step": 4714
    },
    {
      "epoch": 0.645713503149822,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008863352490997479,
      "loss": 5.0877,
      "step": 4715
    },
    {
      "epoch": 0.6458504519309778,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008863252287437459,
      "loss": 5.1056,
      "step": 4716
    },
    {
      "epoch": 0.6459874007121337,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008863152047718129,
      "loss": 5.1092,
      "step": 4717
    },
    {
      "epoch": 0.6461243494932896,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008863051771840319,
      "loss": 5.0806,
      "step": 4718
    },
    {
      "epoch": 0.6462612982744453,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008862951459804857,
      "loss": 5.0152,
      "step": 4719
    },
    {
      "epoch": 0.6463982470556012,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008862851111612577,
      "loss": 5.0618,
      "step": 4720
    },
    {
      "epoch": 0.646535195836757,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008862750727264311,
      "loss": 5.0442,
      "step": 4721
    },
    {
      "epoch": 0.6466721446179129,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000886265030676089,
      "loss": 5.0144,
      "step": 4722
    },
    {
      "epoch": 0.6468090933990688,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008862549850103147,
      "loss": 5.1222,
      "step": 4723
    },
    {
      "epoch": 0.6469460421802246,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008862449357291913,
      "loss": 5.0999,
      "step": 4724
    },
    {
      "epoch": 0.6470829909613804,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008862348828328022,
      "loss": 5.0802,
      "step": 4725
    },
    {
      "epoch": 0.6472199397425363,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008862248263212306,
      "loss": 5.1471,
      "step": 4726
    },
    {
      "epoch": 0.6473568885236921,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00088621476619456,
      "loss": 5.0984,
      "step": 4727
    },
    {
      "epoch": 0.647493837304848,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008862047024528737,
      "loss": 5.1063,
      "step": 4728
    },
    {
      "epoch": 0.6476307860860039,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008861946350962551,
      "loss": 5.075,
      "step": 4729
    },
    {
      "epoch": 0.6477677348671597,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008861845641247875,
      "loss": 5.0875,
      "step": 4730
    },
    {
      "epoch": 0.6479046836483155,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008861744895385546,
      "loss": 5.0321,
      "step": 4731
    },
    {
      "epoch": 0.6480416324294713,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008861644113376397,
      "loss": 5.0389,
      "step": 4732
    },
    {
      "epoch": 0.6481785812106272,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008861543295221264,
      "loss": 5.0394,
      "step": 4733
    },
    {
      "epoch": 0.6483155299917831,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008861442440920983,
      "loss": 5.1146,
      "step": 4734
    },
    {
      "epoch": 0.6484524787729389,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008861341550476388,
      "loss": 5.08,
      "step": 4735
    },
    {
      "epoch": 0.6485894275540948,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008861240623888317,
      "loss": 5.0866,
      "step": 4736
    },
    {
      "epoch": 0.6487263763352507,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008861139661157607,
      "loss": 4.9772,
      "step": 4737
    },
    {
      "epoch": 0.6488633251164064,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008861038662285092,
      "loss": 5.0684,
      "step": 4738
    },
    {
      "epoch": 0.6490002738975623,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008860937627271611,
      "loss": 5.1535,
      "step": 4739
    },
    {
      "epoch": 0.6491372226787182,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008860836556118,
      "loss": 5.1095,
      "step": 4740
    },
    {
      "epoch": 0.649274171459874,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008860735448825098,
      "loss": 5.101,
      "step": 4741
    },
    {
      "epoch": 0.6494111202410299,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008860634305393742,
      "loss": 4.9883,
      "step": 4742
    },
    {
      "epoch": 0.6495480690221856,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008860533125824771,
      "loss": 5.1057,
      "step": 4743
    },
    {
      "epoch": 0.6496850178033415,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008860431910119021,
      "loss": 5.0735,
      "step": 4744
    },
    {
      "epoch": 0.6498219665844974,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008860330658277335,
      "loss": 5.0971,
      "step": 4745
    },
    {
      "epoch": 0.6499589153656532,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008860229370300549,
      "loss": 5.049,
      "step": 4746
    },
    {
      "epoch": 0.6500958641468091,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008860128046189503,
      "loss": 5.1113,
      "step": 4747
    },
    {
      "epoch": 0.650232812927965,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008860026685945036,
      "loss": 5.0258,
      "step": 4748
    },
    {
      "epoch": 0.6503697617091208,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008859925289567991,
      "loss": 5.0489,
      "step": 4749
    },
    {
      "epoch": 0.6505067104902766,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008859823857059204,
      "loss": 5.1036,
      "step": 4750
    },
    {
      "epoch": 0.6506436592714325,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000885972238841952,
      "loss": 5.0498,
      "step": 4751
    },
    {
      "epoch": 0.6507806080525883,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008859620883649777,
      "loss": 5.0962,
      "step": 4752
    },
    {
      "epoch": 0.6509175568337442,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008859519342750816,
      "loss": 5.097,
      "step": 4753
    },
    {
      "epoch": 0.6510545056149001,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000885941776572348,
      "loss": 5.1213,
      "step": 4754
    },
    {
      "epoch": 0.6511914543960559,
      "grad_norm": 0.390625,
      "learning_rate": 0.000885931615256861,
      "loss": 5.0442,
      "step": 4755
    },
    {
      "epoch": 0.6513284031772117,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008859214503287048,
      "loss": 4.9843,
      "step": 4756
    },
    {
      "epoch": 0.6514653519583675,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008859112817879638,
      "loss": 5.0922,
      "step": 4757
    },
    {
      "epoch": 0.6516023007395234,
      "grad_norm": 0.359375,
      "learning_rate": 0.000885901109634722,
      "loss": 5.0751,
      "step": 4758
    },
    {
      "epoch": 0.6517392495206793,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008858909338690639,
      "loss": 5.0618,
      "step": 4759
    },
    {
      "epoch": 0.6518761983018351,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008858807544910737,
      "loss": 5.0096,
      "step": 4760
    },
    {
      "epoch": 0.652013147082991,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008858705715008358,
      "loss": 5.1055,
      "step": 4761
    },
    {
      "epoch": 0.6521500958641469,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008858603848984348,
      "loss": 5.0369,
      "step": 4762
    },
    {
      "epoch": 0.6522870446453026,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008858501946839548,
      "loss": 5.0834,
      "step": 4763
    },
    {
      "epoch": 0.6524239934264585,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008858400008574804,
      "loss": 5.0066,
      "step": 4764
    },
    {
      "epoch": 0.6525609422076144,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008858298034190959,
      "loss": 4.9968,
      "step": 4765
    },
    {
      "epoch": 0.6526978909887702,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008858196023688861,
      "loss": 5.0691,
      "step": 4766
    },
    {
      "epoch": 0.6528348397699261,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008858093977069355,
      "loss": 5.0625,
      "step": 4767
    },
    {
      "epoch": 0.6529717885510818,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008857991894333284,
      "loss": 5.0512,
      "step": 4768
    },
    {
      "epoch": 0.6531087373322377,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008857889775481498,
      "loss": 5.0461,
      "step": 4769
    },
    {
      "epoch": 0.6532456861133936,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008857787620514839,
      "loss": 5.0607,
      "step": 4770
    },
    {
      "epoch": 0.6533826348945494,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008857685429434157,
      "loss": 5.0759,
      "step": 4771
    },
    {
      "epoch": 0.6535195836757053,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008857583202240296,
      "loss": 5.1148,
      "step": 4772
    },
    {
      "epoch": 0.6536565324568612,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008857480938934106,
      "loss": 5.1562,
      "step": 4773
    },
    {
      "epoch": 0.653793481238017,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008857378639516432,
      "loss": 5.1136,
      "step": 4774
    },
    {
      "epoch": 0.6539304300191728,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008857276303988124,
      "loss": 4.9913,
      "step": 4775
    },
    {
      "epoch": 0.6540673788003287,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008857173932350027,
      "loss": 5.0701,
      "step": 4776
    },
    {
      "epoch": 0.6542043275814845,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008857071524602993,
      "loss": 5.0826,
      "step": 4777
    },
    {
      "epoch": 0.6543412763626404,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000885696908074787,
      "loss": 5.0411,
      "step": 4778
    },
    {
      "epoch": 0.6544782251437962,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008856866600785503,
      "loss": 5.0677,
      "step": 4779
    },
    {
      "epoch": 0.654615173924952,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008856764084716747,
      "loss": 5.0362,
      "step": 4780
    },
    {
      "epoch": 0.6547521227061079,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008856661532542448,
      "loss": 5.0818,
      "step": 4781
    },
    {
      "epoch": 0.6548890714872637,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008856558944263458,
      "loss": 5.0883,
      "step": 4782
    },
    {
      "epoch": 0.6550260202684196,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008856456319880624,
      "loss": 5.1698,
      "step": 4783
    },
    {
      "epoch": 0.6551629690495755,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008856353659394798,
      "loss": 4.9326,
      "step": 4784
    },
    {
      "epoch": 0.6552999178307313,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008856250962806832,
      "loss": 5.0779,
      "step": 4785
    },
    {
      "epoch": 0.6554368666118872,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008856148230117575,
      "loss": 5.0697,
      "step": 4786
    },
    {
      "epoch": 0.655573815393043,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008856045461327882,
      "loss": 5.0769,
      "step": 4787
    },
    {
      "epoch": 0.6557107641741988,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008855942656438603,
      "loss": 5.0413,
      "step": 4788
    },
    {
      "epoch": 0.6558477129553547,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008855839815450587,
      "loss": 5.081,
      "step": 4789
    },
    {
      "epoch": 0.6559846617365106,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008855736938364688,
      "loss": 5.0101,
      "step": 4790
    },
    {
      "epoch": 0.6561216105176664,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000885563402518176,
      "loss": 5.0142,
      "step": 4791
    },
    {
      "epoch": 0.6562585592988223,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008855531075902656,
      "loss": 5.0469,
      "step": 4792
    },
    {
      "epoch": 0.656395508079978,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008855428090528227,
      "loss": 5.0535,
      "step": 4793
    },
    {
      "epoch": 0.6565324568611339,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008855325069059328,
      "loss": 5.0311,
      "step": 4794
    },
    {
      "epoch": 0.6566694056422898,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008855222011496811,
      "loss": 5.0908,
      "step": 4795
    },
    {
      "epoch": 0.6568063544234456,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008855118917841533,
      "loss": 5.1427,
      "step": 4796
    },
    {
      "epoch": 0.6569433032046015,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008855015788094346,
      "loss": 5.1209,
      "step": 4797
    },
    {
      "epoch": 0.6570802519857574,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008854912622256104,
      "loss": 5.0988,
      "step": 4798
    },
    {
      "epoch": 0.6572172007669131,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008854809420327666,
      "loss": 5.1121,
      "step": 4799
    },
    {
      "epoch": 0.657354149548069,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008854706182309884,
      "loss": 5.0503,
      "step": 4800
    },
    {
      "epoch": 0.6574910983292249,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008854602908203614,
      "loss": 5.1303,
      "step": 4801
    },
    {
      "epoch": 0.6576280471103807,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008854499598009712,
      "loss": 5.0928,
      "step": 4802
    },
    {
      "epoch": 0.6577649958915366,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008854396251729034,
      "loss": 5.1446,
      "step": 4803
    },
    {
      "epoch": 0.6579019446726924,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008854292869362436,
      "loss": 5.1444,
      "step": 4804
    },
    {
      "epoch": 0.6580388934538483,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008854189450910777,
      "loss": 5.0414,
      "step": 4805
    },
    {
      "epoch": 0.6581758422350041,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008854085996374912,
      "loss": 5.0653,
      "step": 4806
    },
    {
      "epoch": 0.6583127910161599,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008853982505755698,
      "loss": 5.0582,
      "step": 4807
    },
    {
      "epoch": 0.6584497397973158,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008853878979053994,
      "loss": 5.0698,
      "step": 4808
    },
    {
      "epoch": 0.6585866885784717,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008853775416270657,
      "loss": 5.0424,
      "step": 4809
    },
    {
      "epoch": 0.6587236373596275,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008853671817406546,
      "loss": 5.0592,
      "step": 4810
    },
    {
      "epoch": 0.6588605861407834,
      "grad_norm": 0.875,
      "learning_rate": 0.0008853568182462519,
      "loss": 5.0397,
      "step": 4811
    },
    {
      "epoch": 0.6589975349219392,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008853464511439436,
      "loss": 5.1139,
      "step": 4812
    },
    {
      "epoch": 0.659134483703095,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008853360804338154,
      "loss": 5.0464,
      "step": 4813
    },
    {
      "epoch": 0.6592714324842509,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008853257061159534,
      "loss": 5.0885,
      "step": 4814
    },
    {
      "epoch": 0.6594083812654067,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008853153281904436,
      "loss": 5.0405,
      "step": 4815
    },
    {
      "epoch": 0.6595453300465626,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008853049466573719,
      "loss": 5.0153,
      "step": 4816
    },
    {
      "epoch": 0.6596822788277185,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008852945615168244,
      "loss": 5.1421,
      "step": 4817
    },
    {
      "epoch": 0.6598192276088742,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008852841727688872,
      "loss": 4.9383,
      "step": 4818
    },
    {
      "epoch": 0.6599561763900301,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008852737804136463,
      "loss": 5.1023,
      "step": 4819
    },
    {
      "epoch": 0.660093125171186,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008852633844511879,
      "loss": 5.0177,
      "step": 4820
    },
    {
      "epoch": 0.6602300739523418,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008852529848815979,
      "loss": 5.0819,
      "step": 4821
    },
    {
      "epoch": 0.6603670227334977,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008852425817049628,
      "loss": 5.0504,
      "step": 4822
    },
    {
      "epoch": 0.6605039715146536,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008852321749213689,
      "loss": 5.0992,
      "step": 4823
    },
    {
      "epoch": 0.6606409202958093,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008852217645309021,
      "loss": 5.0105,
      "step": 4824
    },
    {
      "epoch": 0.6607778690769652,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008852113505336488,
      "loss": 5.0319,
      "step": 4825
    },
    {
      "epoch": 0.6609148178581211,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008852009329296955,
      "loss": 5.0919,
      "step": 4826
    },
    {
      "epoch": 0.6610517666392769,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008851905117191281,
      "loss": 5.0338,
      "step": 4827
    },
    {
      "epoch": 0.6611887154204328,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008851800869020334,
      "loss": 5.01,
      "step": 4828
    },
    {
      "epoch": 0.6613256642015886,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008851696584784975,
      "loss": 5.1308,
      "step": 4829
    },
    {
      "epoch": 0.6614626129827444,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008851592264486071,
      "loss": 5.065,
      "step": 4830
    },
    {
      "epoch": 0.6615995617639003,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008851487908124485,
      "loss": 5.0569,
      "step": 4831
    },
    {
      "epoch": 0.6617365105450561,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008851383515701081,
      "loss": 5.0833,
      "step": 4832
    },
    {
      "epoch": 0.661873459326212,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008851279087216725,
      "loss": 5.1006,
      "step": 4833
    },
    {
      "epoch": 0.6620104081073679,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008851174622672283,
      "loss": 5.0673,
      "step": 4834
    },
    {
      "epoch": 0.6621473568885237,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008851070122068618,
      "loss": 5.076,
      "step": 4835
    },
    {
      "epoch": 0.6622843056696796,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00088509655854066,
      "loss": 5.0539,
      "step": 4836
    },
    {
      "epoch": 0.6624212544508354,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008850861012687094,
      "loss": 4.9931,
      "step": 4837
    },
    {
      "epoch": 0.6625582032319912,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008850756403910965,
      "loss": 5.1007,
      "step": 4838
    },
    {
      "epoch": 0.6626951520131471,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008850651759079081,
      "loss": 5.0344,
      "step": 4839
    },
    {
      "epoch": 0.6628321007943029,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008850547078192309,
      "loss": 5.0113,
      "step": 4840
    },
    {
      "epoch": 0.6629690495754588,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008850442361251517,
      "loss": 5.0001,
      "step": 4841
    },
    {
      "epoch": 0.6631059983566147,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008850337608257572,
      "loss": 4.9648,
      "step": 4842
    },
    {
      "epoch": 0.6632429471377704,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008850232819211343,
      "loss": 5.1125,
      "step": 4843
    },
    {
      "epoch": 0.6633798959189263,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008850127994113698,
      "loss": 5.1349,
      "step": 4844
    },
    {
      "epoch": 0.6635168447000822,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008850023132965506,
      "loss": 5.0183,
      "step": 4845
    },
    {
      "epoch": 0.663653793481238,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008849918235767635,
      "loss": 5.0564,
      "step": 4846
    },
    {
      "epoch": 0.6637907422623939,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008849813302520956,
      "loss": 5.0608,
      "step": 4847
    },
    {
      "epoch": 0.6639276910435498,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008849708333226338,
      "loss": 5.0534,
      "step": 4848
    },
    {
      "epoch": 0.6640646398247055,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008849603327884649,
      "loss": 5.1507,
      "step": 4849
    },
    {
      "epoch": 0.6642015886058614,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008849498286496762,
      "loss": 5.0101,
      "step": 4850
    },
    {
      "epoch": 0.6643385373870172,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008849393209063545,
      "loss": 5.0495,
      "step": 4851
    },
    {
      "epoch": 0.6644754861681731,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008849288095585872,
      "loss": 5.1058,
      "step": 4852
    },
    {
      "epoch": 0.664612434949329,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008849182946064611,
      "loss": 5.11,
      "step": 4853
    },
    {
      "epoch": 0.6647493837304848,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008849077760500634,
      "loss": 5.0743,
      "step": 4854
    },
    {
      "epoch": 0.6648863325116406,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008848972538894814,
      "loss": 5.0707,
      "step": 4855
    },
    {
      "epoch": 0.6650232812927965,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008848867281248022,
      "loss": 5.0539,
      "step": 4856
    },
    {
      "epoch": 0.6651602300739523,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008848761987561131,
      "loss": 5.0886,
      "step": 4857
    },
    {
      "epoch": 0.6652971788551082,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008848656657835012,
      "loss": 5.0691,
      "step": 4858
    },
    {
      "epoch": 0.6654341276362641,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008848551292070542,
      "loss": 5.1453,
      "step": 4859
    },
    {
      "epoch": 0.6655710764174199,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008848445890268588,
      "loss": 5.0634,
      "step": 4860
    },
    {
      "epoch": 0.6657080251985757,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008848340452430027,
      "loss": 5.0893,
      "step": 4861
    },
    {
      "epoch": 0.6658449739797316,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008848234978555735,
      "loss": 5.0577,
      "step": 4862
    },
    {
      "epoch": 0.6659819227608874,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008848129468646581,
      "loss": 5.0021,
      "step": 4863
    },
    {
      "epoch": 0.6661188715420433,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008848023922703443,
      "loss": 5.0798,
      "step": 4864
    },
    {
      "epoch": 0.6662558203231991,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008847918340727195,
      "loss": 5.1199,
      "step": 4865
    },
    {
      "epoch": 0.666392769104355,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008847812722718711,
      "loss": 5.1206,
      "step": 4866
    },
    {
      "epoch": 0.6665297178855109,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008847707068678867,
      "loss": 5.0688,
      "step": 4867
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000884760137860854,
      "loss": 5.0922,
      "step": 4868
    },
    {
      "epoch": 0.6668036154478225,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008847495652508603,
      "loss": 5.1542,
      "step": 4869
    },
    {
      "epoch": 0.6669405642289784,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008847389890379934,
      "loss": 5.0964,
      "step": 4870
    },
    {
      "epoch": 0.6670775130101342,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008847284092223409,
      "loss": 5.1498,
      "step": 4871
    },
    {
      "epoch": 0.6672144617912901,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008847178258039903,
      "loss": 5.0763,
      "step": 4872
    },
    {
      "epoch": 0.667351410572446,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008847072387830298,
      "loss": 4.9772,
      "step": 4873
    },
    {
      "epoch": 0.6674883593536017,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008846966481595467,
      "loss": 5.0801,
      "step": 4874
    },
    {
      "epoch": 0.6676253081347576,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008846860539336287,
      "loss": 5.0215,
      "step": 4875
    },
    {
      "epoch": 0.6677622569159134,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008846754561053641,
      "loss": 5.0857,
      "step": 4876
    },
    {
      "epoch": 0.6678992056970693,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008846648546748402,
      "loss": 5.0874,
      "step": 4877
    },
    {
      "epoch": 0.6680361544782252,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008846542496421452,
      "loss": 5.1452,
      "step": 4878
    },
    {
      "epoch": 0.668173103259381,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008846436410073668,
      "loss": 5.1225,
      "step": 4879
    },
    {
      "epoch": 0.6683100520405368,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000884633028770593,
      "loss": 5.0821,
      "step": 4880
    },
    {
      "epoch": 0.6684470008216927,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008846224129319118,
      "loss": 5.1131,
      "step": 4881
    },
    {
      "epoch": 0.6685839496028485,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008846117934914109,
      "loss": 5.0848,
      "step": 4882
    },
    {
      "epoch": 0.6687208983840044,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008846011704491787,
      "loss": 5.0651,
      "step": 4883
    },
    {
      "epoch": 0.6688578471651603,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000884590543805303,
      "loss": 5.014,
      "step": 4884
    },
    {
      "epoch": 0.6689947959463161,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008845799135598718,
      "loss": 5.1365,
      "step": 4885
    },
    {
      "epoch": 0.6691317447274719,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008845692797129734,
      "loss": 5.1362,
      "step": 4886
    },
    {
      "epoch": 0.6692686935086277,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008845586422646958,
      "loss": 5.1085,
      "step": 4887
    },
    {
      "epoch": 0.6694056422897836,
      "grad_norm": 0.5,
      "learning_rate": 0.0008845480012151271,
      "loss": 5.1404,
      "step": 4888
    },
    {
      "epoch": 0.6695425910709395,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008845373565643558,
      "loss": 5.0181,
      "step": 4889
    },
    {
      "epoch": 0.6696795398520953,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008845267083124696,
      "loss": 5.1213,
      "step": 4890
    },
    {
      "epoch": 0.6698164886332512,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008845160564595573,
      "loss": 5.1217,
      "step": 4891
    },
    {
      "epoch": 0.669953437414407,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008845054010057067,
      "loss": 5.0689,
      "step": 4892
    },
    {
      "epoch": 0.6700903861955628,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008844947419510064,
      "loss": 5.0133,
      "step": 4893
    },
    {
      "epoch": 0.6702273349767187,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008844840792955446,
      "loss": 5.0979,
      "step": 4894
    },
    {
      "epoch": 0.6703642837578746,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008844734130394096,
      "loss": 5.0547,
      "step": 4895
    },
    {
      "epoch": 0.6705012325390304,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008844627431826901,
      "loss": 5.1592,
      "step": 4896
    },
    {
      "epoch": 0.6706381813201863,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008844520697254742,
      "loss": 5.1258,
      "step": 4897
    },
    {
      "epoch": 0.670775130101342,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008844413926678504,
      "loss": 5.0572,
      "step": 4898
    },
    {
      "epoch": 0.6709120788824979,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008844307120099074,
      "loss": 5.0699,
      "step": 4899
    },
    {
      "epoch": 0.6710490276636538,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008844200277517335,
      "loss": 5.0812,
      "step": 4900
    },
    {
      "epoch": 0.6711859764448096,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008844093398934174,
      "loss": 5.0232,
      "step": 4901
    },
    {
      "epoch": 0.6713229252259655,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008843986484350475,
      "loss": 5.1154,
      "step": 4902
    },
    {
      "epoch": 0.6714598740071214,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008843879533767126,
      "loss": 5.0504,
      "step": 4903
    },
    {
      "epoch": 0.6715968227882771,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008843772547185011,
      "loss": 5.0518,
      "step": 4904
    },
    {
      "epoch": 0.671733771569433,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008843665524605019,
      "loss": 5.0806,
      "step": 4905
    },
    {
      "epoch": 0.6718707203505889,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008843558466028033,
      "loss": 5.0631,
      "step": 4906
    },
    {
      "epoch": 0.6720076691317447,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008843451371454945,
      "loss": 5.0076,
      "step": 4907
    },
    {
      "epoch": 0.6721446179129006,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000884334424088664,
      "loss": 5.0731,
      "step": 4908
    },
    {
      "epoch": 0.6722815666940565,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008843237074324008,
      "loss": 5.0667,
      "step": 4909
    },
    {
      "epoch": 0.6724185154752123,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008843129871767934,
      "loss": 5.0756,
      "step": 4910
    },
    {
      "epoch": 0.6725554642563681,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008843022633219307,
      "loss": 5.0895,
      "step": 4911
    },
    {
      "epoch": 0.6726924130375239,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008842915358679019,
      "loss": 4.992,
      "step": 4912
    },
    {
      "epoch": 0.6728293618186798,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008842808048147954,
      "loss": 5.0621,
      "step": 4913
    },
    {
      "epoch": 0.6729663105998357,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008842700701627005,
      "loss": 5.0407,
      "step": 4914
    },
    {
      "epoch": 0.6731032593809915,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008842593319117061,
      "loss": 4.9676,
      "step": 4915
    },
    {
      "epoch": 0.6732402081621474,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008842485900619011,
      "loss": 5.104,
      "step": 4916
    },
    {
      "epoch": 0.6733771569433032,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008842378446133744,
      "loss": 5.0655,
      "step": 4917
    },
    {
      "epoch": 0.673514105724459,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008842270955662155,
      "loss": 5.0001,
      "step": 4918
    },
    {
      "epoch": 0.6736510545056149,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000884216342920513,
      "loss": 4.979,
      "step": 4919
    },
    {
      "epoch": 0.6737880032867708,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008842055866763563,
      "loss": 5.0528,
      "step": 4920
    },
    {
      "epoch": 0.6739249520679266,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008841948268338343,
      "loss": 5.105,
      "step": 4921
    },
    {
      "epoch": 0.6740619008490825,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008841840633930363,
      "loss": 5.117,
      "step": 4922
    },
    {
      "epoch": 0.6741988496302382,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008841732963540517,
      "loss": 5.0886,
      "step": 4923
    },
    {
      "epoch": 0.6743357984113941,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008841625257169694,
      "loss": 5.0769,
      "step": 4924
    },
    {
      "epoch": 0.67447274719255,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008841517514818788,
      "loss": 5.0838,
      "step": 4925
    },
    {
      "epoch": 0.6746096959737058,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008841409736488692,
      "loss": 5.0851,
      "step": 4926
    },
    {
      "epoch": 0.6747466447548617,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008841301922180298,
      "loss": 5.0563,
      "step": 4927
    },
    {
      "epoch": 0.6748835935360176,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008841194071894502,
      "loss": 5.071,
      "step": 4928
    },
    {
      "epoch": 0.6750205423171733,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008841086185632196,
      "loss": 5.1276,
      "step": 4929
    },
    {
      "epoch": 0.6751574910983292,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008840978263394274,
      "loss": 5.0668,
      "step": 4930
    },
    {
      "epoch": 0.6752944398794851,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008840870305181631,
      "loss": 5.0778,
      "step": 4931
    },
    {
      "epoch": 0.6754313886606409,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008840762310995161,
      "loss": 5.0907,
      "step": 4932
    },
    {
      "epoch": 0.6755683374417968,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000884065428083576,
      "loss": 5.1123,
      "step": 4933
    },
    {
      "epoch": 0.6757052862229526,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008840546214704322,
      "loss": 5.0247,
      "step": 4934
    },
    {
      "epoch": 0.6758422350041084,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008840438112601744,
      "loss": 5.0713,
      "step": 4935
    },
    {
      "epoch": 0.6759791837852643,
      "grad_norm": 0.4375,
      "learning_rate": 0.000884032997452892,
      "loss": 5.149,
      "step": 4936
    },
    {
      "epoch": 0.6761161325664201,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008840221800486748,
      "loss": 5.0921,
      "step": 4937
    },
    {
      "epoch": 0.676253081347576,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008840113590476124,
      "loss": 5.1363,
      "step": 4938
    },
    {
      "epoch": 0.6763900301287319,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008840005344497944,
      "loss": 5.0603,
      "step": 4939
    },
    {
      "epoch": 0.6765269789098877,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008839897062553105,
      "loss": 4.997,
      "step": 4940
    },
    {
      "epoch": 0.6766639276910436,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008839788744642506,
      "loss": 4.9944,
      "step": 4941
    },
    {
      "epoch": 0.6768008764721994,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008839680390767044,
      "loss": 5.0966,
      "step": 4942
    },
    {
      "epoch": 0.6769378252533552,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008839572000927615,
      "loss": 4.9943,
      "step": 4943
    },
    {
      "epoch": 0.6770747740345111,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008839463575125119,
      "loss": 5.0361,
      "step": 4944
    },
    {
      "epoch": 0.677211722815667,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008839355113360454,
      "loss": 5.1328,
      "step": 4945
    },
    {
      "epoch": 0.6773486715968228,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000883924661563452,
      "loss": 5.0255,
      "step": 4946
    },
    {
      "epoch": 0.6774856203779787,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008839138081948216,
      "loss": 5.1041,
      "step": 4947
    },
    {
      "epoch": 0.6776225691591344,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000883902951230244,
      "loss": 5.0555,
      "step": 4948
    },
    {
      "epoch": 0.6777595179402903,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008838920906698092,
      "loss": 5.0579,
      "step": 4949
    },
    {
      "epoch": 0.6778964667214462,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008838812265136072,
      "loss": 5.0434,
      "step": 4950
    },
    {
      "epoch": 0.678033415502602,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008838703587617283,
      "loss": 4.996,
      "step": 4951
    },
    {
      "epoch": 0.6781703642837579,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008838594874142622,
      "loss": 5.0289,
      "step": 4952
    },
    {
      "epoch": 0.6783073130649138,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008838486124712992,
      "loss": 5.0428,
      "step": 4953
    },
    {
      "epoch": 0.6784442618460695,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008838377339329294,
      "loss": 5.0693,
      "step": 4954
    },
    {
      "epoch": 0.6785812106272254,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008838268517992428,
      "loss": 5.0124,
      "step": 4955
    },
    {
      "epoch": 0.6787181594083813,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008838159660703298,
      "loss": 5.0307,
      "step": 4956
    },
    {
      "epoch": 0.6788551081895371,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008838050767462804,
      "loss": 4.9916,
      "step": 4957
    },
    {
      "epoch": 0.678992056970693,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008837941838271851,
      "loss": 5.1258,
      "step": 4958
    },
    {
      "epoch": 0.6791290057518488,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000883783287313134,
      "loss": 5.0465,
      "step": 4959
    },
    {
      "epoch": 0.6792659545330046,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008837723872042174,
      "loss": 5.0801,
      "step": 4960
    },
    {
      "epoch": 0.6794029033141605,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008837614835005256,
      "loss": 5.1041,
      "step": 4961
    },
    {
      "epoch": 0.6795398520953163,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008837505762021491,
      "loss": 5.0801,
      "step": 4962
    },
    {
      "epoch": 0.6796768008764722,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008837396653091783,
      "loss": 5.0737,
      "step": 4963
    },
    {
      "epoch": 0.6798137496576281,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008837287508217034,
      "loss": 5.0085,
      "step": 4964
    },
    {
      "epoch": 0.6799506984387839,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008837178327398152,
      "loss": 5.0713,
      "step": 4965
    },
    {
      "epoch": 0.6800876472199398,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008837069110636037,
      "loss": 5.0158,
      "step": 4966
    },
    {
      "epoch": 0.6802245960010956,
      "grad_norm": 0.75,
      "learning_rate": 0.0008836959857931598,
      "loss": 5.0811,
      "step": 4967
    },
    {
      "epoch": 0.6803615447822514,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008836850569285739,
      "loss": 5.0554,
      "step": 4968
    },
    {
      "epoch": 0.6804984935634073,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008836741244699366,
      "loss": 5.0822,
      "step": 4969
    },
    {
      "epoch": 0.6806354423445631,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008836631884173385,
      "loss": 5.0167,
      "step": 4970
    },
    {
      "epoch": 0.680772391125719,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008836522487708703,
      "loss": 5.096,
      "step": 4971
    },
    {
      "epoch": 0.6809093399068749,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008836413055306225,
      "loss": 5.0371,
      "step": 4972
    },
    {
      "epoch": 0.6810462886880306,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008836303586966858,
      "loss": 5.0988,
      "step": 4973
    },
    {
      "epoch": 0.6811832374691865,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008836194082691511,
      "loss": 5.0589,
      "step": 4974
    },
    {
      "epoch": 0.6813201862503424,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000883608454248109,
      "loss": 5.0965,
      "step": 4975
    },
    {
      "epoch": 0.6814571350314982,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008835974966336502,
      "loss": 5.1157,
      "step": 4976
    },
    {
      "epoch": 0.6815940838126541,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008835865354258658,
      "loss": 5.0379,
      "step": 4977
    },
    {
      "epoch": 0.68173103259381,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008835755706248465,
      "loss": 5.1008,
      "step": 4978
    },
    {
      "epoch": 0.6818679813749657,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000883564602230683,
      "loss": 5.0632,
      "step": 4979
    },
    {
      "epoch": 0.6820049301561216,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008835536302434664,
      "loss": 5.059,
      "step": 4980
    },
    {
      "epoch": 0.6821418789372775,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008835426546632874,
      "loss": 5.0388,
      "step": 4981
    },
    {
      "epoch": 0.6822788277184333,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008835316754902373,
      "loss": 5.0411,
      "step": 4982
    },
    {
      "epoch": 0.6824157764995892,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008835206927244069,
      "loss": 5.0429,
      "step": 4983
    },
    {
      "epoch": 0.682552725280745,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000883509706365887,
      "loss": 5.0655,
      "step": 4984
    },
    {
      "epoch": 0.6826896740619008,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008834987164147691,
      "loss": 5.0783,
      "step": 4985
    },
    {
      "epoch": 0.6828266228430567,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008834877228711441,
      "loss": 5.0315,
      "step": 4986
    },
    {
      "epoch": 0.6829635716242125,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008834767257351029,
      "loss": 5.0703,
      "step": 4987
    },
    {
      "epoch": 0.6831005204053684,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008834657250067366,
      "loss": 5.1125,
      "step": 4988
    },
    {
      "epoch": 0.6832374691865243,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008834547206861368,
      "loss": 4.9894,
      "step": 4989
    },
    {
      "epoch": 0.6833744179676801,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008834437127733944,
      "loss": 4.99,
      "step": 4990
    },
    {
      "epoch": 0.683511366748836,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008834327012686006,
      "loss": 5.1148,
      "step": 4991
    },
    {
      "epoch": 0.6836483155299918,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008834216861718468,
      "loss": 5.0368,
      "step": 4992
    },
    {
      "epoch": 0.6837852643111476,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008834106674832241,
      "loss": 5.0712,
      "step": 4993
    },
    {
      "epoch": 0.6839222130923035,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008833996452028241,
      "loss": 5.0233,
      "step": 4994
    },
    {
      "epoch": 0.6840591618734593,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008833886193307378,
      "loss": 5.093,
      "step": 4995
    },
    {
      "epoch": 0.6841961106546152,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008833775898670568,
      "loss": 5.078,
      "step": 4996
    },
    {
      "epoch": 0.684333059435771,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008833665568118726,
      "loss": 4.9847,
      "step": 4997
    },
    {
      "epoch": 0.6844700082169268,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008833555201652761,
      "loss": 5.1066,
      "step": 4998
    },
    {
      "epoch": 0.6846069569980827,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008833444799273595,
      "loss": 5.0867,
      "step": 4999
    },
    {
      "epoch": 0.6847439057792386,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008833334360982138,
      "loss": 5.1179,
      "step": 5000
    },
    {
      "epoch": 0.6848808545603944,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008833223886779306,
      "loss": 5.0635,
      "step": 5001
    },
    {
      "epoch": 0.6850178033415503,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008833113376666016,
      "loss": 5.0358,
      "step": 5002
    },
    {
      "epoch": 0.6851547521227062,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008833002830643182,
      "loss": 5.0917,
      "step": 5003
    },
    {
      "epoch": 0.6852917009038619,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008832892248711722,
      "loss": 5.1027,
      "step": 5004
    },
    {
      "epoch": 0.6854286496850178,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008832781630872551,
      "loss": 5.0795,
      "step": 5005
    },
    {
      "epoch": 0.6855655984661736,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008832670977126587,
      "loss": 5.0778,
      "step": 5006
    },
    {
      "epoch": 0.6857025472473295,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008832560287474745,
      "loss": 5.0515,
      "step": 5007
    },
    {
      "epoch": 0.6858394960284854,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008832449561917944,
      "loss": 5.0488,
      "step": 5008
    },
    {
      "epoch": 0.6859764448096412,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008832338800457101,
      "loss": 5.0113,
      "step": 5009
    },
    {
      "epoch": 0.686113393590797,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008832228003093134,
      "loss": 5.1303,
      "step": 5010
    },
    {
      "epoch": 0.6862503423719529,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008832117169826962,
      "loss": 5.0451,
      "step": 5011
    },
    {
      "epoch": 0.6863872911531087,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008832006300659503,
      "loss": 5.0554,
      "step": 5012
    },
    {
      "epoch": 0.6865242399342646,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008831895395591674,
      "loss": 4.9901,
      "step": 5013
    },
    {
      "epoch": 0.6866611887154205,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008831784454624396,
      "loss": 5.0611,
      "step": 5014
    },
    {
      "epoch": 0.6867981374965763,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008831673477758587,
      "loss": 5.0048,
      "step": 5015
    },
    {
      "epoch": 0.6869350862777321,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008831562464995169,
      "loss": 5.0698,
      "step": 5016
    },
    {
      "epoch": 0.687072035058888,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008831451416335061,
      "loss": 5.0669,
      "step": 5017
    },
    {
      "epoch": 0.6872089838400438,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008831340331779184,
      "loss": 5.0092,
      "step": 5018
    },
    {
      "epoch": 0.6873459326211997,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008831229211328455,
      "loss": 5.1449,
      "step": 5019
    },
    {
      "epoch": 0.6874828814023555,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008831118054983798,
      "loss": 5.0713,
      "step": 5020
    },
    {
      "epoch": 0.6876198301835114,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008831006862746135,
      "loss": 5.0618,
      "step": 5021
    },
    {
      "epoch": 0.6877567789646672,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008830895634616387,
      "loss": 5.0954,
      "step": 5022
    },
    {
      "epoch": 0.687893727745823,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008830784370595472,
      "loss": 5.064,
      "step": 5023
    },
    {
      "epoch": 0.6880306765269789,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008830673070684316,
      "loss": 5.0457,
      "step": 5024
    },
    {
      "epoch": 0.6881676253081348,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008830561734883842,
      "loss": 5.1473,
      "step": 5025
    },
    {
      "epoch": 0.6883045740892906,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008830450363194968,
      "loss": 5.0749,
      "step": 5026
    },
    {
      "epoch": 0.6884415228704465,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008830338955618622,
      "loss": 5.1028,
      "step": 5027
    },
    {
      "epoch": 0.6885784716516024,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008830227512155724,
      "loss": 5.0206,
      "step": 5028
    },
    {
      "epoch": 0.6887154204327581,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008830116032807199,
      "loss": 5.0114,
      "step": 5029
    },
    {
      "epoch": 0.688852369213914,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008830004517573972,
      "loss": 5.094,
      "step": 5030
    },
    {
      "epoch": 0.6889893179950698,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008829892966456964,
      "loss": 5.0698,
      "step": 5031
    },
    {
      "epoch": 0.6891262667762257,
      "grad_norm": 0.375,
      "learning_rate": 0.0008829781379457103,
      "loss": 5.0741,
      "step": 5032
    },
    {
      "epoch": 0.6892632155573816,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000882966975657531,
      "loss": 5.1913,
      "step": 5033
    },
    {
      "epoch": 0.6894001643385373,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008829558097812514,
      "loss": 5.0778,
      "step": 5034
    },
    {
      "epoch": 0.6895371131196932,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008829446403169638,
      "loss": 5.1101,
      "step": 5035
    },
    {
      "epoch": 0.6896740619008491,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008829334672647608,
      "loss": 5.094,
      "step": 5036
    },
    {
      "epoch": 0.6898110106820049,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008829222906247349,
      "loss": 4.9505,
      "step": 5037
    },
    {
      "epoch": 0.6899479594631608,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008829111103969788,
      "loss": 5.0113,
      "step": 5038
    },
    {
      "epoch": 0.6900849082443167,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008828999265815854,
      "loss": 5.0648,
      "step": 5039
    },
    {
      "epoch": 0.6902218570254725,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008828887391786469,
      "loss": 5.0938,
      "step": 5040
    },
    {
      "epoch": 0.6903588058066283,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008828775481882563,
      "loss": 5.0736,
      "step": 5041
    },
    {
      "epoch": 0.6904957545877841,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008828663536105063,
      "loss": 5.0125,
      "step": 5042
    },
    {
      "epoch": 0.69063270336894,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008828551554454898,
      "loss": 5.088,
      "step": 5043
    },
    {
      "epoch": 0.6907696521500959,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008828439536932994,
      "loss": 5.0339,
      "step": 5044
    },
    {
      "epoch": 0.6909066009312517,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008828327483540281,
      "loss": 5.1079,
      "step": 5045
    },
    {
      "epoch": 0.6910435497124076,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008828215394277686,
      "loss": 5.0982,
      "step": 5046
    },
    {
      "epoch": 0.6911804984935634,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008828103269146138,
      "loss": 5.0515,
      "step": 5047
    },
    {
      "epoch": 0.6913174472747192,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008827991108146568,
      "loss": 5.0709,
      "step": 5048
    },
    {
      "epoch": 0.6914543960558751,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008827878911279904,
      "loss": 5.103,
      "step": 5049
    },
    {
      "epoch": 0.691591344837031,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008827766678547076,
      "loss": 5.0833,
      "step": 5050
    },
    {
      "epoch": 0.6917282936181868,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008827654409949015,
      "loss": 5.0341,
      "step": 5051
    },
    {
      "epoch": 0.6918652423993427,
      "grad_norm": 0.40625,
      "learning_rate": 0.000882754210548665,
      "loss": 5.0705,
      "step": 5052
    },
    {
      "epoch": 0.6920021911804984,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008827429765160912,
      "loss": 5.0476,
      "step": 5053
    },
    {
      "epoch": 0.6921391399616543,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008827317388972733,
      "loss": 4.9666,
      "step": 5054
    },
    {
      "epoch": 0.6922760887428102,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008827204976923045,
      "loss": 5.1538,
      "step": 5055
    },
    {
      "epoch": 0.692413037523966,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008827092529012776,
      "loss": 5.021,
      "step": 5056
    },
    {
      "epoch": 0.6925499863051219,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008826980045242861,
      "loss": 5.1005,
      "step": 5057
    },
    {
      "epoch": 0.6926869350862778,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008826867525614232,
      "loss": 5.0862,
      "step": 5058
    },
    {
      "epoch": 0.6928238838674335,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008826754970127819,
      "loss": 5.103,
      "step": 5059
    },
    {
      "epoch": 0.6929608326485894,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008826642378784559,
      "loss": 5.0049,
      "step": 5060
    },
    {
      "epoch": 0.6930977814297453,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008826529751585381,
      "loss": 5.0677,
      "step": 5061
    },
    {
      "epoch": 0.6932347302109011,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000882641708853122,
      "loss": 5.0451,
      "step": 5062
    },
    {
      "epoch": 0.693371678992057,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000882630438962301,
      "loss": 5.0641,
      "step": 5063
    },
    {
      "epoch": 0.6935086277732129,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008826191654861683,
      "loss": 5.075,
      "step": 5064
    },
    {
      "epoch": 0.6936455765543686,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008826078884248178,
      "loss": 5.0733,
      "step": 5065
    },
    {
      "epoch": 0.6937825253355245,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008825966077783424,
      "loss": 5.016,
      "step": 5066
    },
    {
      "epoch": 0.6939194741166803,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008825853235468359,
      "loss": 5.1299,
      "step": 5067
    },
    {
      "epoch": 0.6940564228978362,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008825740357303917,
      "loss": 5.0526,
      "step": 5068
    },
    {
      "epoch": 0.6941933716789921,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008825627443291036,
      "loss": 4.9782,
      "step": 5069
    },
    {
      "epoch": 0.6943303204601479,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008825514493430647,
      "loss": 5.0359,
      "step": 5070
    },
    {
      "epoch": 0.6944672692413038,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008825401507723691,
      "loss": 5.0686,
      "step": 5071
    },
    {
      "epoch": 0.6946042180224596,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00088252884861711,
      "loss": 5.0267,
      "step": 5072
    },
    {
      "epoch": 0.6947411668036154,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008825175428773814,
      "loss": 5.0897,
      "step": 5073
    },
    {
      "epoch": 0.6948781155847713,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008825062335532768,
      "loss": 5.0957,
      "step": 5074
    },
    {
      "epoch": 0.6950150643659272,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008824949206448899,
      "loss": 5.0873,
      "step": 5075
    },
    {
      "epoch": 0.695152013147083,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008824836041523148,
      "loss": 4.9679,
      "step": 5076
    },
    {
      "epoch": 0.6952889619282389,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008824722840756447,
      "loss": 5.1776,
      "step": 5077
    },
    {
      "epoch": 0.6954259107093946,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000882460960414974,
      "loss": 5.1389,
      "step": 5078
    },
    {
      "epoch": 0.6955628594905505,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000882449633170396,
      "loss": 4.974,
      "step": 5079
    },
    {
      "epoch": 0.6956998082717064,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008824383023420051,
      "loss": 5.0522,
      "step": 5080
    },
    {
      "epoch": 0.6958367570528622,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008824269679298948,
      "loss": 5.051,
      "step": 5081
    },
    {
      "epoch": 0.6959737058340181,
      "grad_norm": 0.375,
      "learning_rate": 0.0008824156299341591,
      "loss": 5.0491,
      "step": 5082
    },
    {
      "epoch": 0.696110654615174,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008824042883548922,
      "loss": 5.0926,
      "step": 5083
    },
    {
      "epoch": 0.6962476033963297,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008823929431921878,
      "loss": 5.0387,
      "step": 5084
    },
    {
      "epoch": 0.6963845521774856,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008823815944461402,
      "loss": 5.06,
      "step": 5085
    },
    {
      "epoch": 0.6965215009586415,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008823702421168431,
      "loss": 5.058,
      "step": 5086
    },
    {
      "epoch": 0.6966584497397973,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000882358886204391,
      "loss": 5.0386,
      "step": 5087
    },
    {
      "epoch": 0.6967953985209532,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008823475267088776,
      "loss": 5.037,
      "step": 5088
    },
    {
      "epoch": 0.696932347302109,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008823361636303972,
      "loss": 5.0385,
      "step": 5089
    },
    {
      "epoch": 0.6970692960832648,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008823247969690441,
      "loss": 5.096,
      "step": 5090
    },
    {
      "epoch": 0.6972062448644207,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008823134267249124,
      "loss": 5.0233,
      "step": 5091
    },
    {
      "epoch": 0.6973431936455765,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008823020528980962,
      "loss": 5.0502,
      "step": 5092
    },
    {
      "epoch": 0.6974801424267324,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00088229067548869,
      "loss": 5.0026,
      "step": 5093
    },
    {
      "epoch": 0.6976170912078883,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000882279294496788,
      "loss": 5.0725,
      "step": 5094
    },
    {
      "epoch": 0.6977540399890441,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008822679099224844,
      "loss": 5.0391,
      "step": 5095
    },
    {
      "epoch": 0.6978909887702,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008822565217658737,
      "loss": 5.0911,
      "step": 5096
    },
    {
      "epoch": 0.6980279375513558,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008822451300270501,
      "loss": 5.0849,
      "step": 5097
    },
    {
      "epoch": 0.6981648863325116,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008822337347061081,
      "loss": 5.0015,
      "step": 5098
    },
    {
      "epoch": 0.6983018351136675,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008822223358031422,
      "loss": 5.1098,
      "step": 5099
    },
    {
      "epoch": 0.6984387838948234,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008822109333182468,
      "loss": 5.0508,
      "step": 5100
    },
    {
      "epoch": 0.6985757326759792,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008821995272515164,
      "loss": 5.0789,
      "step": 5101
    },
    {
      "epoch": 0.698712681457135,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008821881176030455,
      "loss": 5.034,
      "step": 5102
    },
    {
      "epoch": 0.6988496302382908,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008821767043729287,
      "loss": 5.0541,
      "step": 5103
    },
    {
      "epoch": 0.6989865790194467,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008821652875612607,
      "loss": 5.1041,
      "step": 5104
    },
    {
      "epoch": 0.6991235278006026,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008821538671681359,
      "loss": 5.0867,
      "step": 5105
    },
    {
      "epoch": 0.6992604765817584,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008821424431936489,
      "loss": 5.1043,
      "step": 5106
    },
    {
      "epoch": 0.6993974253629143,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008821310156378946,
      "loss": 5.0405,
      "step": 5107
    },
    {
      "epoch": 0.6995343741440702,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008821195845009676,
      "loss": 5.0738,
      "step": 5108
    },
    {
      "epoch": 0.6996713229252259,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008821081497829626,
      "loss": 5.0462,
      "step": 5109
    },
    {
      "epoch": 0.6998082717063818,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008820967114839746,
      "loss": 5.0553,
      "step": 5110
    },
    {
      "epoch": 0.6999452204875377,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008820852696040978,
      "loss": 5.0939,
      "step": 5111
    },
    {
      "epoch": 0.7000821692686935,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008820738241434277,
      "loss": 5.0253,
      "step": 5112
    },
    {
      "epoch": 0.7002191180498494,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008820623751020588,
      "loss": 5.0937,
      "step": 5113
    },
    {
      "epoch": 0.7003560668310052,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000882050922480086,
      "loss": 5.1489,
      "step": 5114
    },
    {
      "epoch": 0.700493015612161,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008820394662776044,
      "loss": 5.0861,
      "step": 5115
    },
    {
      "epoch": 0.7006299643933169,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008820280064947086,
      "loss": 5.0585,
      "step": 5116
    },
    {
      "epoch": 0.7007669131744727,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008820165431314939,
      "loss": 5.0305,
      "step": 5117
    },
    {
      "epoch": 0.7009038619556286,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000882005076188055,
      "loss": 5.0915,
      "step": 5118
    },
    {
      "epoch": 0.7010408107367845,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008819936056644873,
      "loss": 5.0067,
      "step": 5119
    },
    {
      "epoch": 0.7011777595179403,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008819821315608855,
      "loss": 5.0748,
      "step": 5120
    },
    {
      "epoch": 0.7013147082990961,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008819706538773448,
      "loss": 5.0242,
      "step": 5121
    },
    {
      "epoch": 0.701451657080252,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008819591726139605,
      "loss": 4.9965,
      "step": 5122
    },
    {
      "epoch": 0.7015886058614078,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008819476877708276,
      "loss": 5.0446,
      "step": 5123
    },
    {
      "epoch": 0.7017255546425637,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008819361993480412,
      "loss": 5.0769,
      "step": 5124
    },
    {
      "epoch": 0.7018625034237195,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008819247073456966,
      "loss": 5.0023,
      "step": 5125
    },
    {
      "epoch": 0.7019994522048754,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008819132117638891,
      "loss": 5.0731,
      "step": 5126
    },
    {
      "epoch": 0.7021364009860313,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000881901712602714,
      "loss": 5.043,
      "step": 5127
    },
    {
      "epoch": 0.702273349767187,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008818902098622662,
      "loss": 5.0947,
      "step": 5128
    },
    {
      "epoch": 0.7024102985483429,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008818787035426415,
      "loss": 5.0696,
      "step": 5129
    },
    {
      "epoch": 0.7025472473294988,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000881867193643935,
      "loss": 5.0282,
      "step": 5130
    },
    {
      "epoch": 0.7026841961106546,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008818556801662421,
      "loss": 5.0556,
      "step": 5131
    },
    {
      "epoch": 0.7028211448918105,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008818441631096585,
      "loss": 5.0663,
      "step": 5132
    },
    {
      "epoch": 0.7029580936729664,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008818326424742792,
      "loss": 5.1084,
      "step": 5133
    },
    {
      "epoch": 0.7030950424541221,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008818211182602,
      "loss": 5.007,
      "step": 5134
    },
    {
      "epoch": 0.703231991235278,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008818095904675162,
      "loss": 5.023,
      "step": 5135
    },
    {
      "epoch": 0.7033689400164339,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008817980590963237,
      "loss": 5.1561,
      "step": 5136
    },
    {
      "epoch": 0.7035058887975897,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008817865241467175,
      "loss": 5.0951,
      "step": 5137
    },
    {
      "epoch": 0.7036428375787456,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008817749856187936,
      "loss": 5.1094,
      "step": 5138
    },
    {
      "epoch": 0.7037797863599013,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008817634435126474,
      "loss": 5.0491,
      "step": 5139
    },
    {
      "epoch": 0.7039167351410572,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008817518978283748,
      "loss": 5.0136,
      "step": 5140
    },
    {
      "epoch": 0.7040536839222131,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008817403485660713,
      "loss": 5.0751,
      "step": 5141
    },
    {
      "epoch": 0.7041906327033689,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008817287957258326,
      "loss": 5.0598,
      "step": 5142
    },
    {
      "epoch": 0.7043275814845248,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008817172393077546,
      "loss": 5.0214,
      "step": 5143
    },
    {
      "epoch": 0.7044645302656807,
      "grad_norm": 0.421875,
      "learning_rate": 0.000881705679311933,
      "loss": 5.0966,
      "step": 5144
    },
    {
      "epoch": 0.7046014790468365,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008816941157384633,
      "loss": 4.9488,
      "step": 5145
    },
    {
      "epoch": 0.7047384278279923,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008816825485874417,
      "loss": 5.0201,
      "step": 5146
    },
    {
      "epoch": 0.7048753766091482,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008816709778589641,
      "loss": 5.0856,
      "step": 5147
    },
    {
      "epoch": 0.705012325390304,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008816594035531263,
      "loss": 5.0886,
      "step": 5148
    },
    {
      "epoch": 0.7051492741714599,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008816478256700239,
      "loss": 5.0379,
      "step": 5149
    },
    {
      "epoch": 0.7052862229526157,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008816362442097532,
      "loss": 5.0744,
      "step": 5150
    },
    {
      "epoch": 0.7054231717337716,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008816246591724101,
      "loss": 5.1071,
      "step": 5151
    },
    {
      "epoch": 0.7055601205149274,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008816130705580907,
      "loss": 5.0738,
      "step": 5152
    },
    {
      "epoch": 0.7056970692960832,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008816014783668909,
      "loss": 4.9967,
      "step": 5153
    },
    {
      "epoch": 0.7058340180772391,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008815898825989067,
      "loss": 5.0428,
      "step": 5154
    },
    {
      "epoch": 0.705970966858395,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008815782832542345,
      "loss": 5.0068,
      "step": 5155
    },
    {
      "epoch": 0.7061079156395508,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008815666803329699,
      "loss": 4.9712,
      "step": 5156
    },
    {
      "epoch": 0.7062448644207067,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008815550738352097,
      "loss": 5.0742,
      "step": 5157
    },
    {
      "epoch": 0.7063818132018626,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008815434637610497,
      "loss": 5.0108,
      "step": 5158
    },
    {
      "epoch": 0.7065187619830183,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000881531850110586,
      "loss": 5.0646,
      "step": 5159
    },
    {
      "epoch": 0.7066557107641742,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008815202328839152,
      "loss": 4.9707,
      "step": 5160
    },
    {
      "epoch": 0.70679265954533,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008815086120811333,
      "loss": 5.0198,
      "step": 5161
    },
    {
      "epoch": 0.7069296083264859,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008814969877023369,
      "loss": 5.0222,
      "step": 5162
    },
    {
      "epoch": 0.7070665571076418,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000881485359747622,
      "loss": 5.0322,
      "step": 5163
    },
    {
      "epoch": 0.7072035058887975,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008814737282170852,
      "loss": 5.0007,
      "step": 5164
    },
    {
      "epoch": 0.7073404546699534,
      "grad_norm": 0.328125,
      "learning_rate": 0.0008814620931108227,
      "loss": 5.002,
      "step": 5165
    },
    {
      "epoch": 0.7074774034511093,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000881450454428931,
      "loss": 5.0463,
      "step": 5166
    },
    {
      "epoch": 0.7076143522322651,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008814388121715065,
      "loss": 5.0087,
      "step": 5167
    },
    {
      "epoch": 0.707751301013421,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008814271663386459,
      "loss": 5.0882,
      "step": 5168
    },
    {
      "epoch": 0.7078882497945769,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008814155169304456,
      "loss": 5.0424,
      "step": 5169
    },
    {
      "epoch": 0.7080251985757327,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000881403863947002,
      "loss": 5.1077,
      "step": 5170
    },
    {
      "epoch": 0.7081621473568885,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008813922073884119,
      "loss": 5.0815,
      "step": 5171
    },
    {
      "epoch": 0.7082990961380444,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008813805472547717,
      "loss": 5.0913,
      "step": 5172
    },
    {
      "epoch": 0.7084360449192002,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008813688835461783,
      "loss": 4.9917,
      "step": 5173
    },
    {
      "epoch": 0.7085729937003561,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008813572162627278,
      "loss": 5.0886,
      "step": 5174
    },
    {
      "epoch": 0.7087099424815119,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008813455454045176,
      "loss": 4.9705,
      "step": 5175
    },
    {
      "epoch": 0.7088468912626678,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008813338709716439,
      "loss": 5.1266,
      "step": 5176
    },
    {
      "epoch": 0.7089838400438236,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008813221929642038,
      "loss": 5.017,
      "step": 5177
    },
    {
      "epoch": 0.7091207888249794,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008813105113822937,
      "loss": 5.0148,
      "step": 5178
    },
    {
      "epoch": 0.7092577376061353,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008812988262260107,
      "loss": 5.0893,
      "step": 5179
    },
    {
      "epoch": 0.7093946863872912,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008812871374954515,
      "loss": 5.0815,
      "step": 5180
    },
    {
      "epoch": 0.709531635168447,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008812754451907131,
      "loss": 5.0021,
      "step": 5181
    },
    {
      "epoch": 0.7096685839496029,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008812637493118922,
      "loss": 5.0688,
      "step": 5182
    },
    {
      "epoch": 0.7098055327307587,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008812520498590862,
      "loss": 5.0829,
      "step": 5183
    },
    {
      "epoch": 0.7099424815119145,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008812403468323914,
      "loss": 5.1148,
      "step": 5184
    },
    {
      "epoch": 0.7100794302930704,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000881228640231905,
      "loss": 5.0586,
      "step": 5185
    },
    {
      "epoch": 0.7102163790742262,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008812169300577243,
      "loss": 5.0996,
      "step": 5186
    },
    {
      "epoch": 0.7103533278553821,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008812052163099461,
      "loss": 5.1177,
      "step": 5187
    },
    {
      "epoch": 0.710490276636538,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008811934989886675,
      "loss": 5.051,
      "step": 5188
    },
    {
      "epoch": 0.7106272254176937,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008811817780939856,
      "loss": 5.0795,
      "step": 5189
    },
    {
      "epoch": 0.7107641741988496,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008811700536259975,
      "loss": 5.1148,
      "step": 5190
    },
    {
      "epoch": 0.7109011229800055,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008811583255848006,
      "loss": 5.0503,
      "step": 5191
    },
    {
      "epoch": 0.7110380717611613,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008811465939704918,
      "loss": 5.0073,
      "step": 5192
    },
    {
      "epoch": 0.7111750205423172,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008811348587831685,
      "loss": 5.0008,
      "step": 5193
    },
    {
      "epoch": 0.7113119693234731,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008811231200229278,
      "loss": 5.0133,
      "step": 5194
    },
    {
      "epoch": 0.7114489181046288,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008811113776898671,
      "loss": 5.0051,
      "step": 5195
    },
    {
      "epoch": 0.7115858668857847,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008810996317840838,
      "loss": 4.9845,
      "step": 5196
    },
    {
      "epoch": 0.7117228156669405,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000881087882305675,
      "loss": 5.0406,
      "step": 5197
    },
    {
      "epoch": 0.7118597644480964,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008810761292547383,
      "loss": 5.0103,
      "step": 5198
    },
    {
      "epoch": 0.7119967132292523,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008810643726313709,
      "loss": 5.0639,
      "step": 5199
    },
    {
      "epoch": 0.7121336620104081,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008810526124356703,
      "loss": 5.0647,
      "step": 5200
    },
    {
      "epoch": 0.712270610791564,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008810408486677341,
      "loss": 5.0314,
      "step": 5201
    },
    {
      "epoch": 0.7124075595727198,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008810290813276597,
      "loss": 5.0445,
      "step": 5202
    },
    {
      "epoch": 0.7125445083538756,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008810173104155445,
      "loss": 5.0641,
      "step": 5203
    },
    {
      "epoch": 0.7126814571350315,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008810055359314862,
      "loss": 5.0328,
      "step": 5204
    },
    {
      "epoch": 0.7128184059161874,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008809937578755822,
      "loss": 4.9956,
      "step": 5205
    },
    {
      "epoch": 0.7129553546973432,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008809819762479303,
      "loss": 4.9851,
      "step": 5206
    },
    {
      "epoch": 0.7130923034784991,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008809701910486281,
      "loss": 5.047,
      "step": 5207
    },
    {
      "epoch": 0.7132292522596549,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008809584022777732,
      "loss": 4.9737,
      "step": 5208
    },
    {
      "epoch": 0.7133662010408107,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008809466099354634,
      "loss": 5.0941,
      "step": 5209
    },
    {
      "epoch": 0.7135031498219666,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008809348140217964,
      "loss": 4.9697,
      "step": 5210
    },
    {
      "epoch": 0.7136400986031224,
      "grad_norm": 0.5,
      "learning_rate": 0.0008809230145368697,
      "loss": 5.1278,
      "step": 5211
    },
    {
      "epoch": 0.7137770473842783,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008809112114807814,
      "loss": 4.9748,
      "step": 5212
    },
    {
      "epoch": 0.7139139961654342,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008808994048536291,
      "loss": 5.056,
      "step": 5213
    },
    {
      "epoch": 0.7140509449465899,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008808875946555109,
      "loss": 5.0144,
      "step": 5214
    },
    {
      "epoch": 0.7141878937277458,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008808757808865245,
      "loss": 5.0649,
      "step": 5215
    },
    {
      "epoch": 0.7143248425089017,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008808639635467678,
      "loss": 4.9879,
      "step": 5216
    },
    {
      "epoch": 0.7144617912900575,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008808521426363388,
      "loss": 5.0101,
      "step": 5217
    },
    {
      "epoch": 0.7145987400712134,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008808403181553354,
      "loss": 5.0664,
      "step": 5218
    },
    {
      "epoch": 0.7147356888523693,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008808284901038558,
      "loss": 5.0883,
      "step": 5219
    },
    {
      "epoch": 0.714872637633525,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008808166584819975,
      "loss": 5.0454,
      "step": 5220
    },
    {
      "epoch": 0.7150095864146809,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008808048232898593,
      "loss": 5.0403,
      "step": 5221
    },
    {
      "epoch": 0.7151465351958367,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008807929845275385,
      "loss": 5.0386,
      "step": 5222
    },
    {
      "epoch": 0.7152834839769926,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008807811421951338,
      "loss": 5.0134,
      "step": 5223
    },
    {
      "epoch": 0.7154204327581485,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000880769296292743,
      "loss": 5.1044,
      "step": 5224
    },
    {
      "epoch": 0.7155573815393043,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008807574468204644,
      "loss": 5.0449,
      "step": 5225
    },
    {
      "epoch": 0.7156943303204601,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008807455937783963,
      "loss": 5.0422,
      "step": 5226
    },
    {
      "epoch": 0.715831279101616,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008807337371666367,
      "loss": 5.0111,
      "step": 5227
    },
    {
      "epoch": 0.7159682278827718,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008807218769852841,
      "loss": 5.0073,
      "step": 5228
    },
    {
      "epoch": 0.7161051766639277,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008807100132344367,
      "loss": 5.0561,
      "step": 5229
    },
    {
      "epoch": 0.7162421254450836,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008806981459141926,
      "loss": 5.0267,
      "step": 5230
    },
    {
      "epoch": 0.7163790742262394,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008806862750246504,
      "loss": 5.0246,
      "step": 5231
    },
    {
      "epoch": 0.7165160230073953,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008806744005659085,
      "loss": 5.0669,
      "step": 5232
    },
    {
      "epoch": 0.716652971788551,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008806625225380652,
      "loss": 5.1089,
      "step": 5233
    },
    {
      "epoch": 0.7167899205697069,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008806506409412189,
      "loss": 5.1094,
      "step": 5234
    },
    {
      "epoch": 0.7169268693508628,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008806387557754681,
      "loss": 4.9988,
      "step": 5235
    },
    {
      "epoch": 0.7170638181320186,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008806268670409114,
      "loss": 5.0578,
      "step": 5236
    },
    {
      "epoch": 0.7172007669131745,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008806149747376473,
      "loss": 5.0415,
      "step": 5237
    },
    {
      "epoch": 0.7173377156943304,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008806030788657741,
      "loss": 5.0615,
      "step": 5238
    },
    {
      "epoch": 0.7174746644754861,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008805911794253907,
      "loss": 5.0957,
      "step": 5239
    },
    {
      "epoch": 0.717611613256642,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008805792764165956,
      "loss": 5.1165,
      "step": 5240
    },
    {
      "epoch": 0.7177485620377979,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008805673698394874,
      "loss": 5.1027,
      "step": 5241
    },
    {
      "epoch": 0.7178855108189537,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008805554596941648,
      "loss": 5.0079,
      "step": 5242
    },
    {
      "epoch": 0.7180224596001096,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008805435459807265,
      "loss": 5.0691,
      "step": 5243
    },
    {
      "epoch": 0.7181594083812654,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008805316286992712,
      "loss": 5.1298,
      "step": 5244
    },
    {
      "epoch": 0.7182963571624212,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008805197078498978,
      "loss": 5.0152,
      "step": 5245
    },
    {
      "epoch": 0.7184333059435771,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000880507783432705,
      "loss": 5.0438,
      "step": 5246
    },
    {
      "epoch": 0.7185702547247329,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008804958554477915,
      "loss": 4.9298,
      "step": 5247
    },
    {
      "epoch": 0.7187072035058888,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008804839238952564,
      "loss": 5.0459,
      "step": 5248
    },
    {
      "epoch": 0.7188441522870447,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008804719887751984,
      "loss": 4.9939,
      "step": 5249
    },
    {
      "epoch": 0.7189811010682005,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008804600500877164,
      "loss": 5.0165,
      "step": 5250
    },
    {
      "epoch": 0.7191180498493563,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008804481078329094,
      "loss": 5.0208,
      "step": 5251
    },
    {
      "epoch": 0.7192549986305122,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008804361620108765,
      "loss": 5.0591,
      "step": 5252
    },
    {
      "epoch": 0.719391947411668,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008804242126217162,
      "loss": 5.0707,
      "step": 5253
    },
    {
      "epoch": 0.7195288961928239,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008804122596655282,
      "loss": 4.9893,
      "step": 5254
    },
    {
      "epoch": 0.7196658449739798,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008804003031424113,
      "loss": 4.954,
      "step": 5255
    },
    {
      "epoch": 0.7198027937551356,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008803883430524643,
      "loss": 5.0245,
      "step": 5256
    },
    {
      "epoch": 0.7199397425362914,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008803763793957867,
      "loss": 5.0423,
      "step": 5257
    },
    {
      "epoch": 0.7200766913174472,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008803644121724774,
      "loss": 5.0379,
      "step": 5258
    },
    {
      "epoch": 0.7202136400986031,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008803524413826357,
      "loss": 5.0398,
      "step": 5259
    },
    {
      "epoch": 0.720350588879759,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008803404670263608,
      "loss": 5.0564,
      "step": 5260
    },
    {
      "epoch": 0.7204875376609148,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008803284891037518,
      "loss": 5.0837,
      "step": 5261
    },
    {
      "epoch": 0.7206244864420707,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008803165076149082,
      "loss": 5.0607,
      "step": 5262
    },
    {
      "epoch": 0.7207614352232266,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008803045225599292,
      "loss": 5.0581,
      "step": 5263
    },
    {
      "epoch": 0.7208983840043823,
      "grad_norm": 0.375,
      "learning_rate": 0.0008802925339389139,
      "loss": 5.0388,
      "step": 5264
    },
    {
      "epoch": 0.7210353327855382,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000880280541751962,
      "loss": 5.0084,
      "step": 5265
    },
    {
      "epoch": 0.7211722815666941,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008802685459991725,
      "loss": 5.0881,
      "step": 5266
    },
    {
      "epoch": 0.7213092303478499,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008802565466806452,
      "loss": 5.009,
      "step": 5267
    },
    {
      "epoch": 0.7214461791290058,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008802445437964795,
      "loss": 4.9976,
      "step": 5268
    },
    {
      "epoch": 0.7215831279101615,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008802325373467745,
      "loss": 4.964,
      "step": 5269
    },
    {
      "epoch": 0.7217200766913174,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008802205273316301,
      "loss": 5.0468,
      "step": 5270
    },
    {
      "epoch": 0.7218570254724733,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008802085137511456,
      "loss": 5.1101,
      "step": 5271
    },
    {
      "epoch": 0.7219939742536291,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008801964966054206,
      "loss": 5.0881,
      "step": 5272
    },
    {
      "epoch": 0.722130923034785,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008801844758945548,
      "loss": 5.0708,
      "step": 5273
    },
    {
      "epoch": 0.7222678718159409,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008801724516186478,
      "loss": 5.0733,
      "step": 5274
    },
    {
      "epoch": 0.7224048205970967,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000880160423777799,
      "loss": 5.1324,
      "step": 5275
    },
    {
      "epoch": 0.7225417693782525,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008801483923721084,
      "loss": 5.0745,
      "step": 5276
    },
    {
      "epoch": 0.7226787181594084,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008801363574016756,
      "loss": 4.9643,
      "step": 5277
    },
    {
      "epoch": 0.7228156669405642,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008801243188666001,
      "loss": 5.058,
      "step": 5278
    },
    {
      "epoch": 0.7229526157217201,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008801122767669821,
      "loss": 5.0034,
      "step": 5279
    },
    {
      "epoch": 0.7230895645028759,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008801002311029211,
      "loss": 5.0062,
      "step": 5280
    },
    {
      "epoch": 0.7232265132840318,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008800881818745171,
      "loss": 5.084,
      "step": 5281
    },
    {
      "epoch": 0.7233634620651876,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008800761290818697,
      "loss": 5.0282,
      "step": 5282
    },
    {
      "epoch": 0.7235004108463434,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008800640727250791,
      "loss": 5.0052,
      "step": 5283
    },
    {
      "epoch": 0.7236373596274993,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008800520128042448,
      "loss": 5.0239,
      "step": 5284
    },
    {
      "epoch": 0.7237743084086552,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008800399493194671,
      "loss": 5.1164,
      "step": 5285
    },
    {
      "epoch": 0.723911257189811,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000880027882270846,
      "loss": 5.0359,
      "step": 5286
    },
    {
      "epoch": 0.7240482059709669,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008800158116584811,
      "loss": 5.0829,
      "step": 5287
    },
    {
      "epoch": 0.7241851547521228,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000880003737482473,
      "loss": 5.1052,
      "step": 5288
    },
    {
      "epoch": 0.7243221035332785,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008799916597429213,
      "loss": 5.0436,
      "step": 5289
    },
    {
      "epoch": 0.7244590523144344,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008799795784399264,
      "loss": 5.0077,
      "step": 5290
    },
    {
      "epoch": 0.7245960010955903,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008799674935735881,
      "loss": 5.0637,
      "step": 5291
    },
    {
      "epoch": 0.7247329498767461,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008799554051440067,
      "loss": 5.0288,
      "step": 5292
    },
    {
      "epoch": 0.724869898657902,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008799433131512826,
      "loss": 5.0104,
      "step": 5293
    },
    {
      "epoch": 0.7250068474390577,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008799312175955157,
      "loss": 5.1036,
      "step": 5294
    },
    {
      "epoch": 0.7251437962202136,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008799191184768063,
      "loss": 5.0956,
      "step": 5295
    },
    {
      "epoch": 0.7252807450013695,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008799070157952548,
      "loss": 5.0577,
      "step": 5296
    },
    {
      "epoch": 0.7254176937825253,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008798949095509614,
      "loss": 5.0828,
      "step": 5297
    },
    {
      "epoch": 0.7255546425636812,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008798827997440264,
      "loss": 5.0319,
      "step": 5298
    },
    {
      "epoch": 0.7256915913448371,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008798706863745503,
      "loss": 5.0491,
      "step": 5299
    },
    {
      "epoch": 0.7258285401259928,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008798585694426333,
      "loss": 5.0491,
      "step": 5300
    },
    {
      "epoch": 0.7259654889071487,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008798464489483761,
      "loss": 5.0264,
      "step": 5301
    },
    {
      "epoch": 0.7261024376883046,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008798343248918787,
      "loss": 5.0612,
      "step": 5302
    },
    {
      "epoch": 0.7262393864694604,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008798221972732421,
      "loss": 5.0527,
      "step": 5303
    },
    {
      "epoch": 0.7263763352506163,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008798100660925662,
      "loss": 5.0357,
      "step": 5304
    },
    {
      "epoch": 0.7265132840317721,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008797979313499521,
      "loss": 5.0574,
      "step": 5305
    },
    {
      "epoch": 0.726650232812928,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008797857930455,
      "loss": 5.0791,
      "step": 5306
    },
    {
      "epoch": 0.7267871815940838,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008797736511793106,
      "loss": 4.9849,
      "step": 5307
    },
    {
      "epoch": 0.7269241303752396,
      "grad_norm": 0.375,
      "learning_rate": 0.0008797615057514847,
      "loss": 5.1222,
      "step": 5308
    },
    {
      "epoch": 0.7270610791563955,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008797493567621226,
      "loss": 4.9909,
      "step": 5309
    },
    {
      "epoch": 0.7271980279375514,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008797372042113254,
      "loss": 5.0158,
      "step": 5310
    },
    {
      "epoch": 0.7273349767187072,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008797250480991935,
      "loss": 5.0736,
      "step": 5311
    },
    {
      "epoch": 0.7274719254998631,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008797128884258276,
      "loss": 5.0309,
      "step": 5312
    },
    {
      "epoch": 0.727608874281019,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008797007251913286,
      "loss": 5.0343,
      "step": 5313
    },
    {
      "epoch": 0.7277458230621747,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008796885583957975,
      "loss": 5.0069,
      "step": 5314
    },
    {
      "epoch": 0.7278827718433306,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008796763880393347,
      "loss": 5.0446,
      "step": 5315
    },
    {
      "epoch": 0.7280197206244864,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008796642141220415,
      "loss": 5.0057,
      "step": 5316
    },
    {
      "epoch": 0.7281566694056423,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008796520366440186,
      "loss": 5.1196,
      "step": 5317
    },
    {
      "epoch": 0.7282936181867982,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008796398556053666,
      "loss": 5.1018,
      "step": 5318
    },
    {
      "epoch": 0.7284305669679539,
      "grad_norm": 0.453125,
      "learning_rate": 0.000879627671006187,
      "loss": 5.0672,
      "step": 5319
    },
    {
      "epoch": 0.7285675157491098,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008796154828465805,
      "loss": 5.0374,
      "step": 5320
    },
    {
      "epoch": 0.7287044645302657,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000879603291126648,
      "loss": 5.1476,
      "step": 5321
    },
    {
      "epoch": 0.7288414133114215,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008795910958464907,
      "loss": 5.0489,
      "step": 5322
    },
    {
      "epoch": 0.7289783620925774,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008795788970062095,
      "loss": 5.0091,
      "step": 5323
    },
    {
      "epoch": 0.7291153108737333,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008795666946059059,
      "loss": 5.1312,
      "step": 5324
    },
    {
      "epoch": 0.729252259654889,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008795544886456805,
      "loss": 5.0637,
      "step": 5325
    },
    {
      "epoch": 0.7293892084360449,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008795422791256348,
      "loss": 5.1244,
      "step": 5326
    },
    {
      "epoch": 0.7295261572172008,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00087953006604587,
      "loss": 5.022,
      "step": 5327
    },
    {
      "epoch": 0.7296631059983566,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000879517849406487,
      "loss": 5.0239,
      "step": 5328
    },
    {
      "epoch": 0.7298000547795125,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008795056292075872,
      "loss": 5.0712,
      "step": 5329
    },
    {
      "epoch": 0.7299370035606683,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008794934054492721,
      "loss": 5.0348,
      "step": 5330
    },
    {
      "epoch": 0.7300739523418242,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008794811781316427,
      "loss": 5.0421,
      "step": 5331
    },
    {
      "epoch": 0.73021090112298,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008794689472548003,
      "loss": 5.0835,
      "step": 5332
    },
    {
      "epoch": 0.7303478499041358,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008794567128188466,
      "loss": 5.0353,
      "step": 5333
    },
    {
      "epoch": 0.7304847986852917,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008794444748238827,
      "loss": 5.1095,
      "step": 5334
    },
    {
      "epoch": 0.7306217474664476,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008794322332700103,
      "loss": 5.0756,
      "step": 5335
    },
    {
      "epoch": 0.7307586962476034,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008794199881573304,
      "loss": 5.0153,
      "step": 5336
    },
    {
      "epoch": 0.7308956450287593,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008794077394859447,
      "loss": 5.0515,
      "step": 5337
    },
    {
      "epoch": 0.7310325938099151,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008793954872559548,
      "loss": 5.0342,
      "step": 5338
    },
    {
      "epoch": 0.7311695425910709,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008793832314674624,
      "loss": 5.009,
      "step": 5339
    },
    {
      "epoch": 0.7313064913722268,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008793709721205686,
      "loss": 5.0263,
      "step": 5340
    },
    {
      "epoch": 0.7314434401533826,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008793587092153752,
      "loss": 5.0701,
      "step": 5341
    },
    {
      "epoch": 0.7315803889345385,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008793464427519841,
      "loss": 4.9526,
      "step": 5342
    },
    {
      "epoch": 0.7317173377156944,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008793341727304966,
      "loss": 5.0218,
      "step": 5343
    },
    {
      "epoch": 0.7318542864968501,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008793218991510146,
      "loss": 5.0972,
      "step": 5344
    },
    {
      "epoch": 0.731991235278006,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008793096220136395,
      "loss": 5.036,
      "step": 5345
    },
    {
      "epoch": 0.7321281840591619,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008792973413184735,
      "loss": 4.9765,
      "step": 5346
    },
    {
      "epoch": 0.7322651328403177,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000879285057065618,
      "loss": 5.05,
      "step": 5347
    },
    {
      "epoch": 0.7324020816214736,
      "grad_norm": 0.482421875,
      "learning_rate": 0.000879272769255175,
      "loss": 5.0604,
      "step": 5348
    },
    {
      "epoch": 0.7325390304026295,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008792604778872461,
      "loss": 5.0254,
      "step": 5349
    },
    {
      "epoch": 0.7326759791837852,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008792481829619335,
      "loss": 4.997,
      "step": 5350
    },
    {
      "epoch": 0.7328129279649411,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000879235884479339,
      "loss": 5.0513,
      "step": 5351
    },
    {
      "epoch": 0.7329498767460969,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008792235824395643,
      "loss": 5.1115,
      "step": 5352
    },
    {
      "epoch": 0.7330868255272528,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008792112768427115,
      "loss": 5.0446,
      "step": 5353
    },
    {
      "epoch": 0.7332237743084087,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008791989676888826,
      "loss": 4.9998,
      "step": 5354
    },
    {
      "epoch": 0.7333607230895645,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008791866549781796,
      "loss": 5.0464,
      "step": 5355
    },
    {
      "epoch": 0.7334976718707203,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008791743387107045,
      "loss": 5.0548,
      "step": 5356
    },
    {
      "epoch": 0.7336346206518762,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008791620188865595,
      "loss": 5.126,
      "step": 5357
    },
    {
      "epoch": 0.733771569433032,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008791496955058464,
      "loss": 4.9763,
      "step": 5358
    },
    {
      "epoch": 0.7339085182141879,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008791373685686676,
      "loss": 5.053,
      "step": 5359
    },
    {
      "epoch": 0.7340454669953438,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008791250380751252,
      "loss": 5.0889,
      "step": 5360
    },
    {
      "epoch": 0.7341824157764996,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008791127040253213,
      "loss": 5.0398,
      "step": 5361
    },
    {
      "epoch": 0.7343193645576555,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008791003664193582,
      "loss": 5.0082,
      "step": 5362
    },
    {
      "epoch": 0.7344563133388113,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008790880252573381,
      "loss": 5.0421,
      "step": 5363
    },
    {
      "epoch": 0.7345932621199671,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008790756805393632,
      "loss": 4.9691,
      "step": 5364
    },
    {
      "epoch": 0.734730210901123,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008790633322655361,
      "loss": 5.0834,
      "step": 5365
    },
    {
      "epoch": 0.7348671596822788,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008790509804359588,
      "loss": 4.9829,
      "step": 5366
    },
    {
      "epoch": 0.7350041084634347,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008790386250507338,
      "loss": 4.9879,
      "step": 5367
    },
    {
      "epoch": 0.7351410572445906,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008790262661099635,
      "loss": 5.0227,
      "step": 5368
    },
    {
      "epoch": 0.7352780060257463,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008790139036137502,
      "loss": 5.0137,
      "step": 5369
    },
    {
      "epoch": 0.7354149548069022,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008790015375621966,
      "loss": 5.0061,
      "step": 5370
    },
    {
      "epoch": 0.7355519035880581,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000878989167955405,
      "loss": 4.9945,
      "step": 5371
    },
    {
      "epoch": 0.7356888523692139,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008789767947934778,
      "loss": 5.0484,
      "step": 5372
    },
    {
      "epoch": 0.7358258011503698,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008789644180765178,
      "loss": 5.132,
      "step": 5373
    },
    {
      "epoch": 0.7359627499315257,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008789520378046275,
      "loss": 5.0546,
      "step": 5374
    },
    {
      "epoch": 0.7360996987126814,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008789396539779094,
      "loss": 5.0287,
      "step": 5375
    },
    {
      "epoch": 0.7362366474938373,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008789272665964661,
      "loss": 5.1069,
      "step": 5376
    },
    {
      "epoch": 0.7363735962749931,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008789148756604005,
      "loss": 5.027,
      "step": 5377
    },
    {
      "epoch": 0.736510545056149,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000878902481169815,
      "loss": 4.99,
      "step": 5378
    },
    {
      "epoch": 0.7366474938373049,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008788900831248124,
      "loss": 5.0162,
      "step": 5379
    },
    {
      "epoch": 0.7367844426184607,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008788776815254956,
      "loss": 4.9722,
      "step": 5380
    },
    {
      "epoch": 0.7369213913996165,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008788652763719671,
      "loss": 5.0683,
      "step": 5381
    },
    {
      "epoch": 0.7370583401807724,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008788528676643301,
      "loss": 5.1273,
      "step": 5382
    },
    {
      "epoch": 0.7371952889619282,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000878840455402687,
      "loss": 5.0465,
      "step": 5383
    },
    {
      "epoch": 0.7373322377430841,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000878828039587141,
      "loss": 4.9695,
      "step": 5384
    },
    {
      "epoch": 0.73746918652424,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008788156202177946,
      "loss": 5.0861,
      "step": 5385
    },
    {
      "epoch": 0.7376061353053958,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008788031972947512,
      "loss": 5.0215,
      "step": 5386
    },
    {
      "epoch": 0.7377430840865516,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008787907708181135,
      "loss": 5.0707,
      "step": 5387
    },
    {
      "epoch": 0.7378800328677074,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008787783407879845,
      "loss": 5.066,
      "step": 5388
    },
    {
      "epoch": 0.7380169816488633,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008787659072044671,
      "loss": 5.0205,
      "step": 5389
    },
    {
      "epoch": 0.7381539304300192,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008787534700676646,
      "loss": 5.0152,
      "step": 5390
    },
    {
      "epoch": 0.738290879211175,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008787410293776799,
      "loss": 5.0685,
      "step": 5391
    },
    {
      "epoch": 0.7384278279923309,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008787285851346161,
      "loss": 5.0658,
      "step": 5392
    },
    {
      "epoch": 0.7385647767734868,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008787161373385764,
      "loss": 5.0346,
      "step": 5393
    },
    {
      "epoch": 0.7387017255546425,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008787036859896638,
      "loss": 4.9643,
      "step": 5394
    },
    {
      "epoch": 0.7388386743357984,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008786912310879816,
      "loss": 5.049,
      "step": 5395
    },
    {
      "epoch": 0.7389756231169543,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008786787726336331,
      "loss": 5.0911,
      "step": 5396
    },
    {
      "epoch": 0.7391125718981101,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008786663106267215,
      "loss": 5.0223,
      "step": 5397
    },
    {
      "epoch": 0.739249520679266,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008786538450673499,
      "loss": 5.0238,
      "step": 5398
    },
    {
      "epoch": 0.7393864694604217,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008786413759556219,
      "loss": 5.0276,
      "step": 5399
    },
    {
      "epoch": 0.7395234182415776,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008786289032916405,
      "loss": 4.999,
      "step": 5400
    },
    {
      "epoch": 0.7396603670227335,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008786164270755094,
      "loss": 5.0934,
      "step": 5401
    },
    {
      "epoch": 0.7397973158038893,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008786039473073319,
      "loss": 5.087,
      "step": 5402
    },
    {
      "epoch": 0.7399342645850452,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008785914639872111,
      "loss": 5.0562,
      "step": 5403
    },
    {
      "epoch": 0.7400712133662011,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008785789771152509,
      "loss": 5.0865,
      "step": 5404
    },
    {
      "epoch": 0.7402081621473569,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008785664866915545,
      "loss": 5.0165,
      "step": 5405
    },
    {
      "epoch": 0.7403451109285127,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008785539927162255,
      "loss": 4.9692,
      "step": 5406
    },
    {
      "epoch": 0.7404820597096686,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008785414951893676,
      "loss": 5.1238,
      "step": 5407
    },
    {
      "epoch": 0.7406190084908244,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008785289941110841,
      "loss": 5.076,
      "step": 5408
    },
    {
      "epoch": 0.7407559572719803,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008785164894814787,
      "loss": 4.9359,
      "step": 5409
    },
    {
      "epoch": 0.7408929060531362,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008785039813006551,
      "loss": 5.0494,
      "step": 5410
    },
    {
      "epoch": 0.741029854834292,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008784914695687168,
      "loss": 5.0855,
      "step": 5411
    },
    {
      "epoch": 0.7411668036154478,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008784789542857676,
      "loss": 4.9791,
      "step": 5412
    },
    {
      "epoch": 0.7413037523966036,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008784664354519113,
      "loss": 4.9622,
      "step": 5413
    },
    {
      "epoch": 0.7414407011777595,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008784539130672515,
      "loss": 5.0768,
      "step": 5414
    },
    {
      "epoch": 0.7415776499589154,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000878441387131892,
      "loss": 5.0059,
      "step": 5415
    },
    {
      "epoch": 0.7417145987400712,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008784288576459367,
      "loss": 5.0741,
      "step": 5416
    },
    {
      "epoch": 0.7418515475212271,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008784163246094894,
      "loss": 5.1233,
      "step": 5417
    },
    {
      "epoch": 0.741988496302383,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008784037880226538,
      "loss": 5.085,
      "step": 5418
    },
    {
      "epoch": 0.7421254450835387,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000878391247885534,
      "loss": 5.0115,
      "step": 5419
    },
    {
      "epoch": 0.7422623938646946,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008783787041982338,
      "loss": 5.0381,
      "step": 5420
    },
    {
      "epoch": 0.7423993426458505,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008783661569608573,
      "loss": 5.054,
      "step": 5421
    },
    {
      "epoch": 0.7425362914270063,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008783536061735082,
      "loss": 5.0591,
      "step": 5422
    },
    {
      "epoch": 0.7426732402081622,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008783410518362908,
      "loss": 5.0342,
      "step": 5423
    },
    {
      "epoch": 0.7428101889893179,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008783284939493091,
      "loss": 5.0248,
      "step": 5424
    },
    {
      "epoch": 0.7429471377704738,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008783159325126671,
      "loss": 5.0018,
      "step": 5425
    },
    {
      "epoch": 0.7430840865516297,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008783033675264688,
      "loss": 5.0208,
      "step": 5426
    },
    {
      "epoch": 0.7432210353327855,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008782907989908185,
      "loss": 5.0536,
      "step": 5427
    },
    {
      "epoch": 0.7433579841139414,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008782782269058203,
      "loss": 5.1046,
      "step": 5428
    },
    {
      "epoch": 0.7434949328950973,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008782656512715784,
      "loss": 5.0155,
      "step": 5429
    },
    {
      "epoch": 0.743631881676253,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000878253072088197,
      "loss": 5.0645,
      "step": 5430
    },
    {
      "epoch": 0.7437688304574089,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008782404893557803,
      "loss": 5.0989,
      "step": 5431
    },
    {
      "epoch": 0.7439057792385648,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008782279030744327,
      "loss": 5.0338,
      "step": 5432
    },
    {
      "epoch": 0.7440427280197206,
      "grad_norm": 0.375,
      "learning_rate": 0.0008782153132442582,
      "loss": 5.0304,
      "step": 5433
    },
    {
      "epoch": 0.7441796768008765,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008782027198653616,
      "loss": 5.0318,
      "step": 5434
    },
    {
      "epoch": 0.7443166255820323,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008781901229378469,
      "loss": 5.0554,
      "step": 5435
    },
    {
      "epoch": 0.7444535743631882,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008781775224618187,
      "loss": 5.0881,
      "step": 5436
    },
    {
      "epoch": 0.744590523144344,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008781649184373812,
      "loss": 5.1311,
      "step": 5437
    },
    {
      "epoch": 0.7447274719254998,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008781523108646391,
      "loss": 5.0542,
      "step": 5438
    },
    {
      "epoch": 0.7448644207066557,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008781396997436968,
      "loss": 5.0419,
      "step": 5439
    },
    {
      "epoch": 0.7450013694878116,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008781270850746588,
      "loss": 5.1124,
      "step": 5440
    },
    {
      "epoch": 0.7451383182689674,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008781144668576296,
      "loss": 5.083,
      "step": 5441
    },
    {
      "epoch": 0.7452752670501233,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008781018450927139,
      "loss": 4.9879,
      "step": 5442
    },
    {
      "epoch": 0.7454122158312791,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008780892197800159,
      "loss": 5.0712,
      "step": 5443
    },
    {
      "epoch": 0.7455491646124349,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008780765909196409,
      "loss": 5.0108,
      "step": 5444
    },
    {
      "epoch": 0.7456861133935908,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008780639585116929,
      "loss": 5.0234,
      "step": 5445
    },
    {
      "epoch": 0.7458230621747467,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000878051322556277,
      "loss": 4.9648,
      "step": 5446
    },
    {
      "epoch": 0.7459600109559025,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008780386830534977,
      "loss": 5.0157,
      "step": 5447
    },
    {
      "epoch": 0.7460969597370584,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008780260400034599,
      "loss": 5.0509,
      "step": 5448
    },
    {
      "epoch": 0.7462339085182141,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008780133934062684,
      "loss": 5.0757,
      "step": 5449
    },
    {
      "epoch": 0.74637085729937,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008780007432620279,
      "loss": 5.0079,
      "step": 5450
    },
    {
      "epoch": 0.7465078060805259,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000877988089570843,
      "loss": 4.9791,
      "step": 5451
    },
    {
      "epoch": 0.7466447548616817,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008779754323328191,
      "loss": 5.0423,
      "step": 5452
    },
    {
      "epoch": 0.7467817036428376,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008779627715480607,
      "loss": 4.9931,
      "step": 5453
    },
    {
      "epoch": 0.7469186524239935,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008779501072166729,
      "loss": 5.0498,
      "step": 5454
    },
    {
      "epoch": 0.7470556012051492,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008779374393387605,
      "loss": 5.0294,
      "step": 5455
    },
    {
      "epoch": 0.7471925499863051,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008779247679144286,
      "loss": 5.0022,
      "step": 5456
    },
    {
      "epoch": 0.747329498767461,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008779120929437822,
      "loss": 5.083,
      "step": 5457
    },
    {
      "epoch": 0.7474664475486168,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008778994144269263,
      "loss": 5.0369,
      "step": 5458
    },
    {
      "epoch": 0.7476033963297727,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000877886732363966,
      "loss": 5.0641,
      "step": 5459
    },
    {
      "epoch": 0.7477403451109285,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008778740467550064,
      "loss": 4.9981,
      "step": 5460
    },
    {
      "epoch": 0.7478772938920843,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008778613576001526,
      "loss": 5.0678,
      "step": 5461
    },
    {
      "epoch": 0.7480142426732402,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008778486648995098,
      "loss": 5.0702,
      "step": 5462
    },
    {
      "epoch": 0.748151191454396,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008778359686531832,
      "loss": 5.0396,
      "step": 5463
    },
    {
      "epoch": 0.7482881402355519,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008778232688612778,
      "loss": 5.1039,
      "step": 5464
    },
    {
      "epoch": 0.7484250890167078,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008778105655238992,
      "loss": 4.9957,
      "step": 5465
    },
    {
      "epoch": 0.7485620377978636,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008777978586411524,
      "loss": 5.0186,
      "step": 5466
    },
    {
      "epoch": 0.7486989865790195,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008777851482131429,
      "loss": 5.0604,
      "step": 5467
    },
    {
      "epoch": 0.7488359353601753,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008777724342399759,
      "loss": 5.122,
      "step": 5468
    },
    {
      "epoch": 0.7489728841413311,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008777597167217568,
      "loss": 5.0139,
      "step": 5469
    },
    {
      "epoch": 0.749109832922487,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000877746995658591,
      "loss": 4.982,
      "step": 5470
    },
    {
      "epoch": 0.7492467817036428,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000877734271050584,
      "loss": 4.9876,
      "step": 5471
    },
    {
      "epoch": 0.7493837304847987,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000877721542897841,
      "loss": 5.0358,
      "step": 5472
    },
    {
      "epoch": 0.7495206792659546,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008777088112004678,
      "loss": 5.0433,
      "step": 5473
    },
    {
      "epoch": 0.7496576280471103,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008776960759585697,
      "loss": 5.0401,
      "step": 5474
    },
    {
      "epoch": 0.7497945768282662,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008776833371722522,
      "loss": 5.0431,
      "step": 5475
    },
    {
      "epoch": 0.7499315256094221,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008776705948416212,
      "loss": 4.9793,
      "step": 5476
    },
    {
      "epoch": 0.7500684743905779,
      "grad_norm": 0.515625,
      "learning_rate": 0.000877657848966782,
      "loss": 5.0265,
      "step": 5477
    },
    {
      "epoch": 0.7502054231717338,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008776450995478402,
      "loss": 4.8894,
      "step": 5478
    },
    {
      "epoch": 0.7503423719528897,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008776323465849017,
      "loss": 5.0073,
      "step": 5479
    },
    {
      "epoch": 0.7504793207340454,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008776195900780719,
      "loss": 5.0624,
      "step": 5480
    },
    {
      "epoch": 0.7506162695152013,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008776068300274567,
      "loss": 5.048,
      "step": 5481
    },
    {
      "epoch": 0.7507532182963572,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008775940664331618,
      "loss": 5.017,
      "step": 5482
    },
    {
      "epoch": 0.750890167077513,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008775812992952931,
      "loss": 5.0346,
      "step": 5483
    },
    {
      "epoch": 0.7510271158586689,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008775685286139561,
      "loss": 4.9506,
      "step": 5484
    },
    {
      "epoch": 0.7511640646398247,
      "grad_norm": 0.5,
      "learning_rate": 0.000877555754389257,
      "loss": 5.0487,
      "step": 5485
    },
    {
      "epoch": 0.7513010134209805,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008775429766213013,
      "loss": 4.9889,
      "step": 5486
    },
    {
      "epoch": 0.7514379622021364,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008775301953101953,
      "loss": 4.9958,
      "step": 5487
    },
    {
      "epoch": 0.7515749109832922,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008775174104560446,
      "loss": 5.017,
      "step": 5488
    },
    {
      "epoch": 0.7517118597644481,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008775046220589551,
      "loss": 4.9848,
      "step": 5489
    },
    {
      "epoch": 0.751848808545604,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008774918301190331,
      "loss": 5.0416,
      "step": 5490
    },
    {
      "epoch": 0.7519857573267598,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008774790346363844,
      "loss": 5.0909,
      "step": 5491
    },
    {
      "epoch": 0.7521227061079157,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0008774662356111151,
      "loss": 5.0989,
      "step": 5492
    },
    {
      "epoch": 0.7522596548890715,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008774534330433312,
      "loss": 5.0215,
      "step": 5493
    },
    {
      "epoch": 0.7523966036702273,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0008774406269331388,
      "loss": 5.0095,
      "step": 5494
    },
    {
      "epoch": 0.7525335524513832,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008774278172806442,
      "loss": 5.0004,
      "step": 5495
    },
    {
      "epoch": 0.752670501232539,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008774150040859532,
      "loss": 5.0949,
      "step": 5496
    },
    {
      "epoch": 0.7528074500136949,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008774021873491724,
      "loss": 4.9995,
      "step": 5497
    },
    {
      "epoch": 0.7529443987948508,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008773893670704078,
      "loss": 5.0313,
      "step": 5498
    },
    {
      "epoch": 0.7530813475760065,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008773765432497657,
      "loss": 5.0703,
      "step": 5499
    },
    {
      "epoch": 0.7532182963571624,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008773637158873524,
      "loss": 4.9974,
      "step": 5500
    },
    {
      "epoch": 0.7533552451383183,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008773508849832739,
      "loss": 5.076,
      "step": 5501
    },
    {
      "epoch": 0.7534921939194741,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008773380505376371,
      "loss": 5.0774,
      "step": 5502
    },
    {
      "epoch": 0.75362914270063,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008773252125505479,
      "loss": 5.0263,
      "step": 5503
    },
    {
      "epoch": 0.7537660914817859,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008773123710221128,
      "loss": 5.0619,
      "step": 5504
    },
    {
      "epoch": 0.7539030402629416,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008772995259524383,
      "loss": 4.998,
      "step": 5505
    },
    {
      "epoch": 0.7540399890440975,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008772866773416308,
      "loss": 5.0506,
      "step": 5506
    },
    {
      "epoch": 0.7541769378252533,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008772738251897967,
      "loss": 5.0167,
      "step": 5507
    },
    {
      "epoch": 0.7543138866064092,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008772609694970427,
      "loss": 5.0734,
      "step": 5508
    },
    {
      "epoch": 0.7544508353875651,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008772481102634753,
      "loss": 5.0332,
      "step": 5509
    },
    {
      "epoch": 0.7545877841687209,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008772352474892009,
      "loss": 5.0713,
      "step": 5510
    },
    {
      "epoch": 0.7547247329498767,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008772223811743262,
      "loss": 5.027,
      "step": 5511
    },
    {
      "epoch": 0.7548616817310326,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008772095113189579,
      "loss": 4.9902,
      "step": 5512
    },
    {
      "epoch": 0.7549986305121884,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008771966379232026,
      "loss": 5.0241,
      "step": 5513
    },
    {
      "epoch": 0.7551355792933443,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008771837609871669,
      "loss": 5.1049,
      "step": 5514
    },
    {
      "epoch": 0.7552725280745002,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008771708805109575,
      "loss": 5.0002,
      "step": 5515
    },
    {
      "epoch": 0.755409476855656,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008771579964946813,
      "loss": 5.0521,
      "step": 5516
    },
    {
      "epoch": 0.7555464256368118,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000877145108938445,
      "loss": 4.983,
      "step": 5517
    },
    {
      "epoch": 0.7556833744179677,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008771322178423554,
      "loss": 5.1287,
      "step": 5518
    },
    {
      "epoch": 0.7558203231991235,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008771193232065193,
      "loss": 5.0739,
      "step": 5519
    },
    {
      "epoch": 0.7559572719802794,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008771064250310437,
      "loss": 5.0174,
      "step": 5520
    },
    {
      "epoch": 0.7560942207614352,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008770935233160352,
      "loss": 5.0006,
      "step": 5521
    },
    {
      "epoch": 0.7562311695425911,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000877080618061601,
      "loss": 5.0676,
      "step": 5522
    },
    {
      "epoch": 0.756368118323747,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008770677092678479,
      "loss": 5.076,
      "step": 5523
    },
    {
      "epoch": 0.7565050671049027,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000877054796934883,
      "loss": 5.0291,
      "step": 5524
    },
    {
      "epoch": 0.7566420158860586,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008770418810628133,
      "loss": 4.9588,
      "step": 5525
    },
    {
      "epoch": 0.7567789646672145,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008770289616517456,
      "loss": 5.0009,
      "step": 5526
    },
    {
      "epoch": 0.7569159134483703,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008770160387017872,
      "loss": 5.0428,
      "step": 5527
    },
    {
      "epoch": 0.7570528622295262,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008770031122130453,
      "loss": 5.052,
      "step": 5528
    },
    {
      "epoch": 0.7571898110106821,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008769901821856267,
      "loss": 5.138,
      "step": 5529
    },
    {
      "epoch": 0.7573267597918378,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008769772486196387,
      "loss": 5.1299,
      "step": 5530
    },
    {
      "epoch": 0.7574637085729937,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008769643115151886,
      "loss": 4.9932,
      "step": 5531
    },
    {
      "epoch": 0.7576006573541495,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008769513708723834,
      "loss": 4.9813,
      "step": 5532
    },
    {
      "epoch": 0.7577376061353054,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008769384266913304,
      "loss": 5.0651,
      "step": 5533
    },
    {
      "epoch": 0.7578745549164613,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000876925478972137,
      "loss": 5.0198,
      "step": 5534
    },
    {
      "epoch": 0.758011503697617,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008769125277149106,
      "loss": 5.0498,
      "step": 5535
    },
    {
      "epoch": 0.7581484524787729,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008768995729197582,
      "loss": 5.0503,
      "step": 5536
    },
    {
      "epoch": 0.7582854012599288,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008768866145867873,
      "loss": 5.0443,
      "step": 5537
    },
    {
      "epoch": 0.7584223500410846,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008768736527161054,
      "loss": 5.006,
      "step": 5538
    },
    {
      "epoch": 0.7585592988222405,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008768606873078197,
      "loss": 4.997,
      "step": 5539
    },
    {
      "epoch": 0.7586962476033964,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008768477183620379,
      "loss": 5.02,
      "step": 5540
    },
    {
      "epoch": 0.7588331963845522,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008768347458788672,
      "loss": 5.0447,
      "step": 5541
    },
    {
      "epoch": 0.758970145165708,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008768217698584153,
      "loss": 4.9788,
      "step": 5542
    },
    {
      "epoch": 0.7591070939468638,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008768087903007896,
      "loss": 5.0605,
      "step": 5543
    },
    {
      "epoch": 0.7592440427280197,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008767958072060978,
      "loss": 5.0049,
      "step": 5544
    },
    {
      "epoch": 0.7593809915091756,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008767828205744476,
      "loss": 5.0385,
      "step": 5545
    },
    {
      "epoch": 0.7595179402903314,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008767698304059463,
      "loss": 5.0084,
      "step": 5546
    },
    {
      "epoch": 0.7596548890714873,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008767568367007018,
      "loss": 4.9973,
      "step": 5547
    },
    {
      "epoch": 0.7597918378526431,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008767438394588217,
      "loss": 5.0157,
      "step": 5548
    },
    {
      "epoch": 0.7599287866337989,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008767308386804137,
      "loss": 4.9986,
      "step": 5549
    },
    {
      "epoch": 0.7600657354149548,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008767178343655856,
      "loss": 5.0765,
      "step": 5550
    },
    {
      "epoch": 0.7602026841961107,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008767048265144451,
      "loss": 5.0853,
      "step": 5551
    },
    {
      "epoch": 0.7603396329772665,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008766918151271,
      "loss": 4.9931,
      "step": 5552
    },
    {
      "epoch": 0.7604765817584224,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008766788002036581,
      "loss": 5.01,
      "step": 5553
    },
    {
      "epoch": 0.7606135305395783,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008766657817442273,
      "loss": 5.1067,
      "step": 5554
    },
    {
      "epoch": 0.760750479320734,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008766527597489157,
      "loss": 4.9977,
      "step": 5555
    },
    {
      "epoch": 0.7608874281018899,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008766397342178309,
      "loss": 5.0977,
      "step": 5556
    },
    {
      "epoch": 0.7610243768830457,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008766267051510809,
      "loss": 5.0234,
      "step": 5557
    },
    {
      "epoch": 0.7611613256642016,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008766136725487737,
      "loss": 5.0498,
      "step": 5558
    },
    {
      "epoch": 0.7612982744453575,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008766006364110175,
      "loss": 5.0432,
      "step": 5559
    },
    {
      "epoch": 0.7614352232265132,
      "grad_norm": 0.40625,
      "learning_rate": 0.00087658759673792,
      "loss": 5.02,
      "step": 5560
    },
    {
      "epoch": 0.7615721720076691,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008765745535295895,
      "loss": 5.1226,
      "step": 5561
    },
    {
      "epoch": 0.761709120788825,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008765615067861342,
      "loss": 4.9663,
      "step": 5562
    },
    {
      "epoch": 0.7618460695699808,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008765484565076619,
      "loss": 5.113,
      "step": 5563
    },
    {
      "epoch": 0.7619830183511367,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008765354026942808,
      "loss": 5.0143,
      "step": 5564
    },
    {
      "epoch": 0.7621199671322926,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008765223453460994,
      "loss": 5.0548,
      "step": 5565
    },
    {
      "epoch": 0.7622569159134484,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008765092844632256,
      "loss": 4.9892,
      "step": 5566
    },
    {
      "epoch": 0.7623938646946042,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008764962200457676,
      "loss": 5.0039,
      "step": 5567
    },
    {
      "epoch": 0.76253081347576,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008764831520938339,
      "loss": 5.0323,
      "step": 5568
    },
    {
      "epoch": 0.7626677622569159,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008764700806075328,
      "loss": 4.9647,
      "step": 5569
    },
    {
      "epoch": 0.7628047110380718,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008764570055869723,
      "loss": 5.0455,
      "step": 5570
    },
    {
      "epoch": 0.7629416598192276,
      "grad_norm": 0.443359375,
      "learning_rate": 0.000876443927032261,
      "loss": 5.0771,
      "step": 5571
    },
    {
      "epoch": 0.7630786086003835,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008764308449435074,
      "loss": 5.0186,
      "step": 5572
    },
    {
      "epoch": 0.7632155573815393,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008764177593208198,
      "loss": 4.9574,
      "step": 5573
    },
    {
      "epoch": 0.7633525061626951,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008764046701643065,
      "loss": 4.9966,
      "step": 5574
    },
    {
      "epoch": 0.763489454943851,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008763915774740762,
      "loss": 5.0754,
      "step": 5575
    },
    {
      "epoch": 0.7636264037250069,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008763784812502372,
      "loss": 5.028,
      "step": 5576
    },
    {
      "epoch": 0.7637633525061627,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008763653814928981,
      "loss": 5.079,
      "step": 5577
    },
    {
      "epoch": 0.7639003012873186,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008763522782021675,
      "loss": 5.0145,
      "step": 5578
    },
    {
      "epoch": 0.7640372500684743,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000876339171378154,
      "loss": 5.0087,
      "step": 5579
    },
    {
      "epoch": 0.7641741988496302,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008763260610209662,
      "loss": 5.0282,
      "step": 5580
    },
    {
      "epoch": 0.7643111476307861,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008763129471307127,
      "loss": 5.0355,
      "step": 5581
    },
    {
      "epoch": 0.7644480964119419,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008762998297075023,
      "loss": 5.0242,
      "step": 5582
    },
    {
      "epoch": 0.7645850451930978,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008762867087514436,
      "loss": 5.0682,
      "step": 5583
    },
    {
      "epoch": 0.7647219939742537,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008762735842626454,
      "loss": 5.0766,
      "step": 5584
    },
    {
      "epoch": 0.7648589427554094,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008762604562412165,
      "loss": 5.1066,
      "step": 5585
    },
    {
      "epoch": 0.7649958915365653,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008762473246872655,
      "loss": 5.0046,
      "step": 5586
    },
    {
      "epoch": 0.7651328403177212,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008762341896009013,
      "loss": 5.0446,
      "step": 5587
    },
    {
      "epoch": 0.765269789098877,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008762210509822329,
      "loss": 5.048,
      "step": 5588
    },
    {
      "epoch": 0.7654067378800329,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008762079088313692,
      "loss": 5.1103,
      "step": 5589
    },
    {
      "epoch": 0.7655436866611887,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008761947631484188,
      "loss": 4.9623,
      "step": 5590
    },
    {
      "epoch": 0.7656806354423445,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008761816139334908,
      "loss": 5.0277,
      "step": 5591
    },
    {
      "epoch": 0.7658175842235004,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008761684611866944,
      "loss": 5.0106,
      "step": 5592
    },
    {
      "epoch": 0.7659545330046562,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008761553049081383,
      "loss": 5.0777,
      "step": 5593
    },
    {
      "epoch": 0.7660914817858121,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008761421450979317,
      "loss": 5.0021,
      "step": 5594
    },
    {
      "epoch": 0.766228430566968,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008761289817561836,
      "loss": 4.9981,
      "step": 5595
    },
    {
      "epoch": 0.7663653793481238,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008761158148830031,
      "loss": 4.9554,
      "step": 5596
    },
    {
      "epoch": 0.7665023281292797,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008761026444784992,
      "loss": 4.9395,
      "step": 5597
    },
    {
      "epoch": 0.7666392769104355,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008760894705427812,
      "loss": 5.0552,
      "step": 5598
    },
    {
      "epoch": 0.7667762256915913,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008760762930759583,
      "loss": 4.9322,
      "step": 5599
    },
    {
      "epoch": 0.7669131744727472,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008760631120781396,
      "loss": 5.0951,
      "step": 5600
    },
    {
      "epoch": 0.7670501232539031,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008760499275494343,
      "loss": 5.0415,
      "step": 5601
    },
    {
      "epoch": 0.7671870720350589,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008760367394899518,
      "loss": 5.0353,
      "step": 5602
    },
    {
      "epoch": 0.7673240208162148,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008760235478998013,
      "loss": 4.9547,
      "step": 5603
    },
    {
      "epoch": 0.7674609695973705,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008760103527790922,
      "loss": 4.9917,
      "step": 5604
    },
    {
      "epoch": 0.7675979183785264,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008759971541279336,
      "loss": 5.0303,
      "step": 5605
    },
    {
      "epoch": 0.7677348671596823,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008759839519464352,
      "loss": 5.018,
      "step": 5606
    },
    {
      "epoch": 0.7678718159408381,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008759707462347063,
      "loss": 4.9919,
      "step": 5607
    },
    {
      "epoch": 0.768008764721994,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008759575369928563,
      "loss": 4.996,
      "step": 5608
    },
    {
      "epoch": 0.7681457135031499,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008759443242209947,
      "loss": 5.032,
      "step": 5609
    },
    {
      "epoch": 0.7682826622843056,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000875931107919231,
      "loss": 5.0366,
      "step": 5610
    },
    {
      "epoch": 0.7684196110654615,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008759178880876747,
      "loss": 5.0581,
      "step": 5611
    },
    {
      "epoch": 0.7685565598466174,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008759046647264354,
      "loss": 4.9738,
      "step": 5612
    },
    {
      "epoch": 0.7686935086277732,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008758914378356226,
      "loss": 5.0404,
      "step": 5613
    },
    {
      "epoch": 0.7688304574089291,
      "grad_norm": 0.49609375,
      "learning_rate": 0.000875878207415346,
      "loss": 5.0106,
      "step": 5614
    },
    {
      "epoch": 0.7689674061900849,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008758649734657151,
      "loss": 4.9827,
      "step": 5615
    },
    {
      "epoch": 0.7691043549712407,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008758517359868399,
      "loss": 5.0731,
      "step": 5616
    },
    {
      "epoch": 0.7692413037523966,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008758384949788298,
      "loss": 5.0006,
      "step": 5617
    },
    {
      "epoch": 0.7693782525335524,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008758252504417946,
      "loss": 4.965,
      "step": 5618
    },
    {
      "epoch": 0.7695152013147083,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008758120023758441,
      "loss": 5.0261,
      "step": 5619
    },
    {
      "epoch": 0.7696521500958642,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000875798750781088,
      "loss": 5.0863,
      "step": 5620
    },
    {
      "epoch": 0.76978909887702,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008757854956576364,
      "loss": 5.0281,
      "step": 5621
    },
    {
      "epoch": 0.7699260476581758,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008757722370055987,
      "loss": 5.037,
      "step": 5622
    },
    {
      "epoch": 0.7700629964393317,
      "grad_norm": 0.46875,
      "learning_rate": 0.000875758974825085,
      "loss": 5.0202,
      "step": 5623
    },
    {
      "epoch": 0.7701999452204875,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008757457091162054,
      "loss": 4.9431,
      "step": 5624
    },
    {
      "epoch": 0.7703368940016434,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008757324398790696,
      "loss": 4.947,
      "step": 5625
    },
    {
      "epoch": 0.7704738427827992,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008757191671137876,
      "loss": 5.0104,
      "step": 5626
    },
    {
      "epoch": 0.7706107915639551,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008757058908204694,
      "loss": 5.0261,
      "step": 5627
    },
    {
      "epoch": 0.770747740345111,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008756926109992251,
      "loss": 5.0037,
      "step": 5628
    },
    {
      "epoch": 0.7708846891262667,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008756793276501647,
      "loss": 5.0612,
      "step": 5629
    },
    {
      "epoch": 0.7710216379074226,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008756660407733982,
      "loss": 5.0366,
      "step": 5630
    },
    {
      "epoch": 0.7711585866885785,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008756527503690359,
      "loss": 5.058,
      "step": 5631
    },
    {
      "epoch": 0.7712955354697343,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008756394564371877,
      "loss": 5.0093,
      "step": 5632
    },
    {
      "epoch": 0.7714324842508902,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008756261589779639,
      "loss": 5.0118,
      "step": 5633
    },
    {
      "epoch": 0.7715694330320461,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008756128579914749,
      "loss": 5.1309,
      "step": 5634
    },
    {
      "epoch": 0.7717063818132018,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008755995534778305,
      "loss": 5.0489,
      "step": 5635
    },
    {
      "epoch": 0.7718433305943577,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008755862454371413,
      "loss": 4.9845,
      "step": 5636
    },
    {
      "epoch": 0.7719802793755136,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008755729338695174,
      "loss": 4.9987,
      "step": 5637
    },
    {
      "epoch": 0.7721172281566694,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008755596187750693,
      "loss": 5.0272,
      "step": 5638
    },
    {
      "epoch": 0.7722541769378253,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008755463001539072,
      "loss": 5.0347,
      "step": 5639
    },
    {
      "epoch": 0.772391125718981,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008755329780061414,
      "loss": 5.086,
      "step": 5640
    },
    {
      "epoch": 0.7725280745001369,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008755196523318826,
      "loss": 5.0533,
      "step": 5641
    },
    {
      "epoch": 0.7726650232812928,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008755063231312408,
      "loss": 4.9951,
      "step": 5642
    },
    {
      "epoch": 0.7728019720624486,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008754929904043269,
      "loss": 4.9612,
      "step": 5643
    },
    {
      "epoch": 0.7729389208436045,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008754796541512511,
      "loss": 5.0388,
      "step": 5644
    },
    {
      "epoch": 0.7730758696247604,
      "grad_norm": 0.4609375,
      "learning_rate": 0.000875466314372124,
      "loss": 5.0162,
      "step": 5645
    },
    {
      "epoch": 0.7732128184059162,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008754529710670564,
      "loss": 5.0463,
      "step": 5646
    },
    {
      "epoch": 0.773349767187072,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008754396242361584,
      "loss": 4.9517,
      "step": 5647
    },
    {
      "epoch": 0.7734867159682279,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008754262738795409,
      "loss": 5.0043,
      "step": 5648
    },
    {
      "epoch": 0.7736236647493837,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008754129199973145,
      "loss": 5.0456,
      "step": 5649
    },
    {
      "epoch": 0.7737606135305396,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008753995625895899,
      "loss": 5.0687,
      "step": 5650
    },
    {
      "epoch": 0.7738975623116954,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008753862016564778,
      "loss": 5.0486,
      "step": 5651
    },
    {
      "epoch": 0.7740345110928513,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008753728371980889,
      "loss": 5.0171,
      "step": 5652
    },
    {
      "epoch": 0.7741714598740072,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008753594692145338,
      "loss": 5.0572,
      "step": 5653
    },
    {
      "epoch": 0.7743084086551629,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008753460977059235,
      "loss": 4.9994,
      "step": 5654
    },
    {
      "epoch": 0.7744453574363188,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008753327226723688,
      "loss": 5.0474,
      "step": 5655
    },
    {
      "epoch": 0.7745823062174747,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008753193441139804,
      "loss": 4.9438,
      "step": 5656
    },
    {
      "epoch": 0.7747192549986305,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008753059620308692,
      "loss": 5.0245,
      "step": 5657
    },
    {
      "epoch": 0.7748562037797864,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008752925764231461,
      "loss": 4.987,
      "step": 5658
    },
    {
      "epoch": 0.7749931525609423,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008752791872909222,
      "loss": 5.0079,
      "step": 5659
    },
    {
      "epoch": 0.775130101342098,
      "grad_norm": 0.5,
      "learning_rate": 0.0008752657946343084,
      "loss": 5.0174,
      "step": 5660
    },
    {
      "epoch": 0.7752670501232539,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008752523984534155,
      "loss": 5.0059,
      "step": 5661
    },
    {
      "epoch": 0.7754039989044097,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008752389987483547,
      "loss": 4.9955,
      "step": 5662
    },
    {
      "epoch": 0.7755409476855656,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008752255955192371,
      "loss": 5.0914,
      "step": 5663
    },
    {
      "epoch": 0.7756778964667215,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008752121887661735,
      "loss": 4.9956,
      "step": 5664
    },
    {
      "epoch": 0.7758148452478772,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008751987784892751,
      "loss": 4.9735,
      "step": 5665
    },
    {
      "epoch": 0.7759517940290331,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008751853646886533,
      "loss": 4.9753,
      "step": 5666
    },
    {
      "epoch": 0.776088742810189,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008751719473644189,
      "loss": 4.9648,
      "step": 5667
    },
    {
      "epoch": 0.7762256915913448,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008751585265166833,
      "loss": 5.0096,
      "step": 5668
    },
    {
      "epoch": 0.7763626403725007,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008751451021455578,
      "loss": 4.9947,
      "step": 5669
    },
    {
      "epoch": 0.7764995891536566,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008751316742511534,
      "loss": 5.0383,
      "step": 5670
    },
    {
      "epoch": 0.7766365379348124,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008751182428335815,
      "loss": 5.1052,
      "step": 5671
    },
    {
      "epoch": 0.7767734867159682,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008751048078929535,
      "loss": 5.0032,
      "step": 5672
    },
    {
      "epoch": 0.7769104354971241,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008750913694293807,
      "loss": 5.0145,
      "step": 5673
    },
    {
      "epoch": 0.7770473842782799,
      "grad_norm": 0.5,
      "learning_rate": 0.0008750779274429743,
      "loss": 5.0588,
      "step": 5674
    },
    {
      "epoch": 0.7771843330594358,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000875064481933846,
      "loss": 5.0354,
      "step": 5675
    },
    {
      "epoch": 0.7773212818405916,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008750510329021069,
      "loss": 4.9894,
      "step": 5676
    },
    {
      "epoch": 0.7774582306217475,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008750375803478686,
      "loss": 4.98,
      "step": 5677
    },
    {
      "epoch": 0.7775951794029033,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008750241242712427,
      "loss": 4.9486,
      "step": 5678
    },
    {
      "epoch": 0.7777321281840591,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008750106646723405,
      "loss": 4.9394,
      "step": 5679
    },
    {
      "epoch": 0.777869076965215,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008749972015512737,
      "loss": 5.073,
      "step": 5680
    },
    {
      "epoch": 0.7780060257463709,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008749837349081538,
      "loss": 4.9578,
      "step": 5681
    },
    {
      "epoch": 0.7781429745275267,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008749702647430924,
      "loss": 5.0015,
      "step": 5682
    },
    {
      "epoch": 0.7782799233086826,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008749567910562011,
      "loss": 5.037,
      "step": 5683
    },
    {
      "epoch": 0.7784168720898385,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008749433138475918,
      "loss": 5.0188,
      "step": 5684
    },
    {
      "epoch": 0.7785538208709942,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008749298331173759,
      "loss": 5.0387,
      "step": 5685
    },
    {
      "epoch": 0.7786907696521501,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008749163488656652,
      "loss": 5.1267,
      "step": 5686
    },
    {
      "epoch": 0.7788277184333059,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008749028610925715,
      "loss": 5.0505,
      "step": 5687
    },
    {
      "epoch": 0.7789646672144618,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008748893697982064,
      "loss": 5.079,
      "step": 5688
    },
    {
      "epoch": 0.7791016159956177,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008748758749826819,
      "loss": 5.0005,
      "step": 5689
    },
    {
      "epoch": 0.7792385647767734,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008748623766461099,
      "loss": 5.0553,
      "step": 5690
    },
    {
      "epoch": 0.7793755135579293,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008748488747886021,
      "loss": 5.0111,
      "step": 5691
    },
    {
      "epoch": 0.7795124623390852,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008748353694102704,
      "loss": 5.0666,
      "step": 5692
    },
    {
      "epoch": 0.779649411120241,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008748218605112269,
      "loss": 5.0534,
      "step": 5693
    },
    {
      "epoch": 0.7797863599013969,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008748083480915833,
      "loss": 5.0489,
      "step": 5694
    },
    {
      "epoch": 0.7799233086825528,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008747948321514516,
      "loss": 5.0715,
      "step": 5695
    },
    {
      "epoch": 0.7800602574637086,
      "grad_norm": 0.453125,
      "learning_rate": 0.000874781312690944,
      "loss": 5.0577,
      "step": 5696
    },
    {
      "epoch": 0.7801972062448644,
      "grad_norm": 0.5,
      "learning_rate": 0.0008747677897101725,
      "loss": 4.9243,
      "step": 5697
    },
    {
      "epoch": 0.7803341550260202,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000874754263209249,
      "loss": 4.976,
      "step": 5698
    },
    {
      "epoch": 0.7804711038071761,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008747407331882858,
      "loss": 5.0634,
      "step": 5699
    },
    {
      "epoch": 0.780608052588332,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008747271996473949,
      "loss": 4.9938,
      "step": 5700
    },
    {
      "epoch": 0.7807450013694878,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008747136625866885,
      "loss": 5.0081,
      "step": 5701
    },
    {
      "epoch": 0.7808819501506437,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008747001220062787,
      "loss": 5.0259,
      "step": 5702
    },
    {
      "epoch": 0.7810188989317995,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008746865779062778,
      "loss": 5.005,
      "step": 5703
    },
    {
      "epoch": 0.7811558477129553,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008746730302867982,
      "loss": 5.0068,
      "step": 5704
    },
    {
      "epoch": 0.7812927964941112,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008746594791479517,
      "loss": 5.0302,
      "step": 5705
    },
    {
      "epoch": 0.7814297452752671,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008746459244898512,
      "loss": 5.0229,
      "step": 5706
    },
    {
      "epoch": 0.7815666940564229,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008746323663126086,
      "loss": 4.9849,
      "step": 5707
    },
    {
      "epoch": 0.7817036428375788,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008746188046163364,
      "loss": 5.021,
      "step": 5708
    },
    {
      "epoch": 0.7818405916187346,
      "grad_norm": 0.375,
      "learning_rate": 0.000874605239401147,
      "loss": 5.0279,
      "step": 5709
    },
    {
      "epoch": 0.7819775403998904,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008745916706671528,
      "loss": 5.0031,
      "step": 5710
    },
    {
      "epoch": 0.7821144891810463,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008745780984144662,
      "loss": 4.9894,
      "step": 5711
    },
    {
      "epoch": 0.7822514379622021,
      "grad_norm": 0.375,
      "learning_rate": 0.0008745645226432,
      "loss": 4.9921,
      "step": 5712
    },
    {
      "epoch": 0.782388386743358,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008745509433534661,
      "loss": 5.068,
      "step": 5713
    },
    {
      "epoch": 0.7825253355245139,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008745373605453774,
      "loss": 4.9941,
      "step": 5714
    },
    {
      "epoch": 0.7826622843056696,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008745237742190466,
      "loss": 4.9901,
      "step": 5715
    },
    {
      "epoch": 0.7827992330868255,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000874510184374586,
      "loss": 5.0801,
      "step": 5716
    },
    {
      "epoch": 0.7829361818679814,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008744965910121084,
      "loss": 5.0232,
      "step": 5717
    },
    {
      "epoch": 0.7830731306491372,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008744829941317264,
      "loss": 4.9506,
      "step": 5718
    },
    {
      "epoch": 0.7832100794302931,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008744693937335526,
      "loss": 5.0346,
      "step": 5719
    },
    {
      "epoch": 0.783347028211449,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008744557898177,
      "loss": 5.1192,
      "step": 5720
    },
    {
      "epoch": 0.7834839769926047,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008744421823842809,
      "loss": 5.017,
      "step": 5721
    },
    {
      "epoch": 0.7836209257737606,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008744285714334083,
      "loss": 5.0014,
      "step": 5722
    },
    {
      "epoch": 0.7837578745549164,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008744149569651951,
      "loss": 5.0379,
      "step": 5723
    },
    {
      "epoch": 0.7838948233360723,
      "grad_norm": 0.46875,
      "learning_rate": 0.000874401338979754,
      "loss": 5.0203,
      "step": 5724
    },
    {
      "epoch": 0.7840317721172282,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008743877174771979,
      "loss": 5.013,
      "step": 5725
    },
    {
      "epoch": 0.784168720898384,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008743740924576396,
      "loss": 5.0696,
      "step": 5726
    },
    {
      "epoch": 0.7843056696795399,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008743604639211921,
      "loss": 5.1108,
      "step": 5727
    },
    {
      "epoch": 0.7844426184606957,
      "grad_norm": 0.75,
      "learning_rate": 0.0008743468318679683,
      "loss": 5.0095,
      "step": 5728
    },
    {
      "epoch": 0.7845795672418515,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008743331962980811,
      "loss": 5.059,
      "step": 5729
    },
    {
      "epoch": 0.7847165160230074,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008743195572116437,
      "loss": 5.0431,
      "step": 5730
    },
    {
      "epoch": 0.7848534648041633,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008743059146087691,
      "loss": 5.0391,
      "step": 5731
    },
    {
      "epoch": 0.7849904135853191,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008742922684895701,
      "loss": 5.0397,
      "step": 5732
    },
    {
      "epoch": 0.785127362366475,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00087427861885416,
      "loss": 5.1107,
      "step": 5733
    },
    {
      "epoch": 0.7852643111476307,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000874264965702652,
      "loss": 5.0782,
      "step": 5734
    },
    {
      "epoch": 0.7854012599287866,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000874251309035159,
      "loss": 5.0492,
      "step": 5735
    },
    {
      "epoch": 0.7855382087099425,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008742376488517945,
      "loss": 5.0284,
      "step": 5736
    },
    {
      "epoch": 0.7856751574910983,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008742239851526713,
      "loss": 4.9898,
      "step": 5737
    },
    {
      "epoch": 0.7858121062722542,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008742103179379029,
      "loss": 5.0359,
      "step": 5738
    },
    {
      "epoch": 0.7859490550534101,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008741966472076026,
      "loss": 4.9982,
      "step": 5739
    },
    {
      "epoch": 0.7860860038345658,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008741829729618835,
      "loss": 5.0479,
      "step": 5740
    },
    {
      "epoch": 0.7862229526157217,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008741692952008591,
      "loss": 4.963,
      "step": 5741
    },
    {
      "epoch": 0.7863599013968776,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008741556139246426,
      "loss": 5.0119,
      "step": 5742
    },
    {
      "epoch": 0.7864968501780334,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008741419291333476,
      "loss": 5.1294,
      "step": 5743
    },
    {
      "epoch": 0.7866337989591893,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008741282408270872,
      "loss": 5.0385,
      "step": 5744
    },
    {
      "epoch": 0.7867707477403451,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008741145490059752,
      "loss": 5.0202,
      "step": 5745
    },
    {
      "epoch": 0.7869076965215009,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008741008536701248,
      "loss": 4.9459,
      "step": 5746
    },
    {
      "epoch": 0.7870446453026568,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008740871548196494,
      "loss": 5.0935,
      "step": 5747
    },
    {
      "epoch": 0.7871815940838126,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008740734524546629,
      "loss": 4.9555,
      "step": 5748
    },
    {
      "epoch": 0.7873185428649685,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008740597465752785,
      "loss": 5.0276,
      "step": 5749
    },
    {
      "epoch": 0.7874554916461244,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00087404603718161,
      "loss": 5.0654,
      "step": 5750
    },
    {
      "epoch": 0.7875924404272802,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000874032324273771,
      "loss": 5.0726,
      "step": 5751
    },
    {
      "epoch": 0.787729389208436,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000874018607851875,
      "loss": 5.1228,
      "step": 5752
    },
    {
      "epoch": 0.7878663379895919,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008740048879160358,
      "loss": 5.0752,
      "step": 5753
    },
    {
      "epoch": 0.7880032867707477,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000873991164466367,
      "loss": 5.0765,
      "step": 5754
    },
    {
      "epoch": 0.7881402355519036,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008739774375029824,
      "loss": 5.0406,
      "step": 5755
    },
    {
      "epoch": 0.7882771843330595,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008739637070259959,
      "loss": 5.0732,
      "step": 5756
    },
    {
      "epoch": 0.7884141331142153,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008739499730355209,
      "loss": 4.9706,
      "step": 5757
    },
    {
      "epoch": 0.7885510818953712,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008739362355316716,
      "loss": 5.0242,
      "step": 5758
    },
    {
      "epoch": 0.7886880306765269,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008739224945145617,
      "loss": 5.0123,
      "step": 5759
    },
    {
      "epoch": 0.7888249794576828,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000873908749984305,
      "loss": 5.1119,
      "step": 5760
    },
    {
      "epoch": 0.7889619282388387,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008738950019410156,
      "loss": 5.0054,
      "step": 5761
    },
    {
      "epoch": 0.7890988770199945,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008738812503848073,
      "loss": 5.0036,
      "step": 5762
    },
    {
      "epoch": 0.7892358258011504,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008738674953157938,
      "loss": 4.9827,
      "step": 5763
    },
    {
      "epoch": 0.7893727745823063,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008738537367340897,
      "loss": 4.9659,
      "step": 5764
    },
    {
      "epoch": 0.789509723363462,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008738399746398085,
      "loss": 5.0277,
      "step": 5765
    },
    {
      "epoch": 0.7896466721446179,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008738262090330645,
      "loss": 5.0122,
      "step": 5766
    },
    {
      "epoch": 0.7897836209257738,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008738124399139716,
      "loss": 5.1109,
      "step": 5767
    },
    {
      "epoch": 0.7899205697069296,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008737986672826441,
      "loss": 5.017,
      "step": 5768
    },
    {
      "epoch": 0.7900575184880855,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008737848911391961,
      "loss": 4.9435,
      "step": 5769
    },
    {
      "epoch": 0.7901944672692413,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008737711114837417,
      "loss": 5.0002,
      "step": 5770
    },
    {
      "epoch": 0.7903314160503971,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008737573283163952,
      "loss": 5.0499,
      "step": 5771
    },
    {
      "epoch": 0.790468364831553,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008737435416372707,
      "loss": 5.064,
      "step": 5772
    },
    {
      "epoch": 0.7906053136127088,
      "grad_norm": 0.375,
      "learning_rate": 0.0008737297514464824,
      "loss": 5.0162,
      "step": 5773
    },
    {
      "epoch": 0.7907422623938647,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008737159577441447,
      "loss": 5.0625,
      "step": 5774
    },
    {
      "epoch": 0.7908792111750206,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008737021605303719,
      "loss": 4.9865,
      "step": 5775
    },
    {
      "epoch": 0.7910161599561764,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008736883598052784,
      "loss": 5.0176,
      "step": 5776
    },
    {
      "epoch": 0.7911531087373322,
      "grad_norm": 0.375,
      "learning_rate": 0.0008736745555689785,
      "loss": 5.0317,
      "step": 5777
    },
    {
      "epoch": 0.7912900575184881,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008736607478215865,
      "loss": 5.0419,
      "step": 5778
    },
    {
      "epoch": 0.7914270062996439,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008736469365632171,
      "loss": 5.0539,
      "step": 5779
    },
    {
      "epoch": 0.7915639550807998,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008736331217939844,
      "loss": 5.0333,
      "step": 5780
    },
    {
      "epoch": 0.7917009038619556,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008736193035140032,
      "loss": 5.017,
      "step": 5781
    },
    {
      "epoch": 0.7918378526431115,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008736054817233879,
      "loss": 5.082,
      "step": 5782
    },
    {
      "epoch": 0.7919748014242673,
      "grad_norm": 0.421875,
      "learning_rate": 0.000873591656422253,
      "loss": 5.0144,
      "step": 5783
    },
    {
      "epoch": 0.7921117502054231,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008735778276107131,
      "loss": 5.0704,
      "step": 5784
    },
    {
      "epoch": 0.792248698986579,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008735639952888827,
      "loss": 5.0889,
      "step": 5785
    },
    {
      "epoch": 0.7923856477677349,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008735501594568767,
      "loss": 5.0093,
      "step": 5786
    },
    {
      "epoch": 0.7925225965488907,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008735363201148095,
      "loss": 4.9966,
      "step": 5787
    },
    {
      "epoch": 0.7926595453300466,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000873522477262796,
      "loss": 5.0901,
      "step": 5788
    },
    {
      "epoch": 0.7927964941112025,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008735086309009508,
      "loss": 5.0547,
      "step": 5789
    },
    {
      "epoch": 0.7929334428923582,
      "grad_norm": 0.375,
      "learning_rate": 0.0008734947810293886,
      "loss": 5.0165,
      "step": 5790
    },
    {
      "epoch": 0.7930703916735141,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008734809276482241,
      "loss": 5.0417,
      "step": 5791
    },
    {
      "epoch": 0.79320734045467,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008734670707575724,
      "loss": 4.9682,
      "step": 5792
    },
    {
      "epoch": 0.7933442892358258,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008734532103575482,
      "loss": 5.0825,
      "step": 5793
    },
    {
      "epoch": 0.7934812380169817,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008734393464482663,
      "loss": 4.962,
      "step": 5794
    },
    {
      "epoch": 0.7936181867981374,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008734254790298417,
      "loss": 4.975,
      "step": 5795
    },
    {
      "epoch": 0.7937551355792933,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008734116081023891,
      "loss": 5.0376,
      "step": 5796
    },
    {
      "epoch": 0.7938920843604492,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008733977336660237,
      "loss": 5.0729,
      "step": 5797
    },
    {
      "epoch": 0.794029033141605,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008733838557208604,
      "loss": 5.0244,
      "step": 5798
    },
    {
      "epoch": 0.7941659819227609,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008733699742670142,
      "loss": 4.9813,
      "step": 5799
    },
    {
      "epoch": 0.7943029307039168,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008733560893046002,
      "loss": 5.0325,
      "step": 5800
    },
    {
      "epoch": 0.7944398794850726,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008733422008337334,
      "loss": 5.0879,
      "step": 5801
    },
    {
      "epoch": 0.7945768282662284,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008733283088545289,
      "loss": 4.9726,
      "step": 5802
    },
    {
      "epoch": 0.7947137770473843,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008733144133671018,
      "loss": 5.1256,
      "step": 5803
    },
    {
      "epoch": 0.7948507258285401,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008733005143715672,
      "loss": 4.9739,
      "step": 5804
    },
    {
      "epoch": 0.794987674609696,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008732866118680404,
      "loss": 4.9574,
      "step": 5805
    },
    {
      "epoch": 0.7951246233908518,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008732727058566367,
      "loss": 5.0743,
      "step": 5806
    },
    {
      "epoch": 0.7952615721720077,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000873258796337471,
      "loss": 4.9974,
      "step": 5807
    },
    {
      "epoch": 0.7953985209531635,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000873244883310659,
      "loss": 4.9704,
      "step": 5808
    },
    {
      "epoch": 0.7955354697343193,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008732309667763158,
      "loss": 4.9625,
      "step": 5809
    },
    {
      "epoch": 0.7956724185154752,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008732170467345566,
      "loss": 4.9822,
      "step": 5810
    },
    {
      "epoch": 0.7958093672966311,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000873203123185497,
      "loss": 5.0091,
      "step": 5811
    },
    {
      "epoch": 0.7959463160777869,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008731891961292522,
      "loss": 4.9867,
      "step": 5812
    },
    {
      "epoch": 0.7960832648589428,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008731752655659377,
      "loss": 5.0044,
      "step": 5813
    },
    {
      "epoch": 0.7962202136400987,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000873161331495669,
      "loss": 4.9976,
      "step": 5814
    },
    {
      "epoch": 0.7963571624212544,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008731473939185615,
      "loss": 4.9883,
      "step": 5815
    },
    {
      "epoch": 0.7964941112024103,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008731334528347308,
      "loss": 5.0098,
      "step": 5816
    },
    {
      "epoch": 0.7966310599835661,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008731195082442921,
      "loss": 5.0503,
      "step": 5817
    },
    {
      "epoch": 0.796768008764722,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008731055601473613,
      "loss": 5.0635,
      "step": 5818
    },
    {
      "epoch": 0.7969049575458779,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008730916085440541,
      "loss": 5.0206,
      "step": 5819
    },
    {
      "epoch": 0.7970419063270336,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008730776534344858,
      "loss": 4.9726,
      "step": 5820
    },
    {
      "epoch": 0.7971788551081895,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008730636948187721,
      "loss": 5.0176,
      "step": 5821
    },
    {
      "epoch": 0.7973158038893454,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008730497326970287,
      "loss": 4.9752,
      "step": 5822
    },
    {
      "epoch": 0.7974527526705012,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008730357670693716,
      "loss": 5.0301,
      "step": 5823
    },
    {
      "epoch": 0.7975897014516571,
      "grad_norm": 0.40625,
      "learning_rate": 0.000873021797935916,
      "loss": 5.0201,
      "step": 5824
    },
    {
      "epoch": 0.797726650232813,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008730078252967783,
      "loss": 4.9848,
      "step": 5825
    },
    {
      "epoch": 0.7978635990139687,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008729938491520737,
      "loss": 4.9987,
      "step": 5826
    },
    {
      "epoch": 0.7980005477951246,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008729798695019182,
      "loss": 4.9908,
      "step": 5827
    },
    {
      "epoch": 0.7981374965762805,
      "grad_norm": 0.4921875,
      "learning_rate": 0.000872965886346428,
      "loss": 4.9451,
      "step": 5828
    },
    {
      "epoch": 0.7982744453574363,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008729518996857186,
      "loss": 4.9837,
      "step": 5829
    },
    {
      "epoch": 0.7984113941385922,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000872937909519906,
      "loss": 5.0451,
      "step": 5830
    },
    {
      "epoch": 0.798548342919748,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008729239158491062,
      "loss": 5.0147,
      "step": 5831
    },
    {
      "epoch": 0.7986852917009039,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008729099186734351,
      "loss": 4.9925,
      "step": 5832
    },
    {
      "epoch": 0.7988222404820597,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008728959179930087,
      "loss": 5.0106,
      "step": 5833
    },
    {
      "epoch": 0.7989591892632155,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008728819138079431,
      "loss": 4.9609,
      "step": 5834
    },
    {
      "epoch": 0.7990961380443714,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008728679061183543,
      "loss": 5.05,
      "step": 5835
    },
    {
      "epoch": 0.7992330868255273,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008728538949243584,
      "loss": 4.9993,
      "step": 5836
    },
    {
      "epoch": 0.7993700356066831,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008728398802260715,
      "loss": 5.0458,
      "step": 5837
    },
    {
      "epoch": 0.799506984387839,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008728258620236098,
      "loss": 5.0481,
      "step": 5838
    },
    {
      "epoch": 0.7996439331689948,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008728118403170894,
      "loss": 5.0548,
      "step": 5839
    },
    {
      "epoch": 0.7997808819501506,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008727978151066265,
      "loss": 5.0289,
      "step": 5840
    },
    {
      "epoch": 0.7999178307313065,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008727837863923373,
      "loss": 4.9874,
      "step": 5841
    },
    {
      "epoch": 0.8000547795124623,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008727697541743383,
      "loss": 5.0138,
      "step": 5842
    },
    {
      "epoch": 0.8001917282936182,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008727557184527456,
      "loss": 4.9444,
      "step": 5843
    },
    {
      "epoch": 0.8003286770747741,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008727416792276753,
      "loss": 4.9704,
      "step": 5844
    },
    {
      "epoch": 0.8004656258559298,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008727276364992442,
      "loss": 5.0926,
      "step": 5845
    },
    {
      "epoch": 0.8006025746370857,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008727135902675683,
      "loss": 5.0218,
      "step": 5846
    },
    {
      "epoch": 0.8007395234182416,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008726995405327643,
      "loss": 5.0314,
      "step": 5847
    },
    {
      "epoch": 0.8008764721993974,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008726854872949483,
      "loss": 5.0127,
      "step": 5848
    },
    {
      "epoch": 0.8010134209805533,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000872671430554237,
      "loss": 4.97,
      "step": 5849
    },
    {
      "epoch": 0.8011503697617092,
      "grad_norm": 0.375,
      "learning_rate": 0.000872657370310747,
      "loss": 4.9783,
      "step": 5850
    },
    {
      "epoch": 0.801287318542865,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008726433065645946,
      "loss": 5.0695,
      "step": 5851
    },
    {
      "epoch": 0.8014242673240208,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008726292393158962,
      "loss": 5.0671,
      "step": 5852
    },
    {
      "epoch": 0.8015612161051766,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008726151685647688,
      "loss": 5.1094,
      "step": 5853
    },
    {
      "epoch": 0.8016981648863325,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008726010943113287,
      "loss": 5.084,
      "step": 5854
    },
    {
      "epoch": 0.8018351136674884,
      "grad_norm": 0.375,
      "learning_rate": 0.0008725870165556927,
      "loss": 5.0642,
      "step": 5855
    },
    {
      "epoch": 0.8019720624486442,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008725729352979773,
      "loss": 5.0429,
      "step": 5856
    },
    {
      "epoch": 0.8021090112298,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008725588505382992,
      "loss": 5.0277,
      "step": 5857
    },
    {
      "epoch": 0.8022459600109559,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008725447622767755,
      "loss": 5.0865,
      "step": 5858
    },
    {
      "epoch": 0.8023829087921117,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008725306705135224,
      "loss": 5.0368,
      "step": 5859
    },
    {
      "epoch": 0.8025198575732676,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008725165752486571,
      "loss": 4.9548,
      "step": 5860
    },
    {
      "epoch": 0.8026568063544235,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008725024764822963,
      "loss": 5.1256,
      "step": 5861
    },
    {
      "epoch": 0.8027937551355793,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008724883742145567,
      "loss": 4.9813,
      "step": 5862
    },
    {
      "epoch": 0.8029307039167352,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008724742684455553,
      "loss": 4.9677,
      "step": 5863
    },
    {
      "epoch": 0.803067652697891,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000872460159175409,
      "loss": 4.9931,
      "step": 5864
    },
    {
      "epoch": 0.8032046014790468,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008724460464042345,
      "loss": 5.0821,
      "step": 5865
    },
    {
      "epoch": 0.8033415502602027,
      "grad_norm": 0.375,
      "learning_rate": 0.0008724319301321493,
      "loss": 5.034,
      "step": 5866
    },
    {
      "epoch": 0.8034784990413585,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008724178103592698,
      "loss": 5.0597,
      "step": 5867
    },
    {
      "epoch": 0.8036154478225144,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008724036870857133,
      "loss": 4.9293,
      "step": 5868
    },
    {
      "epoch": 0.8037523966036703,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008723895603115968,
      "loss": 4.9592,
      "step": 5869
    },
    {
      "epoch": 0.803889345384826,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008723754300370373,
      "loss": 5.0606,
      "step": 5870
    },
    {
      "epoch": 0.8040262941659819,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000872361296262152,
      "loss": 4.9892,
      "step": 5871
    },
    {
      "epoch": 0.8041632429471378,
      "grad_norm": 0.578125,
      "learning_rate": 0.000872347158987058,
      "loss": 5.0237,
      "step": 5872
    },
    {
      "epoch": 0.8043001917282936,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008723330182118726,
      "loss": 5.0842,
      "step": 5873
    },
    {
      "epoch": 0.8044371405094495,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008723188739367127,
      "loss": 5.0004,
      "step": 5874
    },
    {
      "epoch": 0.8045740892906054,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008723047261616955,
      "loss": 4.9667,
      "step": 5875
    },
    {
      "epoch": 0.8047110380717611,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008722905748869387,
      "loss": 4.914,
      "step": 5876
    },
    {
      "epoch": 0.804847986852917,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008722764201125592,
      "loss": 5.0553,
      "step": 5877
    },
    {
      "epoch": 0.8049849356340728,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008722622618386743,
      "loss": 5.1047,
      "step": 5878
    },
    {
      "epoch": 0.8051218844152287,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008722481000654013,
      "loss": 5.0215,
      "step": 5879
    },
    {
      "epoch": 0.8052588331963846,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008722339347928578,
      "loss": 4.9753,
      "step": 5880
    },
    {
      "epoch": 0.8053957819775404,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008722197660211612,
      "loss": 5.0539,
      "step": 5881
    },
    {
      "epoch": 0.8055327307586962,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008722055937504287,
      "loss": 5.0053,
      "step": 5882
    },
    {
      "epoch": 0.8056696795398521,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008721914179807778,
      "loss": 5.0065,
      "step": 5883
    },
    {
      "epoch": 0.8058066283210079,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008721772387123261,
      "loss": 4.9815,
      "step": 5884
    },
    {
      "epoch": 0.8059435771021638,
      "grad_norm": 0.52734375,
      "learning_rate": 0.000872163055945191,
      "loss": 5.0626,
      "step": 5885
    },
    {
      "epoch": 0.8060805258833197,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00087214886967949,
      "loss": 5.054,
      "step": 5886
    },
    {
      "epoch": 0.8062174746644755,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008721346799153407,
      "loss": 5.0239,
      "step": 5887
    },
    {
      "epoch": 0.8063544234456314,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008721204866528608,
      "loss": 5.0491,
      "step": 5888
    },
    {
      "epoch": 0.8064913722267871,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008721062898921678,
      "loss": 5.0468,
      "step": 5889
    },
    {
      "epoch": 0.806628321007943,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008720920896333794,
      "loss": 4.9768,
      "step": 5890
    },
    {
      "epoch": 0.8067652697890989,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008720778858766132,
      "loss": 4.9995,
      "step": 5891
    },
    {
      "epoch": 0.8069022185702547,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008720636786219871,
      "loss": 5.0692,
      "step": 5892
    },
    {
      "epoch": 0.8070391673514106,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008720494678696188,
      "loss": 4.9914,
      "step": 5893
    },
    {
      "epoch": 0.8071761161325665,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008720352536196257,
      "loss": 4.9844,
      "step": 5894
    },
    {
      "epoch": 0.8073130649137222,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008720210358721262,
      "loss": 5.0297,
      "step": 5895
    },
    {
      "epoch": 0.8074500136948781,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008720068146272376,
      "loss": 5.0114,
      "step": 5896
    },
    {
      "epoch": 0.807586962476034,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008719925898850781,
      "loss": 5.086,
      "step": 5897
    },
    {
      "epoch": 0.8077239112571898,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008719783616457653,
      "loss": 5.1193,
      "step": 5898
    },
    {
      "epoch": 0.8078608600383457,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008719641299094173,
      "loss": 5.0438,
      "step": 5899
    },
    {
      "epoch": 0.8079978088195016,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008719498946761522,
      "loss": 4.944,
      "step": 5900
    },
    {
      "epoch": 0.8081347576006573,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008719356559460876,
      "loss": 5.0104,
      "step": 5901
    },
    {
      "epoch": 0.8082717063818132,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008719214137193418,
      "loss": 5.0038,
      "step": 5902
    },
    {
      "epoch": 0.808408655162969,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008719071679960327,
      "loss": 5.057,
      "step": 5903
    },
    {
      "epoch": 0.8085456039441249,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008718929187762781,
      "loss": 4.9669,
      "step": 5904
    },
    {
      "epoch": 0.8086825527252808,
      "grad_norm": 0.375,
      "learning_rate": 0.0008718786660601966,
      "loss": 5.0285,
      "step": 5905
    },
    {
      "epoch": 0.8088195015064366,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008718644098479061,
      "loss": 4.9256,
      "step": 5906
    },
    {
      "epoch": 0.8089564502875924,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008718501501395246,
      "loss": 4.9294,
      "step": 5907
    },
    {
      "epoch": 0.8090933990687483,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008718358869351704,
      "loss": 5.0066,
      "step": 5908
    },
    {
      "epoch": 0.8092303478499041,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008718216202349616,
      "loss": 5.0507,
      "step": 5909
    },
    {
      "epoch": 0.80936729663106,
      "grad_norm": 0.375,
      "learning_rate": 0.0008718073500390167,
      "loss": 4.9525,
      "step": 5910
    },
    {
      "epoch": 0.8095042454122159,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008717930763474536,
      "loss": 4.9607,
      "step": 5911
    },
    {
      "epoch": 0.8096411941933717,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008717787991603909,
      "loss": 5.0386,
      "step": 5912
    },
    {
      "epoch": 0.8097781429745275,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008717645184779468,
      "loss": 5.021,
      "step": 5913
    },
    {
      "epoch": 0.8099150917556833,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008717502343002396,
      "loss": 5.0165,
      "step": 5914
    },
    {
      "epoch": 0.8100520405368392,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008717359466273876,
      "loss": 4.988,
      "step": 5915
    },
    {
      "epoch": 0.8101889893179951,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008717216554595095,
      "loss": 5.1013,
      "step": 5916
    },
    {
      "epoch": 0.8103259380991509,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008717073607967233,
      "loss": 5.0532,
      "step": 5917
    },
    {
      "epoch": 0.8104628868803068,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000871693062639148,
      "loss": 4.9905,
      "step": 5918
    },
    {
      "epoch": 0.8105998356614627,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008716787609869016,
      "loss": 5.1046,
      "step": 5919
    },
    {
      "epoch": 0.8107367844426184,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008716644558401029,
      "loss": 4.9684,
      "step": 5920
    },
    {
      "epoch": 0.8108737332237743,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008716501471988704,
      "loss": 5.0754,
      "step": 5921
    },
    {
      "epoch": 0.8110106820049302,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008716358350633227,
      "loss": 4.9893,
      "step": 5922
    },
    {
      "epoch": 0.811147630786086,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008716215194335782,
      "loss": 4.9364,
      "step": 5923
    },
    {
      "epoch": 0.8112845795672419,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008716072003097558,
      "loss": 4.9903,
      "step": 5924
    },
    {
      "epoch": 0.8114215283483976,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008715928776919741,
      "loss": 5.0128,
      "step": 5925
    },
    {
      "epoch": 0.8115584771295535,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008715785515803518,
      "loss": 4.999,
      "step": 5926
    },
    {
      "epoch": 0.8116954259107094,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008715642219750075,
      "loss": 5.0102,
      "step": 5927
    },
    {
      "epoch": 0.8118323746918652,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00087154988887606,
      "loss": 5.0389,
      "step": 5928
    },
    {
      "epoch": 0.8119693234730211,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008715355522836282,
      "loss": 5.0536,
      "step": 5929
    },
    {
      "epoch": 0.812106272254177,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008715212121978309,
      "loss": 5.0763,
      "step": 5930
    },
    {
      "epoch": 0.8122432210353328,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008715068686187867,
      "loss": 5.0573,
      "step": 5931
    },
    {
      "epoch": 0.8123801698164886,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008714925215466148,
      "loss": 4.9956,
      "step": 5932
    },
    {
      "epoch": 0.8125171185976445,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008714781709814338,
      "loss": 5.0085,
      "step": 5933
    },
    {
      "epoch": 0.8126540673788003,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008714638169233628,
      "loss": 5.0647,
      "step": 5934
    },
    {
      "epoch": 0.8127910161599562,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008714494593725208,
      "loss": 5.105,
      "step": 5935
    },
    {
      "epoch": 0.812927964941112,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008714350983290266,
      "loss": 5.079,
      "step": 5936
    },
    {
      "epoch": 0.8130649137222679,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008714207337929994,
      "loss": 5.1029,
      "step": 5937
    },
    {
      "epoch": 0.8132018625034237,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008714063657645581,
      "loss": 4.965,
      "step": 5938
    },
    {
      "epoch": 0.8133388112845795,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008713919942438218,
      "loss": 5.0803,
      "step": 5939
    },
    {
      "epoch": 0.8134757600657354,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008713776192309097,
      "loss": 4.998,
      "step": 5940
    },
    {
      "epoch": 0.8136127088468913,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008713632407259409,
      "loss": 5.123,
      "step": 5941
    },
    {
      "epoch": 0.8137496576280471,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008713488587290345,
      "loss": 5.0514,
      "step": 5942
    },
    {
      "epoch": 0.813886606409203,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008713344732403097,
      "loss": 4.9742,
      "step": 5943
    },
    {
      "epoch": 0.8140235551903588,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008713200842598856,
      "loss": 5.0479,
      "step": 5944
    },
    {
      "epoch": 0.8141605039715146,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008713056917878817,
      "loss": 5.0513,
      "step": 5945
    },
    {
      "epoch": 0.8142974527526705,
      "grad_norm": 0.5,
      "learning_rate": 0.0008712912958244171,
      "loss": 4.9153,
      "step": 5946
    },
    {
      "epoch": 0.8144344015338264,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008712768963696112,
      "loss": 4.9988,
      "step": 5947
    },
    {
      "epoch": 0.8145713503149822,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008712624934235832,
      "loss": 4.9507,
      "step": 5948
    },
    {
      "epoch": 0.8147082990961381,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008712480869864525,
      "loss": 5.0155,
      "step": 5949
    },
    {
      "epoch": 0.8148452478772938,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008712336770583386,
      "loss": 5.0164,
      "step": 5950
    },
    {
      "epoch": 0.8149821966584497,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008712192636393608,
      "loss": 5.0419,
      "step": 5951
    },
    {
      "epoch": 0.8151191454396056,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008712048467296386,
      "loss": 5.021,
      "step": 5952
    },
    {
      "epoch": 0.8152560942207614,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008711904263292915,
      "loss": 4.9483,
      "step": 5953
    },
    {
      "epoch": 0.8153930430019173,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008711760024384391,
      "loss": 4.9649,
      "step": 5954
    },
    {
      "epoch": 0.8155299917830732,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008711615750572006,
      "loss": 4.9782,
      "step": 5955
    },
    {
      "epoch": 0.815666940564229,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008711471441856958,
      "loss": 4.9977,
      "step": 5956
    },
    {
      "epoch": 0.8158038893453848,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008711327098240442,
      "loss": 5.1103,
      "step": 5957
    },
    {
      "epoch": 0.8159408381265407,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008711182719723657,
      "loss": 4.9928,
      "step": 5958
    },
    {
      "epoch": 0.8160777869076965,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008711038306307797,
      "loss": 5.0493,
      "step": 5959
    },
    {
      "epoch": 0.8162147356888524,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008710893857994058,
      "loss": 4.9077,
      "step": 5960
    },
    {
      "epoch": 0.8163516844700082,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008710749374783637,
      "loss": 4.9964,
      "step": 5961
    },
    {
      "epoch": 0.816488633251164,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008710604856677735,
      "loss": 4.9127,
      "step": 5962
    },
    {
      "epoch": 0.8166255820323199,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008710460303677546,
      "loss": 5.0815,
      "step": 5963
    },
    {
      "epoch": 0.8167625308134757,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008710315715784269,
      "loss": 4.9887,
      "step": 5964
    },
    {
      "epoch": 0.8168994795946316,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008710171092999102,
      "loss": 4.9222,
      "step": 5965
    },
    {
      "epoch": 0.8170364283757875,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008710026435323246,
      "loss": 5.0397,
      "step": 5966
    },
    {
      "epoch": 0.8171733771569433,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008709881742757897,
      "loss": 5.0162,
      "step": 5967
    },
    {
      "epoch": 0.8173103259380992,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008709737015304254,
      "loss": 5.0132,
      "step": 5968
    },
    {
      "epoch": 0.817447274719255,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008709592252963517,
      "loss": 4.9794,
      "step": 5969
    },
    {
      "epoch": 0.8175842235004108,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008709447455736886,
      "loss": 5.033,
      "step": 5970
    },
    {
      "epoch": 0.8177211722815667,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008709302623625563,
      "loss": 4.9053,
      "step": 5971
    },
    {
      "epoch": 0.8178581210627225,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008709157756630743,
      "loss": 4.9449,
      "step": 5972
    },
    {
      "epoch": 0.8179950698438784,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008709012854753632,
      "loss": 5.0784,
      "step": 5973
    },
    {
      "epoch": 0.8181320186250343,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008708867917995427,
      "loss": 4.9603,
      "step": 5974
    },
    {
      "epoch": 0.81826896740619,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008708722946357332,
      "loss": 4.9747,
      "step": 5975
    },
    {
      "epoch": 0.8184059161873459,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008708577939840546,
      "loss": 4.9225,
      "step": 5976
    },
    {
      "epoch": 0.8185428649685018,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008708432898446272,
      "loss": 4.9731,
      "step": 5977
    },
    {
      "epoch": 0.8186798137496576,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008708287822175712,
      "loss": 5.0455,
      "step": 5978
    },
    {
      "epoch": 0.8188167625308135,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008708142711030067,
      "loss": 5.0136,
      "step": 5979
    },
    {
      "epoch": 0.8189537113119694,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008707997565010542,
      "loss": 5.0213,
      "step": 5980
    },
    {
      "epoch": 0.8190906600931251,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008707852384118338,
      "loss": 4.984,
      "step": 5981
    },
    {
      "epoch": 0.819227608874281,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008707707168354659,
      "loss": 5.0754,
      "step": 5982
    },
    {
      "epoch": 0.8193645576554369,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008707561917720708,
      "loss": 5.0686,
      "step": 5983
    },
    {
      "epoch": 0.8195015064365927,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008707416632217688,
      "loss": 5.028,
      "step": 5984
    },
    {
      "epoch": 0.8196384552177486,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008707271311846804,
      "loss": 4.9599,
      "step": 5985
    },
    {
      "epoch": 0.8197754039989044,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000870712595660926,
      "loss": 5.0195,
      "step": 5986
    },
    {
      "epoch": 0.8199123527800602,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008706980566506262,
      "loss": 4.9975,
      "step": 5987
    },
    {
      "epoch": 0.8200493015612161,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008706835141539012,
      "loss": 4.9926,
      "step": 5988
    },
    {
      "epoch": 0.8201862503423719,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008706689681708717,
      "loss": 5.0172,
      "step": 5989
    },
    {
      "epoch": 0.8203231991235278,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008706544187016583,
      "loss": 5.0311,
      "step": 5990
    },
    {
      "epoch": 0.8204601479046837,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008706398657463815,
      "loss": 5.0712,
      "step": 5991
    },
    {
      "epoch": 0.8205970966858395,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008706253093051619,
      "loss": 5.0011,
      "step": 5992
    },
    {
      "epoch": 0.8207340454669954,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008706107493781201,
      "loss": 5.0056,
      "step": 5993
    },
    {
      "epoch": 0.8208709942481512,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008705961859653767,
      "loss": 4.9682,
      "step": 5994
    },
    {
      "epoch": 0.821007943029307,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008705816190670525,
      "loss": 5.0148,
      "step": 5995
    },
    {
      "epoch": 0.8211448918104629,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008705670486832682,
      "loss": 5.0268,
      "step": 5996
    },
    {
      "epoch": 0.8212818405916187,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008705524748141447,
      "loss": 5.0601,
      "step": 5997
    },
    {
      "epoch": 0.8214187893727746,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008705378974598024,
      "loss": 5.0074,
      "step": 5998
    },
    {
      "epoch": 0.8215557381539305,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008705233166203626,
      "loss": 4.9338,
      "step": 5999
    },
    {
      "epoch": 0.8216926869350862,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008705087322959458,
      "loss": 5.0158,
      "step": 6000
    },
    {
      "epoch": 0.8218296357162421,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008704941444866728,
      "loss": 5.064,
      "step": 6001
    },
    {
      "epoch": 0.821966584497398,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008704795531926648,
      "loss": 4.9223,
      "step": 6002
    },
    {
      "epoch": 0.8221035332785538,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008704649584140424,
      "loss": 5.0347,
      "step": 6003
    },
    {
      "epoch": 0.8222404820597097,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008704503601509268,
      "loss": 5.0644,
      "step": 6004
    },
    {
      "epoch": 0.8223774308408656,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008704357584034388,
      "loss": 4.9901,
      "step": 6005
    },
    {
      "epoch": 0.8225143796220213,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008704211531716995,
      "loss": 5.0605,
      "step": 6006
    },
    {
      "epoch": 0.8226513284031772,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008704065444558299,
      "loss": 4.9685,
      "step": 6007
    },
    {
      "epoch": 0.822788277184333,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008703919322559512,
      "loss": 4.9187,
      "step": 6008
    },
    {
      "epoch": 0.8229252259654889,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008703773165721844,
      "loss": 4.9922,
      "step": 6009
    },
    {
      "epoch": 0.8230621747466448,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008703626974046505,
      "loss": 5.0307,
      "step": 6010
    },
    {
      "epoch": 0.8231991235278006,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008703480747534707,
      "loss": 4.9676,
      "step": 6011
    },
    {
      "epoch": 0.8233360723089564,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008703334486187664,
      "loss": 5.0135,
      "step": 6012
    },
    {
      "epoch": 0.8234730210901123,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008703188190006585,
      "loss": 5.0328,
      "step": 6013
    },
    {
      "epoch": 0.8236099698712681,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008703041858992684,
      "loss": 4.9573,
      "step": 6014
    },
    {
      "epoch": 0.823746918652424,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008702895493147175,
      "loss": 5.0156,
      "step": 6015
    },
    {
      "epoch": 0.8238838674335799,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008702749092471269,
      "loss": 5.0673,
      "step": 6016
    },
    {
      "epoch": 0.8240208162147357,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008702602656966179,
      "loss": 4.9219,
      "step": 6017
    },
    {
      "epoch": 0.8241577649958916,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000870245618663312,
      "loss": 4.9874,
      "step": 6018
    },
    {
      "epoch": 0.8242947137770474,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008702309681473306,
      "loss": 4.9527,
      "step": 6019
    },
    {
      "epoch": 0.8244316625582032,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008702163141487949,
      "loss": 5.0148,
      "step": 6020
    },
    {
      "epoch": 0.8245686113393591,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008702016566678265,
      "loss": 5.0262,
      "step": 6021
    },
    {
      "epoch": 0.8247055601205149,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008701869957045469,
      "loss": 4.9321,
      "step": 6022
    },
    {
      "epoch": 0.8248425089016708,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008701723312590776,
      "loss": 4.9782,
      "step": 6023
    },
    {
      "epoch": 0.8249794576828267,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008701576633315399,
      "loss": 5.0039,
      "step": 6024
    },
    {
      "epoch": 0.8251164064639824,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008701429919220555,
      "loss": 5.0623,
      "step": 6025
    },
    {
      "epoch": 0.8252533552451383,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008701283170307461,
      "loss": 4.982,
      "step": 6026
    },
    {
      "epoch": 0.8253903040262942,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008701136386577333,
      "loss": 4.9838,
      "step": 6027
    },
    {
      "epoch": 0.82552725280745,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008700989568031386,
      "loss": 4.9721,
      "step": 6028
    },
    {
      "epoch": 0.8256642015886059,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008700842714670838,
      "loss": 4.9437,
      "step": 6029
    },
    {
      "epoch": 0.8258011503697618,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008700695826496904,
      "loss": 4.9731,
      "step": 6030
    },
    {
      "epoch": 0.8259380991509175,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008700548903510804,
      "loss": 4.9317,
      "step": 6031
    },
    {
      "epoch": 0.8260750479320734,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008700401945713754,
      "loss": 4.9379,
      "step": 6032
    },
    {
      "epoch": 0.8262119967132292,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008700254953106972,
      "loss": 4.9526,
      "step": 6033
    },
    {
      "epoch": 0.8263489454943851,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008700107925691677,
      "loss": 5.0015,
      "step": 6034
    },
    {
      "epoch": 0.826485894275541,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008699960863469085,
      "loss": 4.9655,
      "step": 6035
    },
    {
      "epoch": 0.8266228430566968,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008699813766440418,
      "loss": 5.0034,
      "step": 6036
    },
    {
      "epoch": 0.8267597918378526,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008699666634606894,
      "loss": 4.9988,
      "step": 6037
    },
    {
      "epoch": 0.8268967406190085,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008699519467969731,
      "loss": 4.9695,
      "step": 6038
    },
    {
      "epoch": 0.8270336894001643,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008699372266530149,
      "loss": 5.0819,
      "step": 6039
    },
    {
      "epoch": 0.8271706381813202,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008699225030289368,
      "loss": 5.0672,
      "step": 6040
    },
    {
      "epoch": 0.8273075869624761,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000869907775924861,
      "loss": 5.0774,
      "step": 6041
    },
    {
      "epoch": 0.8274445357436319,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008698930453409094,
      "loss": 4.9536,
      "step": 6042
    },
    {
      "epoch": 0.8275814845247877,
      "grad_norm": 0.375,
      "learning_rate": 0.0008698783112772041,
      "loss": 5.0961,
      "step": 6043
    },
    {
      "epoch": 0.8277184333059435,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008698635737338671,
      "loss": 5.0218,
      "step": 6044
    },
    {
      "epoch": 0.8278553820870994,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008698488327110206,
      "loss": 4.9456,
      "step": 6045
    },
    {
      "epoch": 0.8279923308682553,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008698340882087868,
      "loss": 5.0144,
      "step": 6046
    },
    {
      "epoch": 0.8281292796494111,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008698193402272878,
      "loss": 5.0146,
      "step": 6047
    },
    {
      "epoch": 0.828266228430567,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008698045887666461,
      "loss": 5.0445,
      "step": 6048
    },
    {
      "epoch": 0.8284031772117229,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008697898338269835,
      "loss": 5.0156,
      "step": 6049
    },
    {
      "epoch": 0.8285401259928786,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008697750754084227,
      "loss": 5.0599,
      "step": 6050
    },
    {
      "epoch": 0.8286770747740345,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008697603135110858,
      "loss": 4.9658,
      "step": 6051
    },
    {
      "epoch": 0.8288140235551904,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008697455481350951,
      "loss": 5.0271,
      "step": 6052
    },
    {
      "epoch": 0.8289509723363462,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000869730779280573,
      "loss": 5.0261,
      "step": 6053
    },
    {
      "epoch": 0.8290879211175021,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008697160069476419,
      "loss": 4.9613,
      "step": 6054
    },
    {
      "epoch": 0.829224869898658,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008697012311364243,
      "loss": 5.0326,
      "step": 6055
    },
    {
      "epoch": 0.8293618186798137,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008696864518470426,
      "loss": 5.0268,
      "step": 6056
    },
    {
      "epoch": 0.8294987674609696,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008696716690796192,
      "loss": 5.0826,
      "step": 6057
    },
    {
      "epoch": 0.8296357162421254,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008696568828342767,
      "loss": 4.9809,
      "step": 6058
    },
    {
      "epoch": 0.8297726650232813,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008696420931111377,
      "loss": 5.0243,
      "step": 6059
    },
    {
      "epoch": 0.8299096138044372,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008696272999103246,
      "loss": 5.0476,
      "step": 6060
    },
    {
      "epoch": 0.830046562585593,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008696125032319601,
      "loss": 4.997,
      "step": 6061
    },
    {
      "epoch": 0.8301835113667488,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008695977030761667,
      "loss": 5.0123,
      "step": 6062
    },
    {
      "epoch": 0.8303204601479047,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008695828994430674,
      "loss": 5.0895,
      "step": 6063
    },
    {
      "epoch": 0.8304574089290605,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008695680923327844,
      "loss": 5.0412,
      "step": 6064
    },
    {
      "epoch": 0.8305943577102164,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008695532817454407,
      "loss": 4.9691,
      "step": 6065
    },
    {
      "epoch": 0.8307313064913723,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000869538467681159,
      "loss": 4.9718,
      "step": 6066
    },
    {
      "epoch": 0.8308682552725281,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000869523650140062,
      "loss": 4.9799,
      "step": 6067
    },
    {
      "epoch": 0.8310052040536839,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008695088291222725,
      "loss": 4.9965,
      "step": 6068
    },
    {
      "epoch": 0.8311421528348397,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008694940046279136,
      "loss": 5.0661,
      "step": 6069
    },
    {
      "epoch": 0.8312791016159956,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008694791766571076,
      "loss": 5.0398,
      "step": 6070
    },
    {
      "epoch": 0.8314160503971515,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008694643452099779,
      "loss": 5.0527,
      "step": 6071
    },
    {
      "epoch": 0.8315529991783073,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008694495102866473,
      "loss": 5.0297,
      "step": 6072
    },
    {
      "epoch": 0.8316899479594632,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008694346718872384,
      "loss": 5.0136,
      "step": 6073
    },
    {
      "epoch": 0.831826896740619,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008694198300118747,
      "loss": 4.9571,
      "step": 6074
    },
    {
      "epoch": 0.8319638455217748,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008694049846606788,
      "loss": 4.9558,
      "step": 6075
    },
    {
      "epoch": 0.8321007943029307,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008693901358337739,
      "loss": 4.9932,
      "step": 6076
    },
    {
      "epoch": 0.8322377430840866,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008693752835312828,
      "loss": 4.9412,
      "step": 6077
    },
    {
      "epoch": 0.8323746918652424,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008693604277533291,
      "loss": 5.0151,
      "step": 6078
    },
    {
      "epoch": 0.8325116406463983,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008693455685000355,
      "loss": 4.9359,
      "step": 6079
    },
    {
      "epoch": 0.832648589427554,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008693307057715251,
      "loss": 4.9882,
      "step": 6080
    },
    {
      "epoch": 0.8327855382087099,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008693158395679213,
      "loss": 4.9086,
      "step": 6081
    },
    {
      "epoch": 0.8329224869898658,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008693009698893474,
      "loss": 4.9873,
      "step": 6082
    },
    {
      "epoch": 0.8330594357710216,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008692860967359263,
      "loss": 5.065,
      "step": 6083
    },
    {
      "epoch": 0.8331963845521775,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008692712201077813,
      "loss": 4.9504,
      "step": 6084
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000869256340005036,
      "loss": 5.0082,
      "step": 6085
    },
    {
      "epoch": 0.8334702821144891,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008692414564278135,
      "loss": 5.0093,
      "step": 6086
    },
    {
      "epoch": 0.833607230895645,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000869226569376237,
      "loss": 4.9983,
      "step": 6087
    },
    {
      "epoch": 0.8337441796768009,
      "grad_norm": 0.515625,
      "learning_rate": 0.00086921167885043,
      "loss": 4.9726,
      "step": 6088
    },
    {
      "epoch": 0.8338811284579567,
      "grad_norm": 0.48828125,
      "learning_rate": 0.000869196784850516,
      "loss": 4.9465,
      "step": 6089
    },
    {
      "epoch": 0.8340180772391126,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008691818873766184,
      "loss": 4.9797,
      "step": 6090
    },
    {
      "epoch": 0.8341550260202684,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008691669864288606,
      "loss": 5.0386,
      "step": 6091
    },
    {
      "epoch": 0.8342919748014243,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008691520820073661,
      "loss": 4.9655,
      "step": 6092
    },
    {
      "epoch": 0.8344289235825801,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008691371741122584,
      "loss": 5.0448,
      "step": 6093
    },
    {
      "epoch": 0.8345658723637359,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008691222627436611,
      "loss": 5.0318,
      "step": 6094
    },
    {
      "epoch": 0.8347028211448918,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008691073479016977,
      "loss": 4.984,
      "step": 6095
    },
    {
      "epoch": 0.8348397699260477,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008690924295864917,
      "loss": 5.045,
      "step": 6096
    },
    {
      "epoch": 0.8349767187072035,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008690775077981671,
      "loss": 5.016,
      "step": 6097
    },
    {
      "epoch": 0.8351136674883594,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008690625825368473,
      "loss": 5.0045,
      "step": 6098
    },
    {
      "epoch": 0.8352506162695152,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008690476538026559,
      "loss": 5.0169,
      "step": 6099
    },
    {
      "epoch": 0.835387565050671,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008690327215957168,
      "loss": 5.0154,
      "step": 6100
    },
    {
      "epoch": 0.8355245138318269,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008690177859161537,
      "loss": 5.081,
      "step": 6101
    },
    {
      "epoch": 0.8356614626129828,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008690028467640904,
      "loss": 4.9834,
      "step": 6102
    },
    {
      "epoch": 0.8357984113941386,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008689879041396506,
      "loss": 5.0654,
      "step": 6103
    },
    {
      "epoch": 0.8359353601752945,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008689729580429581,
      "loss": 5.0308,
      "step": 6104
    },
    {
      "epoch": 0.8360723089564502,
      "grad_norm": 0.390625,
      "learning_rate": 0.000868958008474137,
      "loss": 4.9679,
      "step": 6105
    },
    {
      "epoch": 0.8362092577376061,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008689430554333111,
      "loss": 5.0046,
      "step": 6106
    },
    {
      "epoch": 0.836346206518762,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008689280989206042,
      "loss": 5.0016,
      "step": 6107
    },
    {
      "epoch": 0.8364831552999178,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008689131389361404,
      "loss": 4.9879,
      "step": 6108
    },
    {
      "epoch": 0.8366201040810737,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008688981754800434,
      "loss": 5.0251,
      "step": 6109
    },
    {
      "epoch": 0.8367570528622296,
      "grad_norm": 0.5,
      "learning_rate": 0.0008688832085524375,
      "loss": 4.9679,
      "step": 6110
    },
    {
      "epoch": 0.8368940016433853,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008688682381534466,
      "loss": 5.0023,
      "step": 6111
    },
    {
      "epoch": 0.8370309504245412,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008688532642831949,
      "loss": 4.9848,
      "step": 6112
    },
    {
      "epoch": 0.8371678992056971,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008688382869418064,
      "loss": 4.9727,
      "step": 6113
    },
    {
      "epoch": 0.8373048479868529,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008688233061294051,
      "loss": 5.0184,
      "step": 6114
    },
    {
      "epoch": 0.8374417967680088,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008688083218461154,
      "loss": 4.9382,
      "step": 6115
    },
    {
      "epoch": 0.8375787455491646,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008687933340920612,
      "loss": 4.9933,
      "step": 6116
    },
    {
      "epoch": 0.8377156943303204,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008687783428673669,
      "loss": 5.0375,
      "step": 6117
    },
    {
      "epoch": 0.8378526431114763,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008687633481721567,
      "loss": 4.9667,
      "step": 6118
    },
    {
      "epoch": 0.8379895918926321,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008687483500065548,
      "loss": 4.9764,
      "step": 6119
    },
    {
      "epoch": 0.838126540673788,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008687333483706857,
      "loss": 4.9958,
      "step": 6120
    },
    {
      "epoch": 0.8382634894549439,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008687183432646734,
      "loss": 4.9135,
      "step": 6121
    },
    {
      "epoch": 0.8384004382360997,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008687033346886423,
      "loss": 4.8937,
      "step": 6122
    },
    {
      "epoch": 0.8385373870172556,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008686883226427172,
      "loss": 5.0419,
      "step": 6123
    },
    {
      "epoch": 0.8386743357984114,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000868673307127022,
      "loss": 4.9851,
      "step": 6124
    },
    {
      "epoch": 0.8388112845795672,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008686582881416814,
      "loss": 5.0055,
      "step": 6125
    },
    {
      "epoch": 0.8389482333607231,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008686432656868199,
      "loss": 5.0525,
      "step": 6126
    },
    {
      "epoch": 0.8390851821418789,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008686282397625617,
      "loss": 4.9645,
      "step": 6127
    },
    {
      "epoch": 0.8392221309230348,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008686132103690316,
      "loss": 4.9365,
      "step": 6128
    },
    {
      "epoch": 0.8393590797041907,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008685981775063541,
      "loss": 5.0067,
      "step": 6129
    },
    {
      "epoch": 0.8394960284853464,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008685831411746538,
      "loss": 4.9863,
      "step": 6130
    },
    {
      "epoch": 0.8396329772665023,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008685681013740552,
      "loss": 4.967,
      "step": 6131
    },
    {
      "epoch": 0.8397699260476582,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008685530581046831,
      "loss": 4.9343,
      "step": 6132
    },
    {
      "epoch": 0.839906874828814,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008685380113666619,
      "loss": 5.0063,
      "step": 6133
    },
    {
      "epoch": 0.8400438236099699,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008685229611601166,
      "loss": 4.9803,
      "step": 6134
    },
    {
      "epoch": 0.8401807723911258,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008685079074851717,
      "loss": 4.9164,
      "step": 6135
    },
    {
      "epoch": 0.8403177211722815,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008684928503419521,
      "loss": 4.8987,
      "step": 6136
    },
    {
      "epoch": 0.8404546699534374,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008684777897305827,
      "loss": 4.9419,
      "step": 6137
    },
    {
      "epoch": 0.8405916187345933,
      "grad_norm": 0.375,
      "learning_rate": 0.0008684627256511879,
      "loss": 4.9033,
      "step": 6138
    },
    {
      "epoch": 0.8407285675157491,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008684476581038928,
      "loss": 5.029,
      "step": 6139
    },
    {
      "epoch": 0.840865516296905,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008684325870888224,
      "loss": 4.9814,
      "step": 6140
    },
    {
      "epoch": 0.8410024650780608,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008684175126061012,
      "loss": 5.0091,
      "step": 6141
    },
    {
      "epoch": 0.8411394138592166,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008684024346558545,
      "loss": 4.9492,
      "step": 6142
    },
    {
      "epoch": 0.8412763626403725,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008683873532382072,
      "loss": 5.0916,
      "step": 6143
    },
    {
      "epoch": 0.8414133114215283,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008683722683532841,
      "loss": 4.984,
      "step": 6144
    },
    {
      "epoch": 0.8415502602026842,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008683571800012104,
      "loss": 4.9688,
      "step": 6145
    },
    {
      "epoch": 0.8416872089838401,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008683420881821111,
      "loss": 4.9673,
      "step": 6146
    },
    {
      "epoch": 0.8418241577649959,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000868326992896111,
      "loss": 5.0017,
      "step": 6147
    },
    {
      "epoch": 0.8419611065461517,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008683118941433356,
      "loss": 5.0622,
      "step": 6148
    },
    {
      "epoch": 0.8420980553273076,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008682967919239098,
      "loss": 5.0365,
      "step": 6149
    },
    {
      "epoch": 0.8422350041084634,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008682816862379589,
      "loss": 5.0123,
      "step": 6150
    },
    {
      "epoch": 0.8423719528896193,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000868266577085608,
      "loss": 5.0231,
      "step": 6151
    },
    {
      "epoch": 0.8425089016707751,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008682514644669821,
      "loss": 4.9533,
      "step": 6152
    },
    {
      "epoch": 0.842645850451931,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008682363483822067,
      "loss": 5.0412,
      "step": 6153
    },
    {
      "epoch": 0.8427827992330869,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008682212288314072,
      "loss": 5.0292,
      "step": 6154
    },
    {
      "epoch": 0.8429197480142426,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008682061058147087,
      "loss": 5.0733,
      "step": 6155
    },
    {
      "epoch": 0.8430566967953985,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008681909793322365,
      "loss": 5.0139,
      "step": 6156
    },
    {
      "epoch": 0.8431936455765544,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000868175849384116,
      "loss": 5.0658,
      "step": 6157
    },
    {
      "epoch": 0.8433305943577102,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008681607159704726,
      "loss": 5.0533,
      "step": 6158
    },
    {
      "epoch": 0.8434675431388661,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008681455790914317,
      "loss": 5.0468,
      "step": 6159
    },
    {
      "epoch": 0.843604491920022,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008681304387471187,
      "loss": 5.0038,
      "step": 6160
    },
    {
      "epoch": 0.8437414407011777,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008681152949376591,
      "loss": 4.9617,
      "step": 6161
    },
    {
      "epoch": 0.8438783894823336,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008681001476631785,
      "loss": 4.9084,
      "step": 6162
    },
    {
      "epoch": 0.8440153382634894,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008680849969238023,
      "loss": 4.9718,
      "step": 6163
    },
    {
      "epoch": 0.8441522870446453,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008680698427196561,
      "loss": 5.034,
      "step": 6164
    },
    {
      "epoch": 0.8442892358258012,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008680546850508656,
      "loss": 5.0462,
      "step": 6165
    },
    {
      "epoch": 0.844426184606957,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008680395239175562,
      "loss": 4.95,
      "step": 6166
    },
    {
      "epoch": 0.8445631333881128,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008680243593198536,
      "loss": 5.0045,
      "step": 6167
    },
    {
      "epoch": 0.8447000821692687,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008680091912578835,
      "loss": 5.0738,
      "step": 6168
    },
    {
      "epoch": 0.8448370309504245,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008679940197317719,
      "loss": 4.919,
      "step": 6169
    },
    {
      "epoch": 0.8449739797315804,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008679788447416441,
      "loss": 4.9416,
      "step": 6170
    },
    {
      "epoch": 0.8451109285127363,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008679636662876259,
      "loss": 5.073,
      "step": 6171
    },
    {
      "epoch": 0.8452478772938921,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008679484843698433,
      "loss": 4.9251,
      "step": 6172
    },
    {
      "epoch": 0.845384826075048,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000867933298988422,
      "loss": 4.9554,
      "step": 6173
    },
    {
      "epoch": 0.8455217748562038,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008679181101434879,
      "loss": 5.0586,
      "step": 6174
    },
    {
      "epoch": 0.8456587236373596,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008679029178351669,
      "loss": 5.0421,
      "step": 6175
    },
    {
      "epoch": 0.8457956724185155,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008678877220635846,
      "loss": 4.9592,
      "step": 6176
    },
    {
      "epoch": 0.8459326211996713,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008678725228288673,
      "loss": 4.9615,
      "step": 6177
    },
    {
      "epoch": 0.8460695699808272,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000867857320131141,
      "loss": 5.0037,
      "step": 6178
    },
    {
      "epoch": 0.846206518761983,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008678421139705313,
      "loss": 4.9575,
      "step": 6179
    },
    {
      "epoch": 0.8463434675431388,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008678269043471646,
      "loss": 4.9871,
      "step": 6180
    },
    {
      "epoch": 0.8464804163242947,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008678116912611666,
      "loss": 5.0867,
      "step": 6181
    },
    {
      "epoch": 0.8466173651054506,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008677964747126636,
      "loss": 5.1196,
      "step": 6182
    },
    {
      "epoch": 0.8467543138866064,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008677812547017818,
      "loss": 4.9804,
      "step": 6183
    },
    {
      "epoch": 0.8468912626677623,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008677660312286471,
      "loss": 4.942,
      "step": 6184
    },
    {
      "epoch": 0.8470282114489182,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008677508042933857,
      "loss": 4.9817,
      "step": 6185
    },
    {
      "epoch": 0.8471651602300739,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008677355738961239,
      "loss": 4.9266,
      "step": 6186
    },
    {
      "epoch": 0.8473021090112298,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008677203400369879,
      "loss": 4.9654,
      "step": 6187
    },
    {
      "epoch": 0.8474390577923856,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008677051027161038,
      "loss": 4.9558,
      "step": 6188
    },
    {
      "epoch": 0.8475760065735415,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000867689861933598,
      "loss": 5.0571,
      "step": 6189
    },
    {
      "epoch": 0.8477129553546974,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008676746176895968,
      "loss": 5.01,
      "step": 6190
    },
    {
      "epoch": 0.8478499041358531,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008676593699842266,
      "loss": 4.9611,
      "step": 6191
    },
    {
      "epoch": 0.847986852917009,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008676441188176136,
      "loss": 5.0576,
      "step": 6192
    },
    {
      "epoch": 0.8481238016981649,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008676288641898842,
      "loss": 4.9763,
      "step": 6193
    },
    {
      "epoch": 0.8482607504793207,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000867613606101165,
      "loss": 5.0348,
      "step": 6194
    },
    {
      "epoch": 0.8483976992604766,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008675983445515823,
      "loss": 4.9343,
      "step": 6195
    },
    {
      "epoch": 0.8485346480416325,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008675830795412626,
      "loss": 5.0796,
      "step": 6196
    },
    {
      "epoch": 0.8486715968227883,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008675678110703324,
      "loss": 5.0542,
      "step": 6197
    },
    {
      "epoch": 0.8488085456039441,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008675525391389184,
      "loss": 5.0292,
      "step": 6198
    },
    {
      "epoch": 0.8489454943850999,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008675372637471468,
      "loss": 4.9491,
      "step": 6199
    },
    {
      "epoch": 0.8490824431662558,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008675219848951445,
      "loss": 4.9484,
      "step": 6200
    },
    {
      "epoch": 0.8492193919474117,
      "grad_norm": 0.40625,
      "learning_rate": 0.000867506702583038,
      "loss": 4.8999,
      "step": 6201
    },
    {
      "epoch": 0.8493563407285675,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000867491416810954,
      "loss": 4.9794,
      "step": 6202
    },
    {
      "epoch": 0.8494932895097234,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008674761275790191,
      "loss": 5.0358,
      "step": 6203
    },
    {
      "epoch": 0.8496302382908792,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008674608348873601,
      "loss": 4.9387,
      "step": 6204
    },
    {
      "epoch": 0.849767187072035,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008674455387361036,
      "loss": 4.9833,
      "step": 6205
    },
    {
      "epoch": 0.8499041358531909,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008674302391253765,
      "loss": 4.9791,
      "step": 6206
    },
    {
      "epoch": 0.8500410846343468,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008674149360553056,
      "loss": 4.9761,
      "step": 6207
    },
    {
      "epoch": 0.8501780334155026,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008673996295260175,
      "loss": 5.0029,
      "step": 6208
    },
    {
      "epoch": 0.8503149821966585,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008673843195376394,
      "loss": 5.0382,
      "step": 6209
    },
    {
      "epoch": 0.8504519309778144,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008673690060902979,
      "loss": 5.042,
      "step": 6210
    },
    {
      "epoch": 0.8505888797589701,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008673536891841199,
      "loss": 5.0104,
      "step": 6211
    },
    {
      "epoch": 0.850725828540126,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008673383688192326,
      "loss": 5.0846,
      "step": 6212
    },
    {
      "epoch": 0.8508627773212818,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008673230449957627,
      "loss": 4.9554,
      "step": 6213
    },
    {
      "epoch": 0.8509997261024377,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008673077177138373,
      "loss": 5.0118,
      "step": 6214
    },
    {
      "epoch": 0.8511366748835936,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008672923869735834,
      "loss": 4.989,
      "step": 6215
    },
    {
      "epoch": 0.8512736236647493,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008672770527751281,
      "loss": 4.9478,
      "step": 6216
    },
    {
      "epoch": 0.8514105724459052,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008672617151185983,
      "loss": 5.0243,
      "step": 6217
    },
    {
      "epoch": 0.8515475212270611,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008672463740041213,
      "loss": 4.9572,
      "step": 6218
    },
    {
      "epoch": 0.8516844700082169,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008672310294318243,
      "loss": 5.0488,
      "step": 6219
    },
    {
      "epoch": 0.8518214187893728,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008672156814018342,
      "loss": 4.9278,
      "step": 6220
    },
    {
      "epoch": 0.8519583675705287,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008672003299142785,
      "loss": 4.9915,
      "step": 6221
    },
    {
      "epoch": 0.8520953163516845,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008671849749692839,
      "loss": 5.0299,
      "step": 6222
    },
    {
      "epoch": 0.8522322651328403,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008671696165669783,
      "loss": 5.0632,
      "step": 6223
    },
    {
      "epoch": 0.8523692139139961,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008671542547074886,
      "loss": 4.9996,
      "step": 6224
    },
    {
      "epoch": 0.852506162695152,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008671388893909422,
      "loss": 4.9883,
      "step": 6225
    },
    {
      "epoch": 0.8526431114763079,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008671235206174665,
      "loss": 5.0029,
      "step": 6226
    },
    {
      "epoch": 0.8527800602574637,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008671081483871886,
      "loss": 5.0782,
      "step": 6227
    },
    {
      "epoch": 0.8529170090386196,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008670927727002362,
      "loss": 5.0304,
      "step": 6228
    },
    {
      "epoch": 0.8530539578197754,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008670773935567366,
      "loss": 5.0118,
      "step": 6229
    },
    {
      "epoch": 0.8531909066009312,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008670620109568173,
      "loss": 4.9566,
      "step": 6230
    },
    {
      "epoch": 0.8533278553820871,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008670466249006057,
      "loss": 4.9794,
      "step": 6231
    },
    {
      "epoch": 0.853464804163243,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008670312353882293,
      "loss": 4.9911,
      "step": 6232
    },
    {
      "epoch": 0.8536017529443988,
      "grad_norm": 0.375,
      "learning_rate": 0.0008670158424198157,
      "loss": 5.0305,
      "step": 6233
    },
    {
      "epoch": 0.8537387017255547,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008670004459954924,
      "loss": 4.9631,
      "step": 6234
    },
    {
      "epoch": 0.8538756505067104,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008669850461153872,
      "loss": 4.9696,
      "step": 6235
    },
    {
      "epoch": 0.8540125992878663,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008669696427796274,
      "loss": 5.0074,
      "step": 6236
    },
    {
      "epoch": 0.8541495480690222,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008669542359883409,
      "loss": 4.9431,
      "step": 6237
    },
    {
      "epoch": 0.854286496850178,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008669388257416552,
      "loss": 5.0331,
      "step": 6238
    },
    {
      "epoch": 0.8544234456313339,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008669234120396982,
      "loss": 4.9237,
      "step": 6239
    },
    {
      "epoch": 0.8545603944124898,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008669079948825975,
      "loss": 4.9908,
      "step": 6240
    },
    {
      "epoch": 0.8546973431936455,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008668925742704809,
      "loss": 5.0159,
      "step": 6241
    },
    {
      "epoch": 0.8548342919748014,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008668771502034761,
      "loss": 4.9519,
      "step": 6242
    },
    {
      "epoch": 0.8549712407559573,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008668617226817112,
      "loss": 5.0111,
      "step": 6243
    },
    {
      "epoch": 0.8551081895371131,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008668462917053137,
      "loss": 4.9336,
      "step": 6244
    },
    {
      "epoch": 0.855245138318269,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008668308572744118,
      "loss": 5.037,
      "step": 6245
    },
    {
      "epoch": 0.8553820870994249,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008668154193891332,
      "loss": 5.0555,
      "step": 6246
    },
    {
      "epoch": 0.8555190358805806,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000866799978049606,
      "loss": 4.9451,
      "step": 6247
    },
    {
      "epoch": 0.8556559846617365,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008667845332559578,
      "loss": 4.9834,
      "step": 6248
    },
    {
      "epoch": 0.8557929334428923,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008667690850083171,
      "loss": 4.954,
      "step": 6249
    },
    {
      "epoch": 0.8559298822240482,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008667536333068116,
      "loss": 4.968,
      "step": 6250
    },
    {
      "epoch": 0.8560668310052041,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008667381781515694,
      "loss": 4.9643,
      "step": 6251
    },
    {
      "epoch": 0.8562037797863599,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008667227195427186,
      "loss": 4.9711,
      "step": 6252
    },
    {
      "epoch": 0.8563407285675158,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008667072574803874,
      "loss": 4.985,
      "step": 6253
    },
    {
      "epoch": 0.8564776773486716,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008666917919647038,
      "loss": 4.9516,
      "step": 6254
    },
    {
      "epoch": 0.8566146261298274,
      "grad_norm": 0.453125,
      "learning_rate": 0.000866676322995796,
      "loss": 4.9445,
      "step": 6255
    },
    {
      "epoch": 0.8567515749109833,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008666608505737924,
      "loss": 4.9883,
      "step": 6256
    },
    {
      "epoch": 0.8568885236921392,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008666453746988209,
      "loss": 4.9551,
      "step": 6257
    },
    {
      "epoch": 0.857025472473295,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008666298953710099,
      "loss": 5.0056,
      "step": 6258
    },
    {
      "epoch": 0.8571624212544509,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008666144125904877,
      "loss": 4.9485,
      "step": 6259
    },
    {
      "epoch": 0.8572993700356066,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008665989263573826,
      "loss": 4.9969,
      "step": 6260
    },
    {
      "epoch": 0.8574363188167625,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000866583436671823,
      "loss": 5.0556,
      "step": 6261
    },
    {
      "epoch": 0.8575732675979184,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008665679435339371,
      "loss": 4.9934,
      "step": 6262
    },
    {
      "epoch": 0.8577102163790742,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008665524469438534,
      "loss": 5.0064,
      "step": 6263
    },
    {
      "epoch": 0.8578471651602301,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008665369469017003,
      "loss": 5.0001,
      "step": 6264
    },
    {
      "epoch": 0.857984113941386,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008665214434076063,
      "loss": 4.9934,
      "step": 6265
    },
    {
      "epoch": 0.8581210627225417,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008665059364616998,
      "loss": 5.015,
      "step": 6266
    },
    {
      "epoch": 0.8582580115036976,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008664904260641095,
      "loss": 5.0572,
      "step": 6267
    },
    {
      "epoch": 0.8583949602848535,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008664749122149635,
      "loss": 5.0239,
      "step": 6268
    },
    {
      "epoch": 0.8585319090660093,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000866459394914391,
      "loss": 4.9784,
      "step": 6269
    },
    {
      "epoch": 0.8586688578471652,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00086644387416252,
      "loss": 4.9617,
      "step": 6270
    },
    {
      "epoch": 0.858805806628321,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008664283499594795,
      "loss": 4.9588,
      "step": 6271
    },
    {
      "epoch": 0.8589427554094768,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000866412822305398,
      "loss": 5.0007,
      "step": 6272
    },
    {
      "epoch": 0.8590797041906327,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008663972912004043,
      "loss": 4.9755,
      "step": 6273
    },
    {
      "epoch": 0.8592166529717885,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008663817566446269,
      "loss": 4.9748,
      "step": 6274
    },
    {
      "epoch": 0.8593536017529444,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008663662186381948,
      "loss": 4.9916,
      "step": 6275
    },
    {
      "epoch": 0.8594905505341003,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008663506771812365,
      "loss": 5.0437,
      "step": 6276
    },
    {
      "epoch": 0.8596274993152561,
      "grad_norm": 0.4375,
      "learning_rate": 0.000866335132273881,
      "loss": 5.0101,
      "step": 6277
    },
    {
      "epoch": 0.859764448096412,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008663195839162571,
      "loss": 5.0159,
      "step": 6278
    },
    {
      "epoch": 0.8599013968775678,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008663040321084935,
      "loss": 5.0232,
      "step": 6279
    },
    {
      "epoch": 0.8600383456587236,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008662884768507193,
      "loss": 4.927,
      "step": 6280
    },
    {
      "epoch": 0.8601752944398795,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008662729181430632,
      "loss": 4.9177,
      "step": 6281
    },
    {
      "epoch": 0.8603122432210353,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008662573559856542,
      "loss": 4.9734,
      "step": 6282
    },
    {
      "epoch": 0.8604491920021912,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008662417903786214,
      "loss": 5.0448,
      "step": 6283
    },
    {
      "epoch": 0.860586140783347,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008662262213220936,
      "loss": 4.8853,
      "step": 6284
    },
    {
      "epoch": 0.8607230895645028,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008662106488162,
      "loss": 4.9581,
      "step": 6285
    },
    {
      "epoch": 0.8608600383456587,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008661950728610696,
      "loss": 4.948,
      "step": 6286
    },
    {
      "epoch": 0.8609969871268146,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008661794934568313,
      "loss": 5.025,
      "step": 6287
    },
    {
      "epoch": 0.8611339359079704,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008661639106036145,
      "loss": 5.0028,
      "step": 6288
    },
    {
      "epoch": 0.8612708846891263,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008661483243015483,
      "loss": 4.9753,
      "step": 6289
    },
    {
      "epoch": 0.8614078334702822,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008661327345507615,
      "loss": 4.9926,
      "step": 6290
    },
    {
      "epoch": 0.8615447822514379,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008661171413513838,
      "loss": 4.9583,
      "step": 6291
    },
    {
      "epoch": 0.8616817310325938,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008661015447035441,
      "loss": 4.9711,
      "step": 6292
    },
    {
      "epoch": 0.8618186798137497,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008660859446073718,
      "loss": 4.932,
      "step": 6293
    },
    {
      "epoch": 0.8619556285949055,
      "grad_norm": 0.470703125,
      "learning_rate": 0.000866070341062996,
      "loss": 4.9341,
      "step": 6294
    },
    {
      "epoch": 0.8620925773760614,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008660547340705463,
      "loss": 5.0071,
      "step": 6295
    },
    {
      "epoch": 0.8622295261572172,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008660391236301518,
      "loss": 4.929,
      "step": 6296
    },
    {
      "epoch": 0.862366474938373,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008660235097419421,
      "loss": 5.0284,
      "step": 6297
    },
    {
      "epoch": 0.8625034237195289,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008660078924060462,
      "loss": 5.0072,
      "step": 6298
    },
    {
      "epoch": 0.8626403725006847,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008659922716225938,
      "loss": 4.9662,
      "step": 6299
    },
    {
      "epoch": 0.8627773212818406,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008659766473917145,
      "loss": 4.9262,
      "step": 6300
    },
    {
      "epoch": 0.8629142700629965,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008659610197135374,
      "loss": 5.0113,
      "step": 6301
    },
    {
      "epoch": 0.8630512188441523,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008659453885881923,
      "loss": 4.9869,
      "step": 6302
    },
    {
      "epoch": 0.8631881676253081,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008659297540158088,
      "loss": 4.9338,
      "step": 6303
    },
    {
      "epoch": 0.863325116406464,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008659141159965161,
      "loss": 4.9888,
      "step": 6304
    },
    {
      "epoch": 0.8634620651876198,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008658984745304442,
      "loss": 4.9986,
      "step": 6305
    },
    {
      "epoch": 0.8635990139687757,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008658828296177223,
      "loss": 4.9742,
      "step": 6306
    },
    {
      "epoch": 0.8637359627499315,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008658671812584804,
      "loss": 5.0961,
      "step": 6307
    },
    {
      "epoch": 0.8638729115310874,
      "grad_norm": 0.375,
      "learning_rate": 0.000865851529452848,
      "loss": 4.9884,
      "step": 6308
    },
    {
      "epoch": 0.8640098603122432,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008658358742009551,
      "loss": 5.0911,
      "step": 6309
    },
    {
      "epoch": 0.864146809093399,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000865820215502931,
      "loss": 4.9919,
      "step": 6310
    },
    {
      "epoch": 0.8642837578745549,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008658045533589059,
      "loss": 4.9314,
      "step": 6311
    },
    {
      "epoch": 0.8644207066557108,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008657888877690092,
      "loss": 4.9561,
      "step": 6312
    },
    {
      "epoch": 0.8645576554368666,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000865773218733371,
      "loss": 4.9115,
      "step": 6313
    },
    {
      "epoch": 0.8646946042180225,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000865757546252121,
      "loss": 4.9948,
      "step": 6314
    },
    {
      "epoch": 0.8648315529991784,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008657418703253893,
      "loss": 5.0094,
      "step": 6315
    },
    {
      "epoch": 0.8649685017803341,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008657261909533054,
      "loss": 5.0329,
      "step": 6316
    },
    {
      "epoch": 0.86510545056149,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008657105081359998,
      "loss": 4.9922,
      "step": 6317
    },
    {
      "epoch": 0.8652423993426458,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008656948218736019,
      "loss": 4.9747,
      "step": 6318
    },
    {
      "epoch": 0.8653793481238017,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008656791321662421,
      "loss": 4.957,
      "step": 6319
    },
    {
      "epoch": 0.8655162969049576,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008656634390140502,
      "loss": 4.9249,
      "step": 6320
    },
    {
      "epoch": 0.8656532456861133,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008656477424171564,
      "loss": 5.0787,
      "step": 6321
    },
    {
      "epoch": 0.8657901944672692,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008656320423756906,
      "loss": 4.926,
      "step": 6322
    },
    {
      "epoch": 0.8659271432484251,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008656163388897831,
      "loss": 4.9433,
      "step": 6323
    },
    {
      "epoch": 0.8660640920295809,
      "grad_norm": 0.375,
      "learning_rate": 0.0008656006319595641,
      "loss": 5.0972,
      "step": 6324
    },
    {
      "epoch": 0.8662010408107368,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008655849215851635,
      "loss": 4.9481,
      "step": 6325
    },
    {
      "epoch": 0.8663379895918927,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008655692077667116,
      "loss": 5.0384,
      "step": 6326
    },
    {
      "epoch": 0.8664749383730485,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008655534905043387,
      "loss": 4.9908,
      "step": 6327
    },
    {
      "epoch": 0.8666118871542043,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000865537769798175,
      "loss": 4.9541,
      "step": 6328
    },
    {
      "epoch": 0.8667488359353602,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008655220456483506,
      "loss": 5.0571,
      "step": 6329
    },
    {
      "epoch": 0.866885784716516,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008655063180549962,
      "loss": 5.0371,
      "step": 6330
    },
    {
      "epoch": 0.8670227334976719,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000865490587018242,
      "loss": 4.9905,
      "step": 6331
    },
    {
      "epoch": 0.8671596822788277,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008654748525382182,
      "loss": 4.9891,
      "step": 6332
    },
    {
      "epoch": 0.8672966310599836,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008654591146150551,
      "loss": 5.0364,
      "step": 6333
    },
    {
      "epoch": 0.8674335798411394,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008654433732488837,
      "loss": 5.0059,
      "step": 6334
    },
    {
      "epoch": 0.8675705286222952,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008654276284398339,
      "loss": 4.9891,
      "step": 6335
    },
    {
      "epoch": 0.8677074774034511,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008654118801880362,
      "loss": 4.9265,
      "step": 6336
    },
    {
      "epoch": 0.867844426184607,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008653961284936214,
      "loss": 4.987,
      "step": 6337
    },
    {
      "epoch": 0.8679813749657628,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008653803733567199,
      "loss": 5.0428,
      "step": 6338
    },
    {
      "epoch": 0.8681183237469187,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008653646147774622,
      "loss": 4.975,
      "step": 6339
    },
    {
      "epoch": 0.8682552725280746,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008653488527559792,
      "loss": 4.8936,
      "step": 6340
    },
    {
      "epoch": 0.8683922213092303,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000865333087292401,
      "loss": 4.9277,
      "step": 6341
    },
    {
      "epoch": 0.8685291700903862,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008653173183868586,
      "loss": 4.968,
      "step": 6342
    },
    {
      "epoch": 0.868666118871542,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008653015460394827,
      "loss": 4.9614,
      "step": 6343
    },
    {
      "epoch": 0.8688030676526979,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008652857702504039,
      "loss": 4.8971,
      "step": 6344
    },
    {
      "epoch": 0.8689400164338538,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008652699910197529,
      "loss": 4.9758,
      "step": 6345
    },
    {
      "epoch": 0.8690769652150095,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008652542083476606,
      "loss": 5.0097,
      "step": 6346
    },
    {
      "epoch": 0.8692139139961654,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008652384222342577,
      "loss": 5.0066,
      "step": 6347
    },
    {
      "epoch": 0.8693508627773213,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000865222632679675,
      "loss": 4.9469,
      "step": 6348
    },
    {
      "epoch": 0.8694878115584771,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008652068396840435,
      "loss": 4.9493,
      "step": 6349
    },
    {
      "epoch": 0.869624760339633,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000865191043247494,
      "loss": 4.9708,
      "step": 6350
    },
    {
      "epoch": 0.8697617091207889,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008651752433701573,
      "loss": 5.014,
      "step": 6351
    },
    {
      "epoch": 0.8698986579019446,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008651594400521644,
      "loss": 5.021,
      "step": 6352
    },
    {
      "epoch": 0.8700356066831005,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008651436332936463,
      "loss": 5.021,
      "step": 6353
    },
    {
      "epoch": 0.8701725554642563,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008651278230947341,
      "loss": 5.0218,
      "step": 6354
    },
    {
      "epoch": 0.8703095042454122,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008651120094555586,
      "loss": 4.9699,
      "step": 6355
    },
    {
      "epoch": 0.8704464530265681,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008650961923762511,
      "loss": 4.924,
      "step": 6356
    },
    {
      "epoch": 0.8705834018077239,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008650803718569424,
      "loss": 4.9563,
      "step": 6357
    },
    {
      "epoch": 0.8707203505888798,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008650645478977639,
      "loss": 5.0262,
      "step": 6358
    },
    {
      "epoch": 0.8708572993700356,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008650487204988464,
      "loss": 4.974,
      "step": 6359
    },
    {
      "epoch": 0.8709942481511914,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008650328896603213,
      "loss": 4.9679,
      "step": 6360
    },
    {
      "epoch": 0.8711311969323473,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008650170553823199,
      "loss": 4.9408,
      "step": 6361
    },
    {
      "epoch": 0.8712681457135032,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008650012176649732,
      "loss": 4.9564,
      "step": 6362
    },
    {
      "epoch": 0.871405094494659,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008649853765084123,
      "loss": 5.0003,
      "step": 6363
    },
    {
      "epoch": 0.8715420432758149,
      "grad_norm": 0.486328125,
      "learning_rate": 0.000864969531912769,
      "loss": 5.0175,
      "step": 6364
    },
    {
      "epoch": 0.8716789920569707,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000864953683878174,
      "loss": 5.0522,
      "step": 6365
    },
    {
      "epoch": 0.8718159408381265,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008649378324047592,
      "loss": 4.9644,
      "step": 6366
    },
    {
      "epoch": 0.8719528896192824,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008649219774926556,
      "loss": 4.9346,
      "step": 6367
    },
    {
      "epoch": 0.8720898384004382,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008649061191419948,
      "loss": 5.1576,
      "step": 6368
    },
    {
      "epoch": 0.8722267871815941,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008648902573529081,
      "loss": 5.068,
      "step": 6369
    },
    {
      "epoch": 0.87236373596275,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008648743921255269,
      "loss": 5.0484,
      "step": 6370
    },
    {
      "epoch": 0.8725006847439057,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008648585234599827,
      "loss": 4.9753,
      "step": 6371
    },
    {
      "epoch": 0.8726376335250616,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008648426513564073,
      "loss": 4.9386,
      "step": 6372
    },
    {
      "epoch": 0.8727745823062175,
      "grad_norm": 0.375,
      "learning_rate": 0.0008648267758149318,
      "loss": 5.0141,
      "step": 6373
    },
    {
      "epoch": 0.8729115310873733,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000864810896835688,
      "loss": 4.9974,
      "step": 6374
    },
    {
      "epoch": 0.8730484798685292,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008647950144188074,
      "loss": 4.986,
      "step": 6375
    },
    {
      "epoch": 0.8731854286496851,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008647791285644218,
      "loss": 4.9479,
      "step": 6376
    },
    {
      "epoch": 0.8733223774308408,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008647632392726626,
      "loss": 4.9937,
      "step": 6377
    },
    {
      "epoch": 0.8734593262119967,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008647473465436617,
      "loss": 4.9572,
      "step": 6378
    },
    {
      "epoch": 0.8735962749931525,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008647314503775506,
      "loss": 4.9373,
      "step": 6379
    },
    {
      "epoch": 0.8737332237743084,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008647155507744613,
      "loss": 4.9371,
      "step": 6380
    },
    {
      "epoch": 0.8738701725554643,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008646996477345253,
      "loss": 4.9568,
      "step": 6381
    },
    {
      "epoch": 0.8740071213366201,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008646837412578745,
      "loss": 4.9432,
      "step": 6382
    },
    {
      "epoch": 0.874144070117776,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008646678313446409,
      "loss": 4.9413,
      "step": 6383
    },
    {
      "epoch": 0.8742810188989318,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008646519179949559,
      "loss": 4.986,
      "step": 6384
    },
    {
      "epoch": 0.8744179676800876,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008646360012089518,
      "loss": 4.9671,
      "step": 6385
    },
    {
      "epoch": 0.8745549164612435,
      "grad_norm": 0.375,
      "learning_rate": 0.0008646200809867603,
      "loss": 4.9755,
      "step": 6386
    },
    {
      "epoch": 0.8746918652423994,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008646041573285134,
      "loss": 4.9498,
      "step": 6387
    },
    {
      "epoch": 0.8748288140235552,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008645882302343431,
      "loss": 4.9808,
      "step": 6388
    },
    {
      "epoch": 0.8749657628047111,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008645722997043813,
      "loss": 5.0618,
      "step": 6389
    },
    {
      "epoch": 0.8751027115858668,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008645563657387601,
      "loss": 4.9105,
      "step": 6390
    },
    {
      "epoch": 0.8752396603670227,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008645404283376115,
      "loss": 5.0203,
      "step": 6391
    },
    {
      "epoch": 0.8753766091481786,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008645244875010675,
      "loss": 4.9456,
      "step": 6392
    },
    {
      "epoch": 0.8755135579293344,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008645085432292605,
      "loss": 4.9292,
      "step": 6393
    },
    {
      "epoch": 0.8756505067104903,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008644925955223222,
      "loss": 5.0903,
      "step": 6394
    },
    {
      "epoch": 0.8757874554916462,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008644766443803851,
      "loss": 4.9774,
      "step": 6395
    },
    {
      "epoch": 0.8759244042728019,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008644606898035814,
      "loss": 5.003,
      "step": 6396
    },
    {
      "epoch": 0.8760613530539578,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000864444731792043,
      "loss": 5.0368,
      "step": 6397
    },
    {
      "epoch": 0.8761983018351137,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008644287703459025,
      "loss": 4.9147,
      "step": 6398
    },
    {
      "epoch": 0.8763352506162695,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000864412805465292,
      "loss": 5.0071,
      "step": 6399
    },
    {
      "epoch": 0.8764721993974254,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008643968371503438,
      "loss": 4.9671,
      "step": 6400
    },
    {
      "epoch": 0.8766091481785813,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008643808654011903,
      "loss": 5.0162,
      "step": 6401
    },
    {
      "epoch": 0.876746096959737,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008643648902179636,
      "loss": 5.0082,
      "step": 6402
    },
    {
      "epoch": 0.8768830457408929,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008643489116007967,
      "loss": 5.045,
      "step": 6403
    },
    {
      "epoch": 0.8770199945220487,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008643329295498214,
      "loss": 5.0288,
      "step": 6404
    },
    {
      "epoch": 0.8771569433032046,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008643169440651705,
      "loss": 5.0146,
      "step": 6405
    },
    {
      "epoch": 0.8772938920843605,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008643009551469762,
      "loss": 5.0301,
      "step": 6406
    },
    {
      "epoch": 0.8774308408655163,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008642849627953712,
      "loss": 4.9371,
      "step": 6407
    },
    {
      "epoch": 0.8775677896466721,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008642689670104881,
      "loss": 4.9519,
      "step": 6408
    },
    {
      "epoch": 0.877704738427828,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008642529677924594,
      "loss": 5.0159,
      "step": 6409
    },
    {
      "epoch": 0.8778416872089838,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008642369651414175,
      "loss": 5.0728,
      "step": 6410
    },
    {
      "epoch": 0.8779786359901397,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008642209590574952,
      "loss": 4.9303,
      "step": 6411
    },
    {
      "epoch": 0.8781155847712956,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000864204949540825,
      "loss": 5.0091,
      "step": 6412
    },
    {
      "epoch": 0.8782525335524514,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008641889365915399,
      "loss": 5.0242,
      "step": 6413
    },
    {
      "epoch": 0.8783894823336073,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008641729202097723,
      "loss": 4.9141,
      "step": 6414
    },
    {
      "epoch": 0.878526431114763,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000864156900395655,
      "loss": 5.0475,
      "step": 6415
    },
    {
      "epoch": 0.8786633798959189,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008641408771493207,
      "loss": 5.0607,
      "step": 6416
    },
    {
      "epoch": 0.8788003286770748,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008641248504709024,
      "loss": 4.919,
      "step": 6417
    },
    {
      "epoch": 0.8789372774582306,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008641088203605328,
      "loss": 4.9429,
      "step": 6418
    },
    {
      "epoch": 0.8790742262393865,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008640927868183446,
      "loss": 4.9605,
      "step": 6419
    },
    {
      "epoch": 0.8792111750205424,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008640767498444709,
      "loss": 4.9632,
      "step": 6420
    },
    {
      "epoch": 0.8793481238016981,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008640607094390444,
      "loss": 4.9661,
      "step": 6421
    },
    {
      "epoch": 0.879485072582854,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008640446656021982,
      "loss": 5.004,
      "step": 6422
    },
    {
      "epoch": 0.8796220213640099,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008640286183340652,
      "loss": 4.9706,
      "step": 6423
    },
    {
      "epoch": 0.8797589701451657,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008640125676347784,
      "loss": 4.9552,
      "step": 6424
    },
    {
      "epoch": 0.8798959189263216,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008639965135044707,
      "loss": 4.9613,
      "step": 6425
    },
    {
      "epoch": 0.8800328677074774,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008639804559432753,
      "loss": 4.93,
      "step": 6426
    },
    {
      "epoch": 0.8801698164886332,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008639643949513254,
      "loss": 4.9687,
      "step": 6427
    },
    {
      "epoch": 0.8803067652697891,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008639483305287538,
      "loss": 4.9598,
      "step": 6428
    },
    {
      "epoch": 0.8804437140509449,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008639322626756936,
      "loss": 4.9452,
      "step": 6429
    },
    {
      "epoch": 0.8805806628321008,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008639161913922782,
      "loss": 4.9699,
      "step": 6430
    },
    {
      "epoch": 0.8807176116132567,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008639001166786406,
      "loss": 5.012,
      "step": 6431
    },
    {
      "epoch": 0.8808545603944125,
      "grad_norm": 0.375,
      "learning_rate": 0.0008638840385349143,
      "loss": 4.9566,
      "step": 6432
    },
    {
      "epoch": 0.8809915091755683,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008638679569612322,
      "loss": 4.9709,
      "step": 6433
    },
    {
      "epoch": 0.8811284579567242,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008638518719577278,
      "loss": 4.9254,
      "step": 6434
    },
    {
      "epoch": 0.88126540673788,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008638357835245343,
      "loss": 4.9617,
      "step": 6435
    },
    {
      "epoch": 0.8814023555190359,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000863819691661785,
      "loss": 4.9569,
      "step": 6436
    },
    {
      "epoch": 0.8815393043001917,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008638035963696134,
      "loss": 5.0082,
      "step": 6437
    },
    {
      "epoch": 0.8816762530813476,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008637874976481528,
      "loss": 4.9875,
      "step": 6438
    },
    {
      "epoch": 0.8818132018625034,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008637713954975366,
      "loss": 5.0173,
      "step": 6439
    },
    {
      "epoch": 0.8819501506436592,
      "grad_norm": 0.375,
      "learning_rate": 0.0008637552899178981,
      "loss": 4.9564,
      "step": 6440
    },
    {
      "epoch": 0.8820870994248151,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000863739180909371,
      "loss": 5.0506,
      "step": 6441
    },
    {
      "epoch": 0.882224048205971,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008637230684720887,
      "loss": 5.0233,
      "step": 6442
    },
    {
      "epoch": 0.8823609969871268,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008637069526061848,
      "loss": 4.9791,
      "step": 6443
    },
    {
      "epoch": 0.8824979457682827,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008636908333117929,
      "loss": 5.0364,
      "step": 6444
    },
    {
      "epoch": 0.8826348945494386,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008636747105890464,
      "loss": 4.9681,
      "step": 6445
    },
    {
      "epoch": 0.8827718433305943,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008636585844380789,
      "loss": 4.9849,
      "step": 6446
    },
    {
      "epoch": 0.8829087921117502,
      "grad_norm": 0.375,
      "learning_rate": 0.0008636424548590242,
      "loss": 4.9639,
      "step": 6447
    },
    {
      "epoch": 0.8830457408929061,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008636263218520159,
      "loss": 4.9991,
      "step": 6448
    },
    {
      "epoch": 0.8831826896740619,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008636101854171878,
      "loss": 5.0083,
      "step": 6449
    },
    {
      "epoch": 0.8833196384552178,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008635940455546735,
      "loss": 4.9845,
      "step": 6450
    },
    {
      "epoch": 0.8834565872363735,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008635779022646068,
      "loss": 5.0234,
      "step": 6451
    },
    {
      "epoch": 0.8835935360175294,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008635617555471213,
      "loss": 4.9739,
      "step": 6452
    },
    {
      "epoch": 0.8837304847986853,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008635456054023513,
      "loss": 5.0294,
      "step": 6453
    },
    {
      "epoch": 0.8838674335798411,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008635294518304301,
      "loss": 4.9539,
      "step": 6454
    },
    {
      "epoch": 0.884004382360997,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008635132948314918,
      "loss": 4.9223,
      "step": 6455
    },
    {
      "epoch": 0.8841413311421529,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008634971344056704,
      "loss": 5.0362,
      "step": 6456
    },
    {
      "epoch": 0.8842782799233087,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008634809705530996,
      "loss": 5.0208,
      "step": 6457
    },
    {
      "epoch": 0.8844152287044645,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008634648032739134,
      "loss": 5.0213,
      "step": 6458
    },
    {
      "epoch": 0.8845521774856204,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008634486325682461,
      "loss": 5.0278,
      "step": 6459
    },
    {
      "epoch": 0.8846891262667762,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008634324584362313,
      "loss": 4.9943,
      "step": 6460
    },
    {
      "epoch": 0.8848260750479321,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008634162808780031,
      "loss": 5.048,
      "step": 6461
    },
    {
      "epoch": 0.8849630238290879,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008634000998936959,
      "loss": 4.9529,
      "step": 6462
    },
    {
      "epoch": 0.8850999726102438,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008633839154834435,
      "loss": 4.9974,
      "step": 6463
    },
    {
      "epoch": 0.8852369213913996,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00086336772764738,
      "loss": 5.0012,
      "step": 6464
    },
    {
      "epoch": 0.8853738701725554,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008633515363856398,
      "loss": 4.9543,
      "step": 6465
    },
    {
      "epoch": 0.8855108189537113,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008633353416983568,
      "loss": 4.9234,
      "step": 6466
    },
    {
      "epoch": 0.8856477677348672,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008633191435856654,
      "loss": 5.0337,
      "step": 6467
    },
    {
      "epoch": 0.885784716516023,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008633029420476997,
      "loss": 4.9321,
      "step": 6468
    },
    {
      "epoch": 0.8859216652971789,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000863286737084594,
      "loss": 4.9687,
      "step": 6469
    },
    {
      "epoch": 0.8860586140783347,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008632705286964827,
      "loss": 4.9951,
      "step": 6470
    },
    {
      "epoch": 0.8861955628594905,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008632543168835,
      "loss": 4.957,
      "step": 6471
    },
    {
      "epoch": 0.8863325116406464,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008632381016457803,
      "loss": 4.9209,
      "step": 6472
    },
    {
      "epoch": 0.8864694604218022,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008632218829834581,
      "loss": 4.9327,
      "step": 6473
    },
    {
      "epoch": 0.8866064092029581,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008632056608966675,
      "loss": 4.9691,
      "step": 6474
    },
    {
      "epoch": 0.886743357984114,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008631894353855434,
      "loss": 5.0311,
      "step": 6475
    },
    {
      "epoch": 0.8868803067652697,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008631732064502197,
      "loss": 4.9535,
      "step": 6476
    },
    {
      "epoch": 0.8870172555464256,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008631569740908314,
      "loss": 5.046,
      "step": 6477
    },
    {
      "epoch": 0.8871542043275815,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008631407383075127,
      "loss": 4.9539,
      "step": 6478
    },
    {
      "epoch": 0.8872911531087373,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008631244991003982,
      "loss": 5.0272,
      "step": 6479
    },
    {
      "epoch": 0.8874281018898932,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008631082564696227,
      "loss": 5.0364,
      "step": 6480
    },
    {
      "epoch": 0.8875650506710491,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008630920104153205,
      "loss": 5.0102,
      "step": 6481
    },
    {
      "epoch": 0.8877019994522048,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008630757609376264,
      "loss": 5.0436,
      "step": 6482
    },
    {
      "epoch": 0.8878389482333607,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008630595080366751,
      "loss": 4.9127,
      "step": 6483
    },
    {
      "epoch": 0.8879758970145166,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008630432517126013,
      "loss": 4.9614,
      "step": 6484
    },
    {
      "epoch": 0.8881128457956724,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008630269919655395,
      "loss": 4.9262,
      "step": 6485
    },
    {
      "epoch": 0.8882497945768283,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008630107287956245,
      "loss": 4.8846,
      "step": 6486
    },
    {
      "epoch": 0.8883867433579841,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008629944622029914,
      "loss": 4.9529,
      "step": 6487
    },
    {
      "epoch": 0.88852369213914,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008629781921877746,
      "loss": 4.9868,
      "step": 6488
    },
    {
      "epoch": 0.8886606409202958,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008629619187501093,
      "loss": 4.9971,
      "step": 6489
    },
    {
      "epoch": 0.8887975897014516,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008629456418901301,
      "loss": 5.0057,
      "step": 6490
    },
    {
      "epoch": 0.8889345384826075,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008629293616079719,
      "loss": 4.9429,
      "step": 6491
    },
    {
      "epoch": 0.8890714872637634,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008629130779037698,
      "loss": 4.9828,
      "step": 6492
    },
    {
      "epoch": 0.8892084360449192,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008628967907776585,
      "loss": 5.0376,
      "step": 6493
    },
    {
      "epoch": 0.8893453848260751,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008628805002297732,
      "loss": 4.9398,
      "step": 6494
    },
    {
      "epoch": 0.889482333607231,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008628642062602488,
      "loss": 4.9744,
      "step": 6495
    },
    {
      "epoch": 0.8896192823883867,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008628479088692203,
      "loss": 4.958,
      "step": 6496
    },
    {
      "epoch": 0.8897562311695426,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008628316080568228,
      "loss": 4.9736,
      "step": 6497
    },
    {
      "epoch": 0.8898931799506984,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008628153038231914,
      "loss": 4.9176,
      "step": 6498
    },
    {
      "epoch": 0.8900301287318543,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008627989961684613,
      "loss": 5.0568,
      "step": 6499
    },
    {
      "epoch": 0.8901670775130102,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008627826850927674,
      "loss": 5.0439,
      "step": 6500
    },
    {
      "epoch": 0.8903040262941659,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008627663705962452,
      "loss": 4.9843,
      "step": 6501
    },
    {
      "epoch": 0.8904409750753218,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008627500526790297,
      "loss": 5.0419,
      "step": 6502
    },
    {
      "epoch": 0.8905779238564777,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000862733731341256,
      "loss": 5.0171,
      "step": 6503
    },
    {
      "epoch": 0.8907148726376335,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008627174065830597,
      "loss": 4.998,
      "step": 6504
    },
    {
      "epoch": 0.8908518214187894,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008627010784045758,
      "loss": 5.083,
      "step": 6505
    },
    {
      "epoch": 0.8909887701999453,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008626847468059397,
      "loss": 4.943,
      "step": 6506
    },
    {
      "epoch": 0.891125718981101,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008626684117872869,
      "loss": 5.0142,
      "step": 6507
    },
    {
      "epoch": 0.8912626677622569,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008626520733487526,
      "loss": 5.051,
      "step": 6508
    },
    {
      "epoch": 0.8913996165434127,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008626357314904721,
      "loss": 5.0027,
      "step": 6509
    },
    {
      "epoch": 0.8915365653245686,
      "grad_norm": 0.46875,
      "learning_rate": 0.000862619386212581,
      "loss": 5.0642,
      "step": 6510
    },
    {
      "epoch": 0.8916735141057245,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008626030375152149,
      "loss": 4.9737,
      "step": 6511
    },
    {
      "epoch": 0.8918104628868803,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008625866853985089,
      "loss": 4.9268,
      "step": 6512
    },
    {
      "epoch": 0.8919474116680361,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008625703298625988,
      "loss": 4.8949,
      "step": 6513
    },
    {
      "epoch": 0.892084360449192,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008625539709076202,
      "loss": 4.9692,
      "step": 6514
    },
    {
      "epoch": 0.8922213092303478,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008625376085337084,
      "loss": 5.0437,
      "step": 6515
    },
    {
      "epoch": 0.8923582580115037,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008625212427409992,
      "loss": 5.0185,
      "step": 6516
    },
    {
      "epoch": 0.8924952067926596,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000862504873529628,
      "loss": 4.9972,
      "step": 6517
    },
    {
      "epoch": 0.8926321555738154,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008624885008997308,
      "loss": 4.9691,
      "step": 6518
    },
    {
      "epoch": 0.8927691043549713,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000862472124851443,
      "loss": 4.9586,
      "step": 6519
    },
    {
      "epoch": 0.8929060531361271,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008624557453849004,
      "loss": 4.9298,
      "step": 6520
    },
    {
      "epoch": 0.8930430019172829,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008624393625002389,
      "loss": 4.933,
      "step": 6521
    },
    {
      "epoch": 0.8931799506984388,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008624229761975941,
      "loss": 5.0322,
      "step": 6522
    },
    {
      "epoch": 0.8933168994795946,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008624065864771017,
      "loss": 5.0027,
      "step": 6523
    },
    {
      "epoch": 0.8934538482607505,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008623901933388977,
      "loss": 4.9964,
      "step": 6524
    },
    {
      "epoch": 0.8935907970419064,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008623737967831179,
      "loss": 4.9154,
      "step": 6525
    },
    {
      "epoch": 0.8937277458230621,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008623573968098981,
      "loss": 4.9943,
      "step": 6526
    },
    {
      "epoch": 0.893864694604218,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008623409934193744,
      "loss": 4.9438,
      "step": 6527
    },
    {
      "epoch": 0.8940016433853739,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008623245866116827,
      "loss": 5.0485,
      "step": 6528
    },
    {
      "epoch": 0.8941385921665297,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008623081763869587,
      "loss": 4.9726,
      "step": 6529
    },
    {
      "epoch": 0.8942755409476856,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008622917627453388,
      "loss": 4.9861,
      "step": 6530
    },
    {
      "epoch": 0.8944124897288415,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008622753456869587,
      "loss": 4.9715,
      "step": 6531
    },
    {
      "epoch": 0.8945494385099972,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008622589252119545,
      "loss": 4.9805,
      "step": 6532
    },
    {
      "epoch": 0.8946863872911531,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008622425013204625,
      "loss": 5.0127,
      "step": 6533
    },
    {
      "epoch": 0.8948233360723089,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008622260740126186,
      "loss": 4.999,
      "step": 6534
    },
    {
      "epoch": 0.8949602848534648,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000862209643288559,
      "loss": 4.9596,
      "step": 6535
    },
    {
      "epoch": 0.8950972336346207,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008621932091484199,
      "loss": 4.9793,
      "step": 6536
    },
    {
      "epoch": 0.8952341824157765,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008621767715923374,
      "loss": 4.9828,
      "step": 6537
    },
    {
      "epoch": 0.8953711311969323,
      "grad_norm": 0.5,
      "learning_rate": 0.0008621603306204478,
      "loss": 4.9221,
      "step": 6538
    },
    {
      "epoch": 0.8955080799780882,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008621438862328874,
      "loss": 4.9666,
      "step": 6539
    },
    {
      "epoch": 0.895645028759244,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008621274384297923,
      "loss": 5.0086,
      "step": 6540
    },
    {
      "epoch": 0.8957819775403999,
      "grad_norm": 0.40625,
      "learning_rate": 0.000862110987211299,
      "loss": 5.0422,
      "step": 6541
    },
    {
      "epoch": 0.8959189263215558,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008620945325775436,
      "loss": 4.9731,
      "step": 6542
    },
    {
      "epoch": 0.8960558751027116,
      "grad_norm": 0.375,
      "learning_rate": 0.0008620780745286628,
      "loss": 4.9817,
      "step": 6543
    },
    {
      "epoch": 0.8961928238838675,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008620616130647926,
      "loss": 4.9827,
      "step": 6544
    },
    {
      "epoch": 0.8963297726650232,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008620451481860697,
      "loss": 4.9544,
      "step": 6545
    },
    {
      "epoch": 0.8964667214461791,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008620286798926306,
      "loss": 4.961,
      "step": 6546
    },
    {
      "epoch": 0.896603670227335,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008620122081846115,
      "loss": 5.0047,
      "step": 6547
    },
    {
      "epoch": 0.8967406190084908,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008619957330621492,
      "loss": 4.9732,
      "step": 6548
    },
    {
      "epoch": 0.8968775677896467,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008619792545253799,
      "loss": 4.9844,
      "step": 6549
    },
    {
      "epoch": 0.8970145165708026,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008619627725744404,
      "loss": 4.9172,
      "step": 6550
    },
    {
      "epoch": 0.8971514653519583,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008619462872094673,
      "loss": 4.9818,
      "step": 6551
    },
    {
      "epoch": 0.8972884141331142,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008619297984305972,
      "loss": 4.9935,
      "step": 6552
    },
    {
      "epoch": 0.8974253629142701,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008619133062379667,
      "loss": 4.9972,
      "step": 6553
    },
    {
      "epoch": 0.8975623116954259,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008618968106317123,
      "loss": 5.0915,
      "step": 6554
    },
    {
      "epoch": 0.8976992604765818,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008618803116119711,
      "loss": 4.9979,
      "step": 6555
    },
    {
      "epoch": 0.8978362092577377,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008618638091788794,
      "loss": 4.8619,
      "step": 6556
    },
    {
      "epoch": 0.8979731580388934,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008618473033325743,
      "loss": 4.9604,
      "step": 6557
    },
    {
      "epoch": 0.8981101068200493,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008618307940731926,
      "loss": 4.9653,
      "step": 6558
    },
    {
      "epoch": 0.8982470556012051,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008618142814008706,
      "loss": 4.9911,
      "step": 6559
    },
    {
      "epoch": 0.898384004382361,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008617977653157458,
      "loss": 4.9973,
      "step": 6560
    },
    {
      "epoch": 0.8985209531635169,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008617812458179548,
      "loss": 5.0199,
      "step": 6561
    },
    {
      "epoch": 0.8986579019446727,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008617647229076344,
      "loss": 5.0256,
      "step": 6562
    },
    {
      "epoch": 0.8987948507258285,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008617481965849216,
      "loss": 4.9751,
      "step": 6563
    },
    {
      "epoch": 0.8989317995069844,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008617316668499534,
      "loss": 4.9664,
      "step": 6564
    },
    {
      "epoch": 0.8990687482881402,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008617151337028668,
      "loss": 4.919,
      "step": 6565
    },
    {
      "epoch": 0.8992056970692961,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008616985971437989,
      "loss": 4.906,
      "step": 6566
    },
    {
      "epoch": 0.899342645850452,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008616820571728863,
      "loss": 4.9756,
      "step": 6567
    },
    {
      "epoch": 0.8994795946316078,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008616655137902667,
      "loss": 4.966,
      "step": 6568
    },
    {
      "epoch": 0.8996165434127636,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008616489669960768,
      "loss": 4.9397,
      "step": 6569
    },
    {
      "epoch": 0.8997534921939194,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008616324167904538,
      "loss": 5.0219,
      "step": 6570
    },
    {
      "epoch": 0.8998904409750753,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008616158631735347,
      "loss": 5.0976,
      "step": 6571
    },
    {
      "epoch": 0.9000273897562312,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008615993061454572,
      "loss": 4.9864,
      "step": 6572
    },
    {
      "epoch": 0.900164338537387,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008615827457063579,
      "loss": 4.986,
      "step": 6573
    },
    {
      "epoch": 0.9003012873185429,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008615661818563745,
      "loss": 4.9718,
      "step": 6574
    },
    {
      "epoch": 0.9004382360996988,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008615496145956441,
      "loss": 4.9961,
      "step": 6575
    },
    {
      "epoch": 0.9005751848808545,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008615330439243038,
      "loss": 4.9742,
      "step": 6576
    },
    {
      "epoch": 0.9007121336620104,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008615164698424912,
      "loss": 4.9223,
      "step": 6577
    },
    {
      "epoch": 0.9008490824431663,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008614998923503434,
      "loss": 5.0387,
      "step": 6578
    },
    {
      "epoch": 0.9009860312243221,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008614833114479982,
      "loss": 5.0401,
      "step": 6579
    },
    {
      "epoch": 0.901122980005478,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008614667271355926,
      "loss": 5.0287,
      "step": 6580
    },
    {
      "epoch": 0.9012599287866337,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008614501394132642,
      "loss": 5.0153,
      "step": 6581
    },
    {
      "epoch": 0.9013968775677896,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008614335482811504,
      "loss": 5.0014,
      "step": 6582
    },
    {
      "epoch": 0.9015338263489455,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008614169537393888,
      "loss": 4.9586,
      "step": 6583
    },
    {
      "epoch": 0.9016707751301013,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008614003557881169,
      "loss": 4.988,
      "step": 6584
    },
    {
      "epoch": 0.9018077239112572,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008613837544274722,
      "loss": 5.0334,
      "step": 6585
    },
    {
      "epoch": 0.9019446726924131,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008613671496575922,
      "loss": 4.9753,
      "step": 6586
    },
    {
      "epoch": 0.9020816214735689,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008613505414786148,
      "loss": 4.8875,
      "step": 6587
    },
    {
      "epoch": 0.9022185702547247,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008613339298906773,
      "loss": 4.8973,
      "step": 6588
    },
    {
      "epoch": 0.9023555190358806,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008613173148939175,
      "loss": 4.8868,
      "step": 6589
    },
    {
      "epoch": 0.9024924678170364,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008613006964884732,
      "loss": 5.0318,
      "step": 6590
    },
    {
      "epoch": 0.9026294165981923,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008612840746744819,
      "loss": 5.0634,
      "step": 6591
    },
    {
      "epoch": 0.9027663653793482,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008612674494520815,
      "loss": 4.9236,
      "step": 6592
    },
    {
      "epoch": 0.902903314160504,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008612508208214099,
      "loss": 5.0118,
      "step": 6593
    },
    {
      "epoch": 0.9030402629416598,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008612341887826046,
      "loss": 5.0025,
      "step": 6594
    },
    {
      "epoch": 0.9031772117228156,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008612175533358037,
      "loss": 5.0059,
      "step": 6595
    },
    {
      "epoch": 0.9033141605039715,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008612009144811448,
      "loss": 5.018,
      "step": 6596
    },
    {
      "epoch": 0.9034511092851274,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008611842722187661,
      "loss": 4.9512,
      "step": 6597
    },
    {
      "epoch": 0.9035880580662832,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008611676265488052,
      "loss": 4.9781,
      "step": 6598
    },
    {
      "epoch": 0.9037250068474391,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008611509774714002,
      "loss": 4.9856,
      "step": 6599
    },
    {
      "epoch": 0.903861955628595,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008611343249866891,
      "loss": 5.0091,
      "step": 6600
    },
    {
      "epoch": 0.9039989044097507,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00086111766909481,
      "loss": 4.9634,
      "step": 6601
    },
    {
      "epoch": 0.9041358531909066,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008611010097959007,
      "loss": 5.0138,
      "step": 6602
    },
    {
      "epoch": 0.9042728019720625,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008610843470900995,
      "loss": 4.946,
      "step": 6603
    },
    {
      "epoch": 0.9044097507532183,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008610676809775443,
      "loss": 5.0201,
      "step": 6604
    },
    {
      "epoch": 0.9045466995343742,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008610510114583731,
      "loss": 4.9849,
      "step": 6605
    },
    {
      "epoch": 0.9046836483155299,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008610343385327244,
      "loss": 4.9761,
      "step": 6606
    },
    {
      "epoch": 0.9048205970966858,
      "grad_norm": 0.375,
      "learning_rate": 0.0008610176622007362,
      "loss": 5.0108,
      "step": 6607
    },
    {
      "epoch": 0.9049575458778417,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008610009824625465,
      "loss": 4.998,
      "step": 6608
    },
    {
      "epoch": 0.9050944946589975,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000860984299318294,
      "loss": 4.965,
      "step": 6609
    },
    {
      "epoch": 0.9052314434401534,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008609676127681165,
      "loss": 4.9511,
      "step": 6610
    },
    {
      "epoch": 0.9053683922213093,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008609509228121525,
      "loss": 5.0436,
      "step": 6611
    },
    {
      "epoch": 0.905505341002465,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008609342294505404,
      "loss": 5.0435,
      "step": 6612
    },
    {
      "epoch": 0.9056422897836209,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008609175326834183,
      "loss": 5.0008,
      "step": 6613
    },
    {
      "epoch": 0.9057792385647768,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008609008325109246,
      "loss": 4.9797,
      "step": 6614
    },
    {
      "epoch": 0.9059161873459326,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000860884128933198,
      "loss": 4.9401,
      "step": 6615
    },
    {
      "epoch": 0.9060531361270885,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008608674219503766,
      "loss": 4.989,
      "step": 6616
    },
    {
      "epoch": 0.9061900849082443,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008608507115625989,
      "loss": 5.0198,
      "step": 6617
    },
    {
      "epoch": 0.9063270336894002,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008608339977700036,
      "loss": 4.9447,
      "step": 6618
    },
    {
      "epoch": 0.906463982470556,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008608172805727289,
      "loss": 4.986,
      "step": 6619
    },
    {
      "epoch": 0.9066009312517118,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008608005599709137,
      "loss": 4.9471,
      "step": 6620
    },
    {
      "epoch": 0.9067378800328677,
      "grad_norm": 0.375,
      "learning_rate": 0.0008607838359646963,
      "loss": 4.9121,
      "step": 6621
    },
    {
      "epoch": 0.9068748288140236,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008607671085542153,
      "loss": 5.0296,
      "step": 6622
    },
    {
      "epoch": 0.9070117775951794,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008607503777396094,
      "loss": 5.0547,
      "step": 6623
    },
    {
      "epoch": 0.9071487263763353,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008607336435210173,
      "loss": 4.9649,
      "step": 6624
    },
    {
      "epoch": 0.9072856751574911,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008607169058985776,
      "loss": 4.9438,
      "step": 6625
    },
    {
      "epoch": 0.9074226239386469,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000860700164872429,
      "loss": 5.0597,
      "step": 6626
    },
    {
      "epoch": 0.9075595727198028,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008606834204427103,
      "loss": 4.9334,
      "step": 6627
    },
    {
      "epoch": 0.9076965215009586,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008606666726095602,
      "loss": 4.9563,
      "step": 6628
    },
    {
      "epoch": 0.9078334702821145,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008606499213731176,
      "loss": 4.958,
      "step": 6629
    },
    {
      "epoch": 0.9079704190632704,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008606331667335213,
      "loss": 4.9414,
      "step": 6630
    },
    {
      "epoch": 0.9081073678444261,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008606164086909099,
      "loss": 4.9571,
      "step": 6631
    },
    {
      "epoch": 0.908244316625582,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008605996472454226,
      "loss": 4.9766,
      "step": 6632
    },
    {
      "epoch": 0.9083812654067379,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008605828823971981,
      "loss": 4.9127,
      "step": 6633
    },
    {
      "epoch": 0.9085182141878937,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008605661141463756,
      "loss": 4.9469,
      "step": 6634
    },
    {
      "epoch": 0.9086551629690496,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008605493424930937,
      "loss": 5.0064,
      "step": 6635
    },
    {
      "epoch": 0.9087921117502055,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008605325674374916,
      "loss": 5.043,
      "step": 6636
    },
    {
      "epoch": 0.9089290605313612,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008605157889797083,
      "loss": 5.0113,
      "step": 6637
    },
    {
      "epoch": 0.9090660093125171,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008604990071198828,
      "loss": 4.9544,
      "step": 6638
    },
    {
      "epoch": 0.909202958093673,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008604822218581543,
      "loss": 5.0461,
      "step": 6639
    },
    {
      "epoch": 0.9093399068748288,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008604654331946619,
      "loss": 4.9853,
      "step": 6640
    },
    {
      "epoch": 0.9094768556559847,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008604486411295445,
      "loss": 4.8798,
      "step": 6641
    },
    {
      "epoch": 0.9096138044371405,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008604318456629414,
      "loss": 5.019,
      "step": 6642
    },
    {
      "epoch": 0.9097507532182963,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008604150467949918,
      "loss": 4.9157,
      "step": 6643
    },
    {
      "epoch": 0.9098877019994522,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008603982445258349,
      "loss": 4.9421,
      "step": 6644
    },
    {
      "epoch": 0.910024650780608,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008603814388556101,
      "loss": 4.984,
      "step": 6645
    },
    {
      "epoch": 0.9101615995617639,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008603646297844564,
      "loss": 4.958,
      "step": 6646
    },
    {
      "epoch": 0.9102985483429198,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008603478173125133,
      "loss": 4.9282,
      "step": 6647
    },
    {
      "epoch": 0.9104354971240756,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00086033100143992,
      "loss": 5.0038,
      "step": 6648
    },
    {
      "epoch": 0.9105724459052315,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008603141821668159,
      "loss": 4.97,
      "step": 6649
    },
    {
      "epoch": 0.9107093946863873,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008602973594933404,
      "loss": 5.0092,
      "step": 6650
    },
    {
      "epoch": 0.9108463434675431,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008602805334196331,
      "loss": 4.9361,
      "step": 6651
    },
    {
      "epoch": 0.910983292248699,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008602637039458328,
      "loss": 4.9698,
      "step": 6652
    },
    {
      "epoch": 0.9111202410298548,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008602468710720798,
      "loss": 5.0095,
      "step": 6653
    },
    {
      "epoch": 0.9112571898110107,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008602300347985132,
      "loss": 4.8951,
      "step": 6654
    },
    {
      "epoch": 0.9113941385921666,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008602131951252725,
      "loss": 4.9948,
      "step": 6655
    },
    {
      "epoch": 0.9115310873733223,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008601963520524972,
      "loss": 5.0077,
      "step": 6656
    },
    {
      "epoch": 0.9116680361544782,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008601795055803271,
      "loss": 4.9715,
      "step": 6657
    },
    {
      "epoch": 0.9118049849356341,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008601626557089017,
      "loss": 5.0087,
      "step": 6658
    },
    {
      "epoch": 0.9119419337167899,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008601458024383606,
      "loss": 4.9361,
      "step": 6659
    },
    {
      "epoch": 0.9120788824979458,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008601289457688436,
      "loss": 4.9514,
      "step": 6660
    },
    {
      "epoch": 0.9122158312791017,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008601120857004901,
      "loss": 4.9743,
      "step": 6661
    },
    {
      "epoch": 0.9123527800602574,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008600952222334402,
      "loss": 4.9384,
      "step": 6662
    },
    {
      "epoch": 0.9124897288414133,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008600783553678334,
      "loss": 5.0347,
      "step": 6663
    },
    {
      "epoch": 0.9126266776225691,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008600614851038095,
      "loss": 4.9028,
      "step": 6664
    },
    {
      "epoch": 0.912763626403725,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008600446114415084,
      "loss": 4.9412,
      "step": 6665
    },
    {
      "epoch": 0.9129005751848809,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00086002773438107,
      "loss": 4.9575,
      "step": 6666
    },
    {
      "epoch": 0.9130375239660367,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000860010853922634,
      "loss": 5.0126,
      "step": 6667
    },
    {
      "epoch": 0.9131744727471925,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008599939700663403,
      "loss": 4.9739,
      "step": 6668
    },
    {
      "epoch": 0.9133114215283484,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000859977082812329,
      "loss": 4.9832,
      "step": 6669
    },
    {
      "epoch": 0.9134483703095042,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008599601921607398,
      "loss": 4.9369,
      "step": 6670
    },
    {
      "epoch": 0.9135853190906601,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008599432981117128,
      "loss": 4.9835,
      "step": 6671
    },
    {
      "epoch": 0.913722267871816,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000859926400665388,
      "loss": 4.9694,
      "step": 6672
    },
    {
      "epoch": 0.9138592166529718,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008599094998219054,
      "loss": 5.0506,
      "step": 6673
    },
    {
      "epoch": 0.9139961654341276,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008598925955814052,
      "loss": 4.9446,
      "step": 6674
    },
    {
      "epoch": 0.9141331142152835,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008598756879440274,
      "loss": 4.9879,
      "step": 6675
    },
    {
      "epoch": 0.9142700629964393,
      "grad_norm": 0.486328125,
      "learning_rate": 0.000859858776909912,
      "loss": 4.9329,
      "step": 6676
    },
    {
      "epoch": 0.9144070117775952,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008598418624791994,
      "loss": 5.0082,
      "step": 6677
    },
    {
      "epoch": 0.914543960558751,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008598249446520296,
      "loss": 4.9589,
      "step": 6678
    },
    {
      "epoch": 0.9146809093399069,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008598080234285426,
      "loss": 4.987,
      "step": 6679
    },
    {
      "epoch": 0.9148178581210628,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000859791098808879,
      "loss": 5.0544,
      "step": 6680
    },
    {
      "epoch": 0.9149548069022185,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008597741707931791,
      "loss": 5.0022,
      "step": 6681
    },
    {
      "epoch": 0.9150917556833744,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008597572393815826,
      "loss": 4.9405,
      "step": 6682
    },
    {
      "epoch": 0.9152287044645303,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008597403045742306,
      "loss": 4.8766,
      "step": 6683
    },
    {
      "epoch": 0.9153656532456861,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008597233663712629,
      "loss": 4.958,
      "step": 6684
    },
    {
      "epoch": 0.915502602026842,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008597064247728201,
      "loss": 4.9785,
      "step": 6685
    },
    {
      "epoch": 0.9156395508079979,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008596894797790425,
      "loss": 4.9791,
      "step": 6686
    },
    {
      "epoch": 0.9157764995891536,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008596725313900705,
      "loss": 4.9551,
      "step": 6687
    },
    {
      "epoch": 0.9159134483703095,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008596555796060447,
      "loss": 5.0069,
      "step": 6688
    },
    {
      "epoch": 0.9160503971514653,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008596386244271053,
      "loss": 5.0003,
      "step": 6689
    },
    {
      "epoch": 0.9161873459326212,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008596216658533932,
      "loss": 5.0659,
      "step": 6690
    },
    {
      "epoch": 0.9163242947137771,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008596047038850488,
      "loss": 4.9918,
      "step": 6691
    },
    {
      "epoch": 0.9164612434949329,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008595877385222124,
      "loss": 4.9539,
      "step": 6692
    },
    {
      "epoch": 0.9165981922760887,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000859570769765025,
      "loss": 5.0283,
      "step": 6693
    },
    {
      "epoch": 0.9167351410572446,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008595537976136269,
      "loss": 4.9908,
      "step": 6694
    },
    {
      "epoch": 0.9168720898384004,
      "grad_norm": 0.53125,
      "learning_rate": 0.000859536822068159,
      "loss": 4.9663,
      "step": 6695
    },
    {
      "epoch": 0.9170090386195563,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008595198431287619,
      "loss": 4.9982,
      "step": 6696
    },
    {
      "epoch": 0.9171459874007122,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008595028607955761,
      "loss": 5.0195,
      "step": 6697
    },
    {
      "epoch": 0.917282936181868,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008594858750687426,
      "loss": 4.9949,
      "step": 6698
    },
    {
      "epoch": 0.9174198849630238,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008594688859484022,
      "loss": 4.9201,
      "step": 6699
    },
    {
      "epoch": 0.9175568337441796,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008594518934346955,
      "loss": 4.9461,
      "step": 6700
    },
    {
      "epoch": 0.9176937825253355,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008594348975277633,
      "loss": 4.8633,
      "step": 6701
    },
    {
      "epoch": 0.9178307313064914,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008594178982277465,
      "loss": 4.952,
      "step": 6702
    },
    {
      "epoch": 0.9179676800876472,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008594008955347862,
      "loss": 4.9383,
      "step": 6703
    },
    {
      "epoch": 0.9181046288688031,
      "grad_norm": 0.48828125,
      "learning_rate": 0.000859383889449023,
      "loss": 4.9824,
      "step": 6704
    },
    {
      "epoch": 0.918241577649959,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000859366879970598,
      "loss": 4.954,
      "step": 6705
    },
    {
      "epoch": 0.9183785264311147,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008593498670996521,
      "loss": 4.8957,
      "step": 6706
    },
    {
      "epoch": 0.9185154752122706,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008593328508363263,
      "loss": 5.0543,
      "step": 6707
    },
    {
      "epoch": 0.9186524239934265,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008593158311807616,
      "loss": 4.9635,
      "step": 6708
    },
    {
      "epoch": 0.9187893727745823,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008592988081330992,
      "loss": 4.9363,
      "step": 6709
    },
    {
      "epoch": 0.9189263215557382,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00085928178169348,
      "loss": 4.9649,
      "step": 6710
    },
    {
      "epoch": 0.9190632703368941,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008592647518620452,
      "loss": 4.9431,
      "step": 6711
    },
    {
      "epoch": 0.9192002191180498,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008592477186389357,
      "loss": 5.0122,
      "step": 6712
    },
    {
      "epoch": 0.9193371678992057,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008592306820242929,
      "loss": 5.0557,
      "step": 6713
    },
    {
      "epoch": 0.9194741166803615,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000859213642018258,
      "loss": 4.9921,
      "step": 6714
    },
    {
      "epoch": 0.9196110654615174,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000859196598620972,
      "loss": 5.0412,
      "step": 6715
    },
    {
      "epoch": 0.9197480142426733,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008591795518325764,
      "loss": 5.02,
      "step": 6716
    },
    {
      "epoch": 0.919884963023829,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008591625016532124,
      "loss": 4.9415,
      "step": 6717
    },
    {
      "epoch": 0.9200219118049849,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008591454480830212,
      "loss": 4.8928,
      "step": 6718
    },
    {
      "epoch": 0.9201588605861408,
      "grad_norm": 0.46875,
      "learning_rate": 0.000859128391122144,
      "loss": 4.9633,
      "step": 6719
    },
    {
      "epoch": 0.9202958093672966,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008591113307707226,
      "loss": 5.0402,
      "step": 6720
    },
    {
      "epoch": 0.9204327581484525,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008590942670288979,
      "loss": 5.0446,
      "step": 6721
    },
    {
      "epoch": 0.9205697069296084,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008590771998968118,
      "loss": 5.0225,
      "step": 6722
    },
    {
      "epoch": 0.9207066557107642,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008590601293746051,
      "loss": 4.9013,
      "step": 6723
    },
    {
      "epoch": 0.92084360449192,
      "grad_norm": 0.375,
      "learning_rate": 0.0008590430554624199,
      "loss": 5.0197,
      "step": 6724
    },
    {
      "epoch": 0.9209805532730758,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008590259781603974,
      "loss": 5.0289,
      "step": 6725
    },
    {
      "epoch": 0.9211175020542317,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008590088974686792,
      "loss": 4.9582,
      "step": 6726
    },
    {
      "epoch": 0.9212544508353876,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008589918133874066,
      "loss": 4.925,
      "step": 6727
    },
    {
      "epoch": 0.9213913996165434,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008589747259167216,
      "loss": 4.9218,
      "step": 6728
    },
    {
      "epoch": 0.9215283483976993,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008589576350567655,
      "loss": 4.9824,
      "step": 6729
    },
    {
      "epoch": 0.9216652971788551,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00085894054080768,
      "loss": 4.9341,
      "step": 6730
    },
    {
      "epoch": 0.9218022459600109,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008589234431696067,
      "loss": 4.9128,
      "step": 6731
    },
    {
      "epoch": 0.9219391947411668,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008589063421426875,
      "loss": 4.9467,
      "step": 6732
    },
    {
      "epoch": 0.9220761435223227,
      "grad_norm": 0.375,
      "learning_rate": 0.0008588892377270641,
      "loss": 4.951,
      "step": 6733
    },
    {
      "epoch": 0.9222130923034785,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008588721299228781,
      "loss": 5.001,
      "step": 6734
    },
    {
      "epoch": 0.9223500410846344,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008588550187302713,
      "loss": 5.057,
      "step": 6735
    },
    {
      "epoch": 0.9224869898657901,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008588379041493855,
      "loss": 4.8908,
      "step": 6736
    },
    {
      "epoch": 0.922623938646946,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008588207861803626,
      "loss": 4.9832,
      "step": 6737
    },
    {
      "epoch": 0.9227608874281019,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008588036648233445,
      "loss": 4.9991,
      "step": 6738
    },
    {
      "epoch": 0.9228978362092577,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008587865400784729,
      "loss": 4.976,
      "step": 6739
    },
    {
      "epoch": 0.9230347849904136,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008587694119458898,
      "loss": 4.9455,
      "step": 6740
    },
    {
      "epoch": 0.9231717337715695,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008587522804257374,
      "loss": 4.9927,
      "step": 6741
    },
    {
      "epoch": 0.9233086825527252,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008587351455181571,
      "loss": 5.0227,
      "step": 6742
    },
    {
      "epoch": 0.9234456313338811,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008587180072232916,
      "loss": 4.994,
      "step": 6743
    },
    {
      "epoch": 0.923582580115037,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008587008655412824,
      "loss": 5.0391,
      "step": 6744
    },
    {
      "epoch": 0.9237195288961928,
      "grad_norm": 0.375,
      "learning_rate": 0.0008586837204722717,
      "loss": 4.994,
      "step": 6745
    },
    {
      "epoch": 0.9238564776773487,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008586665720164018,
      "loss": 4.9226,
      "step": 6746
    },
    {
      "epoch": 0.9239934264585046,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008586494201738145,
      "loss": 4.8553,
      "step": 6747
    },
    {
      "epoch": 0.9241303752396604,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008586322649446522,
      "loss": 4.9264,
      "step": 6748
    },
    {
      "epoch": 0.9242673240208162,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008586151063290569,
      "loss": 4.9816,
      "step": 6749
    },
    {
      "epoch": 0.924404272801972,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008585979443271708,
      "loss": 4.9695,
      "step": 6750
    },
    {
      "epoch": 0.9245412215831279,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008585807789391362,
      "loss": 4.9461,
      "step": 6751
    },
    {
      "epoch": 0.9246781703642838,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008585636101650952,
      "loss": 4.8824,
      "step": 6752
    },
    {
      "epoch": 0.9248151191454396,
      "grad_norm": 0.375,
      "learning_rate": 0.0008585464380051903,
      "loss": 4.9275,
      "step": 6753
    },
    {
      "epoch": 0.9249520679265955,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008585292624595638,
      "loss": 4.9566,
      "step": 6754
    },
    {
      "epoch": 0.9250890167077513,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008585120835283579,
      "loss": 4.863,
      "step": 6755
    },
    {
      "epoch": 0.9252259654889071,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000858494901211715,
      "loss": 4.9708,
      "step": 6756
    },
    {
      "epoch": 0.925362914270063,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008584777155097776,
      "loss": 4.9762,
      "step": 6757
    },
    {
      "epoch": 0.9254998630512189,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008584605264226879,
      "loss": 5.0086,
      "step": 6758
    },
    {
      "epoch": 0.9256368118323747,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0008584433339505885,
      "loss": 4.9973,
      "step": 6759
    },
    {
      "epoch": 0.9257737606135306,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000858426138093622,
      "loss": 4.9813,
      "step": 6760
    },
    {
      "epoch": 0.9259107093946863,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008584089388519307,
      "loss": 4.9158,
      "step": 6761
    },
    {
      "epoch": 0.9260476581758422,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008583917362256572,
      "loss": 4.9761,
      "step": 6762
    },
    {
      "epoch": 0.9261846069569981,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000858374530214944,
      "loss": 5.0494,
      "step": 6763
    },
    {
      "epoch": 0.9263215557381539,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008583573208199338,
      "loss": 4.982,
      "step": 6764
    },
    {
      "epoch": 0.9264585045193098,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008583401080407691,
      "loss": 4.9386,
      "step": 6765
    },
    {
      "epoch": 0.9265954533004657,
      "grad_norm": 0.375,
      "learning_rate": 0.0008583228918775928,
      "loss": 5.0051,
      "step": 6766
    },
    {
      "epoch": 0.9267324020816214,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008583056723305473,
      "loss": 4.9881,
      "step": 6767
    },
    {
      "epoch": 0.9268693508627773,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008582884493997754,
      "loss": 4.9892,
      "step": 6768
    },
    {
      "epoch": 0.9270062996439332,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008582712230854199,
      "loss": 4.9831,
      "step": 6769
    },
    {
      "epoch": 0.927143248425089,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008582539933876235,
      "loss": 5.0078,
      "step": 6770
    },
    {
      "epoch": 0.9272801972062449,
      "grad_norm": 0.375,
      "learning_rate": 0.0008582367603065288,
      "loss": 5.0545,
      "step": 6771
    },
    {
      "epoch": 0.9274171459874007,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000858219523842279,
      "loss": 4.8863,
      "step": 6772
    },
    {
      "epoch": 0.9275540947685565,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008582022839950166,
      "loss": 4.8679,
      "step": 6773
    },
    {
      "epoch": 0.9276910435497124,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008581850407648845,
      "loss": 5.0206,
      "step": 6774
    },
    {
      "epoch": 0.9278279923308682,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008581677941520259,
      "loss": 4.9253,
      "step": 6775
    },
    {
      "epoch": 0.9279649411120241,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008581505441565835,
      "loss": 5.0719,
      "step": 6776
    },
    {
      "epoch": 0.92810188989318,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008581332907787002,
      "loss": 4.9991,
      "step": 6777
    },
    {
      "epoch": 0.9282388386743358,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008581160340185191,
      "loss": 4.9954,
      "step": 6778
    },
    {
      "epoch": 0.9283757874554917,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008580987738761831,
      "loss": 4.9182,
      "step": 6779
    },
    {
      "epoch": 0.9285127362366475,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008580815103518356,
      "loss": 4.8908,
      "step": 6780
    },
    {
      "epoch": 0.9286496850178033,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008580642434456191,
      "loss": 4.989,
      "step": 6781
    },
    {
      "epoch": 0.9287866337989592,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008580469731576772,
      "loss": 4.9704,
      "step": 6782
    },
    {
      "epoch": 0.928923582580115,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008580296994881526,
      "loss": 4.965,
      "step": 6783
    },
    {
      "epoch": 0.9290605313612709,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008580124224371888,
      "loss": 5.0106,
      "step": 6784
    },
    {
      "epoch": 0.9291974801424268,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008579951420049286,
      "loss": 4.9604,
      "step": 6785
    },
    {
      "epoch": 0.9293344289235825,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008579778581915157,
      "loss": 4.9589,
      "step": 6786
    },
    {
      "epoch": 0.9294713777047384,
      "grad_norm": 0.359375,
      "learning_rate": 0.000857960570997093,
      "loss": 5.0122,
      "step": 6787
    },
    {
      "epoch": 0.9296083264858943,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008579432804218037,
      "loss": 4.9929,
      "step": 6788
    },
    {
      "epoch": 0.9297452752670501,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008579259864657914,
      "loss": 4.9929,
      "step": 6789
    },
    {
      "epoch": 0.929882224048206,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008579086891291991,
      "loss": 5.0109,
      "step": 6790
    },
    {
      "epoch": 0.9300191728293619,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008578913884121703,
      "loss": 4.9274,
      "step": 6791
    },
    {
      "epoch": 0.9301561216105176,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008578740843148485,
      "loss": 4.9065,
      "step": 6792
    },
    {
      "epoch": 0.9302930703916735,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008578567768373769,
      "loss": 4.9095,
      "step": 6793
    },
    {
      "epoch": 0.9304300191728294,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000857839465979899,
      "loss": 4.9713,
      "step": 6794
    },
    {
      "epoch": 0.9305669679539852,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008578221517425584,
      "loss": 4.9539,
      "step": 6795
    },
    {
      "epoch": 0.9307039167351411,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008578048341254982,
      "loss": 5.0577,
      "step": 6796
    },
    {
      "epoch": 0.9308408655162969,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008577875131288624,
      "loss": 5.0285,
      "step": 6797
    },
    {
      "epoch": 0.9309778142974527,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008577701887527941,
      "loss": 5.0011,
      "step": 6798
    },
    {
      "epoch": 0.9311147630786086,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008577528609974372,
      "loss": 4.9603,
      "step": 6799
    },
    {
      "epoch": 0.9312517118597644,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008577355298629352,
      "loss": 4.9424,
      "step": 6800
    },
    {
      "epoch": 0.9313886606409203,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008577181953494316,
      "loss": 4.9383,
      "step": 6801
    },
    {
      "epoch": 0.9315256094220762,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008577008574570702,
      "loss": 5.0192,
      "step": 6802
    },
    {
      "epoch": 0.931662558203232,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008576835161859947,
      "loss": 5.0275,
      "step": 6803
    },
    {
      "epoch": 0.9317995069843878,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008576661715363487,
      "loss": 4.9855,
      "step": 6804
    },
    {
      "epoch": 0.9319364557655437,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000857648823508276,
      "loss": 4.9395,
      "step": 6805
    },
    {
      "epoch": 0.9320734045466995,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008576314721019205,
      "loss": 4.9877,
      "step": 6806
    },
    {
      "epoch": 0.9322103533278554,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008576141173174257,
      "loss": 5.0321,
      "step": 6807
    },
    {
      "epoch": 0.9323473021090112,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008575967591549356,
      "loss": 5.0025,
      "step": 6808
    },
    {
      "epoch": 0.9324842508901671,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000857579397614594,
      "loss": 4.985,
      "step": 6809
    },
    {
      "epoch": 0.932621199671323,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008575620326965449,
      "loss": 5.0397,
      "step": 6810
    },
    {
      "epoch": 0.9327581484524787,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008575446644009321,
      "loss": 4.9873,
      "step": 6811
    },
    {
      "epoch": 0.9328950972336346,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008575272927278995,
      "loss": 5.0034,
      "step": 6812
    },
    {
      "epoch": 0.9330320460147905,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008575099176775911,
      "loss": 4.8781,
      "step": 6813
    },
    {
      "epoch": 0.9331689947959463,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000857492539250151,
      "loss": 5.0237,
      "step": 6814
    },
    {
      "epoch": 0.9333059435771022,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008574751574457231,
      "loss": 4.9682,
      "step": 6815
    },
    {
      "epoch": 0.9334428923582581,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008574577722644514,
      "loss": 5.0095,
      "step": 6816
    },
    {
      "epoch": 0.9335798411394138,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008574403837064801,
      "loss": 4.9351,
      "step": 6817
    },
    {
      "epoch": 0.9337167899205697,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008574229917719533,
      "loss": 5.0364,
      "step": 6818
    },
    {
      "epoch": 0.9338537387017255,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008574055964610151,
      "loss": 4.9666,
      "step": 6819
    },
    {
      "epoch": 0.9339906874828814,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008573881977738095,
      "loss": 4.9459,
      "step": 6820
    },
    {
      "epoch": 0.9341276362640373,
      "grad_norm": 0.359375,
      "learning_rate": 0.000857370795710481,
      "loss": 4.989,
      "step": 6821
    },
    {
      "epoch": 0.934264585045193,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008573533902711735,
      "loss": 4.9844,
      "step": 6822
    },
    {
      "epoch": 0.9344015338263489,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008573359814560314,
      "loss": 5.0019,
      "step": 6823
    },
    {
      "epoch": 0.9345384826075048,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008573185692651991,
      "loss": 4.9325,
      "step": 6824
    },
    {
      "epoch": 0.9346754313886606,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008573011536988207,
      "loss": 5.0299,
      "step": 6825
    },
    {
      "epoch": 0.9348123801698165,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008572837347570405,
      "loss": 4.9408,
      "step": 6826
    },
    {
      "epoch": 0.9349493289509724,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008572663124400031,
      "loss": 4.9187,
      "step": 6827
    },
    {
      "epoch": 0.9350862777321282,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008572488867478526,
      "loss": 4.9558,
      "step": 6828
    },
    {
      "epoch": 0.935223226513284,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008572314576807334,
      "loss": 4.9706,
      "step": 6829
    },
    {
      "epoch": 0.9353601752944399,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008572140252387904,
      "loss": 4.9478,
      "step": 6830
    },
    {
      "epoch": 0.9354971240755957,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008571965894221674,
      "loss": 5.0034,
      "step": 6831
    },
    {
      "epoch": 0.9356340728567516,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0008571791502310095,
      "loss": 5.0001,
      "step": 6832
    },
    {
      "epoch": 0.9357710216379074,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008571617076654607,
      "loss": 4.9661,
      "step": 6833
    },
    {
      "epoch": 0.9359079704190633,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000857144261725666,
      "loss": 5.005,
      "step": 6834
    },
    {
      "epoch": 0.9360449192002191,
      "grad_norm": 0.375,
      "learning_rate": 0.0008571268124117697,
      "loss": 4.9946,
      "step": 6835
    },
    {
      "epoch": 0.9361818679813749,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008571093597239167,
      "loss": 4.9838,
      "step": 6836
    },
    {
      "epoch": 0.9363188167625308,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008570919036622512,
      "loss": 4.9699,
      "step": 6837
    },
    {
      "epoch": 0.9364557655436867,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008570744442269181,
      "loss": 4.8681,
      "step": 6838
    },
    {
      "epoch": 0.9365927143248425,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008570569814180622,
      "loss": 4.9966,
      "step": 6839
    },
    {
      "epoch": 0.9367296631059984,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000857039515235828,
      "loss": 5.0106,
      "step": 6840
    },
    {
      "epoch": 0.9368666118871543,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008570220456803602,
      "loss": 4.8986,
      "step": 6841
    },
    {
      "epoch": 0.93700356066831,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008570045727518041,
      "loss": 4.8732,
      "step": 6842
    },
    {
      "epoch": 0.9371405094494659,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008569870964503039,
      "loss": 4.9445,
      "step": 6843
    },
    {
      "epoch": 0.9372774582306217,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008569696167760047,
      "loss": 4.9155,
      "step": 6844
    },
    {
      "epoch": 0.9374144070117776,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008569521337290512,
      "loss": 4.9423,
      "step": 6845
    },
    {
      "epoch": 0.9375513557929335,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008569346473095886,
      "loss": 4.9616,
      "step": 6846
    },
    {
      "epoch": 0.9376883045740892,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008569171575177613,
      "loss": 4.8397,
      "step": 6847
    },
    {
      "epoch": 0.9378252533552451,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008568996643537148,
      "loss": 4.9125,
      "step": 6848
    },
    {
      "epoch": 0.937962202136401,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008568821678175938,
      "loss": 4.9631,
      "step": 6849
    },
    {
      "epoch": 0.9380991509175568,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008568646679095433,
      "loss": 4.9623,
      "step": 6850
    },
    {
      "epoch": 0.9382360996987127,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008568471646297083,
      "loss": 4.979,
      "step": 6851
    },
    {
      "epoch": 0.9383730484798686,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008568296579782339,
      "loss": 4.9752,
      "step": 6852
    },
    {
      "epoch": 0.9385099972610244,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008568121479552652,
      "loss": 4.9775,
      "step": 6853
    },
    {
      "epoch": 0.9386469460421802,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008567946345609473,
      "loss": 4.9746,
      "step": 6854
    },
    {
      "epoch": 0.938783894823336,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008567771177954254,
      "loss": 5.0519,
      "step": 6855
    },
    {
      "epoch": 0.9389208436044919,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008567595976588445,
      "loss": 4.9615,
      "step": 6856
    },
    {
      "epoch": 0.9390577923856478,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00085674207415135,
      "loss": 5.0335,
      "step": 6857
    },
    {
      "epoch": 0.9391947411668036,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008567245472730869,
      "loss": 4.9335,
      "step": 6858
    },
    {
      "epoch": 0.9393316899479595,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008567070170242005,
      "loss": 4.9684,
      "step": 6859
    },
    {
      "epoch": 0.9394686387291153,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008566894834048361,
      "loss": 4.9749,
      "step": 6860
    },
    {
      "epoch": 0.9396055875102711,
      "grad_norm": 0.5,
      "learning_rate": 0.0008566719464151393,
      "loss": 4.9751,
      "step": 6861
    },
    {
      "epoch": 0.939742536291427,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008566544060552551,
      "loss": 4.9544,
      "step": 6862
    },
    {
      "epoch": 0.9398794850725829,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008566368623253288,
      "loss": 5.0085,
      "step": 6863
    },
    {
      "epoch": 0.9400164338537387,
      "grad_norm": 0.4609375,
      "learning_rate": 0.000856619315225506,
      "loss": 5.0374,
      "step": 6864
    },
    {
      "epoch": 0.9401533826348946,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000856601764755932,
      "loss": 4.9725,
      "step": 6865
    },
    {
      "epoch": 0.9402903314160505,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008565842109167523,
      "loss": 5.0625,
      "step": 6866
    },
    {
      "epoch": 0.9404272801972062,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008565666537081124,
      "loss": 5.0096,
      "step": 6867
    },
    {
      "epoch": 0.9405642289783621,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008565490931301578,
      "loss": 4.9081,
      "step": 6868
    },
    {
      "epoch": 0.9407011777595179,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000856531529183034,
      "loss": 5.0214,
      "step": 6869
    },
    {
      "epoch": 0.9408381265406738,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008565139618668864,
      "loss": 4.9636,
      "step": 6870
    },
    {
      "epoch": 0.9409750753218297,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0008564963911818609,
      "loss": 4.9808,
      "step": 6871
    },
    {
      "epoch": 0.9411120241029854,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008564788171281028,
      "loss": 4.9592,
      "step": 6872
    },
    {
      "epoch": 0.9412489728841413,
      "grad_norm": 0.5,
      "learning_rate": 0.0008564612397057579,
      "loss": 4.9923,
      "step": 6873
    },
    {
      "epoch": 0.9413859216652972,
      "grad_norm": 0.484375,
      "learning_rate": 0.000856443658914972,
      "loss": 4.9547,
      "step": 6874
    },
    {
      "epoch": 0.941522870446453,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008564260747558906,
      "loss": 4.9108,
      "step": 6875
    },
    {
      "epoch": 0.9416598192276089,
      "grad_norm": 0.5,
      "learning_rate": 0.0008564084872286593,
      "loss": 5.035,
      "step": 6876
    },
    {
      "epoch": 0.9417967680087648,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008563908963334241,
      "loss": 4.9687,
      "step": 6877
    },
    {
      "epoch": 0.9419337167899205,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008563733020703309,
      "loss": 5.0011,
      "step": 6878
    },
    {
      "epoch": 0.9420706655710764,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008563557044395251,
      "loss": 4.9809,
      "step": 6879
    },
    {
      "epoch": 0.9422076143522322,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008563381034411528,
      "loss": 4.8985,
      "step": 6880
    },
    {
      "epoch": 0.9423445631333881,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008563204990753599,
      "loss": 4.9243,
      "step": 6881
    },
    {
      "epoch": 0.942481511914544,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000856302891342292,
      "loss": 5.0047,
      "step": 6882
    },
    {
      "epoch": 0.9426184606956998,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008562852802420954,
      "loss": 5.0127,
      "step": 6883
    },
    {
      "epoch": 0.9427554094768557,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008562676657749158,
      "loss": 4.9376,
      "step": 6884
    },
    {
      "epoch": 0.9428923582580115,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008562500479408993,
      "loss": 4.8931,
      "step": 6885
    },
    {
      "epoch": 0.9430293070391673,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008562324267401918,
      "loss": 4.9674,
      "step": 6886
    },
    {
      "epoch": 0.9431662558203232,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008562148021729394,
      "loss": 5.0298,
      "step": 6887
    },
    {
      "epoch": 0.9433032046014791,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008561971742392881,
      "loss": 4.925,
      "step": 6888
    },
    {
      "epoch": 0.9434401533826349,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008561795429393841,
      "loss": 5.0173,
      "step": 6889
    },
    {
      "epoch": 0.9435771021637908,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008561619082733733,
      "loss": 5.0087,
      "step": 6890
    },
    {
      "epoch": 0.9437140509449465,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000856144270241402,
      "loss": 4.9788,
      "step": 6891
    },
    {
      "epoch": 0.9438509997261024,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008561266288436165,
      "loss": 5.013,
      "step": 6892
    },
    {
      "epoch": 0.9439879485072583,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008561089840801629,
      "loss": 4.9543,
      "step": 6893
    },
    {
      "epoch": 0.9441248972884141,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008560913359511871,
      "loss": 4.9866,
      "step": 6894
    },
    {
      "epoch": 0.94426184606957,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000856073684456836,
      "loss": 4.9658,
      "step": 6895
    },
    {
      "epoch": 0.9443987948507259,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008560560295972552,
      "loss": 4.9822,
      "step": 6896
    },
    {
      "epoch": 0.9445357436318816,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008560383713725915,
      "loss": 5.0246,
      "step": 6897
    },
    {
      "epoch": 0.9446726924130375,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000856020709782991,
      "loss": 4.9942,
      "step": 6898
    },
    {
      "epoch": 0.9448096411941934,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008560030448286001,
      "loss": 5.0625,
      "step": 6899
    },
    {
      "epoch": 0.9449465899753492,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008559853765095652,
      "loss": 4.9211,
      "step": 6900
    },
    {
      "epoch": 0.9450835387565051,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008559677048260328,
      "loss": 4.9556,
      "step": 6901
    },
    {
      "epoch": 0.945220487537661,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008559500297781493,
      "loss": 4.9538,
      "step": 6902
    },
    {
      "epoch": 0.9453574363188167,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008559323513660611,
      "loss": 4.9133,
      "step": 6903
    },
    {
      "epoch": 0.9454943850999726,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0008559146695899148,
      "loss": 5.0474,
      "step": 6904
    },
    {
      "epoch": 0.9456313338811284,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000855896984449857,
      "loss": 4.8826,
      "step": 6905
    },
    {
      "epoch": 0.9457682826622843,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008558792959460342,
      "loss": 4.9812,
      "step": 6906
    },
    {
      "epoch": 0.9459052314434402,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008558616040785928,
      "loss": 4.95,
      "step": 6907
    },
    {
      "epoch": 0.946042180224596,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008558439088476795,
      "loss": 4.9294,
      "step": 6908
    },
    {
      "epoch": 0.9461791290057519,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008558262102534414,
      "loss": 4.8919,
      "step": 6909
    },
    {
      "epoch": 0.9463160777869077,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008558085082960245,
      "loss": 5.0174,
      "step": 6910
    },
    {
      "epoch": 0.9464530265680635,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008557908029755758,
      "loss": 4.9865,
      "step": 6911
    },
    {
      "epoch": 0.9465899753492194,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008557730942922421,
      "loss": 4.8692,
      "step": 6912
    },
    {
      "epoch": 0.9467269241303753,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00085575538224617,
      "loss": 4.9639,
      "step": 6913
    },
    {
      "epoch": 0.9468638729115311,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008557376668375065,
      "loss": 4.921,
      "step": 6914
    },
    {
      "epoch": 0.947000821692687,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008557199480663982,
      "loss": 4.9541,
      "step": 6915
    },
    {
      "epoch": 0.9471377704738427,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008557022259329919,
      "loss": 5.0196,
      "step": 6916
    },
    {
      "epoch": 0.9472747192549986,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008556845004374348,
      "loss": 4.9595,
      "step": 6917
    },
    {
      "epoch": 0.9474116680361545,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008556667715798733,
      "loss": 4.9284,
      "step": 6918
    },
    {
      "epoch": 0.9475486168173103,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008556490393604548,
      "loss": 4.9687,
      "step": 6919
    },
    {
      "epoch": 0.9476855655984662,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000855631303779326,
      "loss": 4.9362,
      "step": 6920
    },
    {
      "epoch": 0.9478225143796221,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008556135648366339,
      "loss": 4.9998,
      "step": 6921
    },
    {
      "epoch": 0.9479594631607778,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008555958225325255,
      "loss": 4.9928,
      "step": 6922
    },
    {
      "epoch": 0.9480964119419337,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000855578076867148,
      "loss": 4.9395,
      "step": 6923
    },
    {
      "epoch": 0.9482333607230896,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008555603278406482,
      "loss": 4.9417,
      "step": 6924
    },
    {
      "epoch": 0.9483703095042454,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008555425754531734,
      "loss": 4.9096,
      "step": 6925
    },
    {
      "epoch": 0.9485072582854013,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008555248197048708,
      "loss": 4.9372,
      "step": 6926
    },
    {
      "epoch": 0.948644207066557,
      "grad_norm": 0.375,
      "learning_rate": 0.0008555070605958872,
      "loss": 4.9645,
      "step": 6927
    },
    {
      "epoch": 0.9487811558477129,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008554892981263699,
      "loss": 4.9944,
      "step": 6928
    },
    {
      "epoch": 0.9489181046288688,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008554715322964664,
      "loss": 5.041,
      "step": 6929
    },
    {
      "epoch": 0.9490550534100246,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008554537631063234,
      "loss": 4.9697,
      "step": 6930
    },
    {
      "epoch": 0.9491920021911805,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008554359905560885,
      "loss": 4.945,
      "step": 6931
    },
    {
      "epoch": 0.9493289509723364,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008554182146459092,
      "loss": 4.9258,
      "step": 6932
    },
    {
      "epoch": 0.9494658997534922,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008554004353759322,
      "loss": 4.9391,
      "step": 6933
    },
    {
      "epoch": 0.949602848534648,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008553826527463055,
      "loss": 5.024,
      "step": 6934
    },
    {
      "epoch": 0.9497397973158039,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000855364866757176,
      "loss": 4.967,
      "step": 6935
    },
    {
      "epoch": 0.9498767460969597,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008553470774086912,
      "loss": 4.9555,
      "step": 6936
    },
    {
      "epoch": 0.9500136948781156,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008553292847009987,
      "loss": 4.9781,
      "step": 6937
    },
    {
      "epoch": 0.9501506436592715,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008553114886342458,
      "loss": 4.9361,
      "step": 6938
    },
    {
      "epoch": 0.9502875924404273,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008552936892085801,
      "loss": 4.9881,
      "step": 6939
    },
    {
      "epoch": 0.9504245412215832,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008552758864241488,
      "loss": 4.9902,
      "step": 6940
    },
    {
      "epoch": 0.9505614900027389,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008552580802810998,
      "loss": 5.0223,
      "step": 6941
    },
    {
      "epoch": 0.9506984387838948,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008552402707795806,
      "loss": 4.8554,
      "step": 6942
    },
    {
      "epoch": 0.9508353875650507,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008552224579197385,
      "loss": 5.0813,
      "step": 6943
    },
    {
      "epoch": 0.9509723363462065,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008552046417017214,
      "loss": 5.0485,
      "step": 6944
    },
    {
      "epoch": 0.9511092851273624,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008551868221256771,
      "loss": 5.0187,
      "step": 6945
    },
    {
      "epoch": 0.9512462339085183,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008551689991917528,
      "loss": 4.9947,
      "step": 6946
    },
    {
      "epoch": 0.951383182689674,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008551511729000965,
      "loss": 4.9407,
      "step": 6947
    },
    {
      "epoch": 0.9515201314708299,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008551333432508559,
      "loss": 4.8697,
      "step": 6948
    },
    {
      "epoch": 0.9516570802519858,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008551155102441789,
      "loss": 4.9371,
      "step": 6949
    },
    {
      "epoch": 0.9517940290331416,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000855097673880213,
      "loss": 4.9793,
      "step": 6950
    },
    {
      "epoch": 0.9519309778142975,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008550798341591062,
      "loss": 4.9715,
      "step": 6951
    },
    {
      "epoch": 0.9520679265954533,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008550619910810062,
      "loss": 4.8966,
      "step": 6952
    },
    {
      "epoch": 0.9522048753766091,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000855044144646061,
      "loss": 4.9285,
      "step": 6953
    },
    {
      "epoch": 0.952341824157765,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008550262948544184,
      "loss": 4.8858,
      "step": 6954
    },
    {
      "epoch": 0.9524787729389208,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008550084417062265,
      "loss": 4.9648,
      "step": 6955
    },
    {
      "epoch": 0.9526157217200767,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000854990585201633,
      "loss": 4.9582,
      "step": 6956
    },
    {
      "epoch": 0.9527526705012326,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008549727253407862,
      "loss": 4.9372,
      "step": 6957
    },
    {
      "epoch": 0.9528896192823884,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008549548621238338,
      "loss": 4.9729,
      "step": 6958
    },
    {
      "epoch": 0.9530265680635442,
      "grad_norm": 0.40625,
      "learning_rate": 0.000854936995550924,
      "loss": 4.9282,
      "step": 6959
    },
    {
      "epoch": 0.9531635168447001,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008549191256222048,
      "loss": 4.9063,
      "step": 6960
    },
    {
      "epoch": 0.9533004656258559,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008549012523378243,
      "loss": 5.0268,
      "step": 6961
    },
    {
      "epoch": 0.9534374144070118,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008548833756979306,
      "loss": 4.9112,
      "step": 6962
    },
    {
      "epoch": 0.9535743631881676,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000854865495702672,
      "loss": 5.0151,
      "step": 6963
    },
    {
      "epoch": 0.9537113119693235,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008548476123521965,
      "loss": 4.8517,
      "step": 6964
    },
    {
      "epoch": 0.9538482607504793,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008548297256466523,
      "loss": 4.8705,
      "step": 6965
    },
    {
      "epoch": 0.9539852095316351,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008548118355861879,
      "loss": 4.9819,
      "step": 6966
    },
    {
      "epoch": 0.954122158312791,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008547939421709512,
      "loss": 4.9254,
      "step": 6967
    },
    {
      "epoch": 0.9542591070939469,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008547760454010908,
      "loss": 5.0074,
      "step": 6968
    },
    {
      "epoch": 0.9543960558751027,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008547581452767547,
      "loss": 4.9726,
      "step": 6969
    },
    {
      "epoch": 0.9545330046562586,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008547402417980914,
      "loss": 4.8566,
      "step": 6970
    },
    {
      "epoch": 0.9546699534374145,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008547223349652493,
      "loss": 4.9707,
      "step": 6971
    },
    {
      "epoch": 0.9548069022185702,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008547044247783768,
      "loss": 4.9766,
      "step": 6972
    },
    {
      "epoch": 0.9549438509997261,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008546865112376223,
      "loss": 5.0101,
      "step": 6973
    },
    {
      "epoch": 0.9550807997808819,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008546685943431342,
      "loss": 4.9174,
      "step": 6974
    },
    {
      "epoch": 0.9552177485620378,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000854650674095061,
      "loss": 5.009,
      "step": 6975
    },
    {
      "epoch": 0.9553546973431937,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008546327504935514,
      "loss": 4.974,
      "step": 6976
    },
    {
      "epoch": 0.9554916461243494,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008546148235387536,
      "loss": 5.0178,
      "step": 6977
    },
    {
      "epoch": 0.9556285949055053,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008545968932308165,
      "loss": 4.9615,
      "step": 6978
    },
    {
      "epoch": 0.9557655436866612,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008545789595698885,
      "loss": 4.9697,
      "step": 6979
    },
    {
      "epoch": 0.955902492467817,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008545610225561181,
      "loss": 4.9388,
      "step": 6980
    },
    {
      "epoch": 0.9560394412489729,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008545430821896542,
      "loss": 4.9829,
      "step": 6981
    },
    {
      "epoch": 0.9561763900301288,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008545251384706455,
      "loss": 4.9925,
      "step": 6982
    },
    {
      "epoch": 0.9563133388112846,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008545071913992404,
      "loss": 4.9416,
      "step": 6983
    },
    {
      "epoch": 0.9564502875924404,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000854489240975588,
      "loss": 4.8246,
      "step": 6984
    },
    {
      "epoch": 0.9565872363735963,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008544712871998368,
      "loss": 4.9819,
      "step": 6985
    },
    {
      "epoch": 0.9567241851547521,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008544533300721356,
      "loss": 4.9314,
      "step": 6986
    },
    {
      "epoch": 0.956861133935908,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008544353695926334,
      "loss": 4.8442,
      "step": 6987
    },
    {
      "epoch": 0.9569980827170638,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008544174057614788,
      "loss": 4.9587,
      "step": 6988
    },
    {
      "epoch": 0.9571350314982197,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008543994385788208,
      "loss": 4.9603,
      "step": 6989
    },
    {
      "epoch": 0.9572719802793755,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008543814680448083,
      "loss": 4.864,
      "step": 6990
    },
    {
      "epoch": 0.9574089290605313,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008543634941595902,
      "loss": 4.9644,
      "step": 6991
    },
    {
      "epoch": 0.9575458778416872,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008543455169233155,
      "loss": 4.8946,
      "step": 6992
    },
    {
      "epoch": 0.9576828266228431,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008543275363361333,
      "loss": 4.9751,
      "step": 6993
    },
    {
      "epoch": 0.9578197754039989,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008543095523981922,
      "loss": 4.9597,
      "step": 6994
    },
    {
      "epoch": 0.9579567241851548,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008542915651096416,
      "loss": 4.9289,
      "step": 6995
    },
    {
      "epoch": 0.9580936729663106,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008542735744706305,
      "loss": 4.968,
      "step": 6996
    },
    {
      "epoch": 0.9582306217474664,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008542555804813078,
      "loss": 4.9779,
      "step": 6997
    },
    {
      "epoch": 0.9583675705286223,
      "grad_norm": 0.375,
      "learning_rate": 0.000854237583141823,
      "loss": 4.9465,
      "step": 6998
    },
    {
      "epoch": 0.9585045193097781,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000854219582452325,
      "loss": 4.9342,
      "step": 6999
    },
    {
      "epoch": 0.958641468090934,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008542015784129628,
      "loss": 4.9139,
      "step": 7000
    },
    {
      "epoch": 0.9587784168720899,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000854183571023886,
      "loss": 5.0123,
      "step": 7001
    },
    {
      "epoch": 0.9589153656532456,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008541655602852436,
      "loss": 5.0772,
      "step": 7002
    },
    {
      "epoch": 0.9590523144344015,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008541475461971848,
      "loss": 5.0006,
      "step": 7003
    },
    {
      "epoch": 0.9591892632155574,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008541295287598592,
      "loss": 4.995,
      "step": 7004
    },
    {
      "epoch": 0.9593262119967132,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008541115079734157,
      "loss": 4.9975,
      "step": 7005
    },
    {
      "epoch": 0.9594631607778691,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008540934838380038,
      "loss": 4.9447,
      "step": 7006
    },
    {
      "epoch": 0.959600109559025,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008540754563537731,
      "loss": 4.926,
      "step": 7007
    },
    {
      "epoch": 0.9597370583401807,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008540574255208725,
      "loss": 4.951,
      "step": 7008
    },
    {
      "epoch": 0.9598740071213366,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000854039391339452,
      "loss": 4.9988,
      "step": 7009
    },
    {
      "epoch": 0.9600109559024924,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008540213538096607,
      "loss": 4.9607,
      "step": 7010
    },
    {
      "epoch": 0.9601479046836483,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008540033129316482,
      "loss": 4.9645,
      "step": 7011
    },
    {
      "epoch": 0.9602848534648042,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008539852687055639,
      "loss": 4.9423,
      "step": 7012
    },
    {
      "epoch": 0.96042180224596,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008539672211315576,
      "loss": 4.9152,
      "step": 7013
    },
    {
      "epoch": 0.9605587510271159,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008539491702097786,
      "loss": 4.9384,
      "step": 7014
    },
    {
      "epoch": 0.9606956998082717,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008539311159403764,
      "loss": 4.9865,
      "step": 7015
    },
    {
      "epoch": 0.9608326485894275,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000853913058323501,
      "loss": 5.0312,
      "step": 7016
    },
    {
      "epoch": 0.9609695973705834,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008538949973593016,
      "loss": 4.9811,
      "step": 7017
    },
    {
      "epoch": 0.9611065461517393,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008538769330479282,
      "loss": 5.0342,
      "step": 7018
    },
    {
      "epoch": 0.9612434949328951,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008538588653895306,
      "loss": 4.8976,
      "step": 7019
    },
    {
      "epoch": 0.961380443714051,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008538407943842583,
      "loss": 4.9923,
      "step": 7020
    },
    {
      "epoch": 0.9615173924952068,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000853822720032261,
      "loss": 5.0077,
      "step": 7021
    },
    {
      "epoch": 0.9616543412763626,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008538046423336885,
      "loss": 5.0178,
      "step": 7022
    },
    {
      "epoch": 0.9617912900575185,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008537865612886909,
      "loss": 4.8871,
      "step": 7023
    },
    {
      "epoch": 0.9619282388386743,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008537684768974178,
      "loss": 4.946,
      "step": 7024
    },
    {
      "epoch": 0.9620651876198302,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008537503891600191,
      "loss": 5.0043,
      "step": 7025
    },
    {
      "epoch": 0.9622021364009861,
      "grad_norm": 0.375,
      "learning_rate": 0.0008537322980766448,
      "loss": 4.9713,
      "step": 7026
    },
    {
      "epoch": 0.9623390851821418,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008537142036474446,
      "loss": 5.0199,
      "step": 7027
    },
    {
      "epoch": 0.9624760339632977,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008536961058725687,
      "loss": 4.9247,
      "step": 7028
    },
    {
      "epoch": 0.9626129827444536,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008536780047521669,
      "loss": 4.9935,
      "step": 7029
    },
    {
      "epoch": 0.9627499315256094,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008536599002863894,
      "loss": 4.9825,
      "step": 7030
    },
    {
      "epoch": 0.9628868803067653,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008536417924753859,
      "loss": 4.9462,
      "step": 7031
    },
    {
      "epoch": 0.9630238290879212,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008536236813193069,
      "loss": 4.9222,
      "step": 7032
    },
    {
      "epoch": 0.963160777869077,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008536055668183022,
      "loss": 4.9819,
      "step": 7033
    },
    {
      "epoch": 0.9632977266502328,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008535874489725221,
      "loss": 4.8763,
      "step": 7034
    },
    {
      "epoch": 0.9634346754313886,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008535693277821164,
      "loss": 4.9067,
      "step": 7035
    },
    {
      "epoch": 0.9635716242125445,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008535512032472357,
      "loss": 4.9545,
      "step": 7036
    },
    {
      "epoch": 0.9637085729937004,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008535330753680299,
      "loss": 4.952,
      "step": 7037
    },
    {
      "epoch": 0.9638455217748562,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008535149441446495,
      "loss": 5.0075,
      "step": 7038
    },
    {
      "epoch": 0.963982470556012,
      "grad_norm": 0.375,
      "learning_rate": 0.0008534968095772445,
      "loss": 4.9369,
      "step": 7039
    },
    {
      "epoch": 0.9641194193371679,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008534786716659652,
      "loss": 4.9789,
      "step": 7040
    },
    {
      "epoch": 0.9642563681183237,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008534605304109622,
      "loss": 4.9357,
      "step": 7041
    },
    {
      "epoch": 0.9643933168994796,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008534423858123856,
      "loss": 4.9841,
      "step": 7042
    },
    {
      "epoch": 0.9645302656806355,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008534242378703857,
      "loss": 4.8868,
      "step": 7043
    },
    {
      "epoch": 0.9646672144617913,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000853406086585113,
      "loss": 4.9236,
      "step": 7044
    },
    {
      "epoch": 0.9648041632429472,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008533879319567181,
      "loss": 4.9857,
      "step": 7045
    },
    {
      "epoch": 0.9649411120241029,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008533697739853513,
      "loss": 5.0381,
      "step": 7046
    },
    {
      "epoch": 0.9650780608052588,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000853351612671163,
      "loss": 5.0015,
      "step": 7047
    },
    {
      "epoch": 0.9652150095864147,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008533334480143037,
      "loss": 4.9215,
      "step": 7048
    },
    {
      "epoch": 0.9653519583675705,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008533152800149239,
      "loss": 4.9408,
      "step": 7049
    },
    {
      "epoch": 0.9654889071487264,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008532971086731744,
      "loss": 4.9995,
      "step": 7050
    },
    {
      "epoch": 0.9656258559298823,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008532789339892057,
      "loss": 4.9661,
      "step": 7051
    },
    {
      "epoch": 0.965762804711038,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008532607559631684,
      "loss": 4.9593,
      "step": 7052
    },
    {
      "epoch": 0.9658997534921939,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000853242574595213,
      "loss": 4.8417,
      "step": 7053
    },
    {
      "epoch": 0.9660367022733498,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008532243898854904,
      "loss": 4.9437,
      "step": 7054
    },
    {
      "epoch": 0.9661736510545056,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008532062018341512,
      "loss": 5.0313,
      "step": 7055
    },
    {
      "epoch": 0.9663105998356615,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008531880104413461,
      "loss": 4.8903,
      "step": 7056
    },
    {
      "epoch": 0.9664475486168174,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008531698157072259,
      "loss": 5.0136,
      "step": 7057
    },
    {
      "epoch": 0.9665844973979731,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008531516176319413,
      "loss": 4.9311,
      "step": 7058
    },
    {
      "epoch": 0.966721446179129,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008531334162156433,
      "loss": 4.9633,
      "step": 7059
    },
    {
      "epoch": 0.9668583949602848,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008531152114584826,
      "loss": 4.9323,
      "step": 7060
    },
    {
      "epoch": 0.9669953437414407,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00085309700336061,
      "loss": 5.0387,
      "step": 7061
    },
    {
      "epoch": 0.9671322925225966,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008530787919221764,
      "loss": 4.9803,
      "step": 7062
    },
    {
      "epoch": 0.9672692413037524,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008530605771433329,
      "loss": 4.9342,
      "step": 7063
    },
    {
      "epoch": 0.9674061900849082,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008530423590242304,
      "loss": 4.995,
      "step": 7064
    },
    {
      "epoch": 0.9675431388660641,
      "grad_norm": 0.375,
      "learning_rate": 0.0008530241375650198,
      "loss": 4.9675,
      "step": 7065
    },
    {
      "epoch": 0.9676800876472199,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008530059127658522,
      "loss": 4.8716,
      "step": 7066
    },
    {
      "epoch": 0.9678170364283758,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008529876846268786,
      "loss": 4.9034,
      "step": 7067
    },
    {
      "epoch": 0.9679539852095317,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008529694531482498,
      "loss": 4.9609,
      "step": 7068
    },
    {
      "epoch": 0.9680909339906875,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008529512183301172,
      "loss": 4.9726,
      "step": 7069
    },
    {
      "epoch": 0.9682278827718434,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008529329801726319,
      "loss": 4.9809,
      "step": 7070
    },
    {
      "epoch": 0.9683648315529991,
      "grad_norm": 0.40625,
      "learning_rate": 0.000852914738675945,
      "loss": 4.9222,
      "step": 7071
    },
    {
      "epoch": 0.968501780334155,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008528964938402077,
      "loss": 5.0339,
      "step": 7072
    },
    {
      "epoch": 0.9686387291153109,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008528782456655711,
      "loss": 4.8552,
      "step": 7073
    },
    {
      "epoch": 0.9687756778964667,
      "grad_norm": 0.375,
      "learning_rate": 0.0008528599941521862,
      "loss": 5.022,
      "step": 7074
    },
    {
      "epoch": 0.9689126266776226,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000852841739300205,
      "loss": 4.9939,
      "step": 7075
    },
    {
      "epoch": 0.9690495754587785,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000852823481109778,
      "loss": 4.9432,
      "step": 7076
    },
    {
      "epoch": 0.9691865242399342,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008528052195810569,
      "loss": 4.9309,
      "step": 7077
    },
    {
      "epoch": 0.9693234730210901,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000852786954714193,
      "loss": 4.9215,
      "step": 7078
    },
    {
      "epoch": 0.969460421802246,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008527686865093377,
      "loss": 4.9801,
      "step": 7079
    },
    {
      "epoch": 0.9695973705834018,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008527504149666422,
      "loss": 4.9342,
      "step": 7080
    },
    {
      "epoch": 0.9697343193645577,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008527321400862581,
      "loss": 4.8765,
      "step": 7081
    },
    {
      "epoch": 0.9698712681457134,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008527138618683366,
      "loss": 4.8934,
      "step": 7082
    },
    {
      "epoch": 0.9700082169268693,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008526955803130296,
      "loss": 5.0213,
      "step": 7083
    },
    {
      "epoch": 0.9701451657080252,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008526772954204883,
      "loss": 4.9523,
      "step": 7084
    },
    {
      "epoch": 0.970282114489181,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008526590071908643,
      "loss": 4.924,
      "step": 7085
    },
    {
      "epoch": 0.9704190632703369,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008526407156243092,
      "loss": 4.9586,
      "step": 7086
    },
    {
      "epoch": 0.9705560120514928,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008526224207209744,
      "loss": 4.9197,
      "step": 7087
    },
    {
      "epoch": 0.9706929608326486,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008526041224810117,
      "loss": 4.9818,
      "step": 7088
    },
    {
      "epoch": 0.9708299096138044,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008525858209045728,
      "loss": 4.9504,
      "step": 7089
    },
    {
      "epoch": 0.9709668583949603,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008525675159918092,
      "loss": 4.9445,
      "step": 7090
    },
    {
      "epoch": 0.9711038071761161,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008525492077428728,
      "loss": 4.9965,
      "step": 7091
    },
    {
      "epoch": 0.971240755957272,
      "grad_norm": 0.47265625,
      "learning_rate": 0.000852530896157915,
      "loss": 4.9749,
      "step": 7092
    },
    {
      "epoch": 0.9713777047384279,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008525125812370879,
      "loss": 4.9372,
      "step": 7093
    },
    {
      "epoch": 0.9715146535195837,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000852494262980543,
      "loss": 5.0376,
      "step": 7094
    },
    {
      "epoch": 0.9716516023007395,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008524759413884324,
      "loss": 5.0083,
      "step": 7095
    },
    {
      "epoch": 0.9717885510818953,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008524576164609076,
      "loss": 4.9746,
      "step": 7096
    },
    {
      "epoch": 0.9719254998630512,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008524392881981207,
      "loss": 4.9354,
      "step": 7097
    },
    {
      "epoch": 0.9720624486442071,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008524209566002236,
      "loss": 5.0037,
      "step": 7098
    },
    {
      "epoch": 0.9721993974253629,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008524026216673681,
      "loss": 5.0235,
      "step": 7099
    },
    {
      "epoch": 0.9723363462065188,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008523842833997061,
      "loss": 4.8689,
      "step": 7100
    },
    {
      "epoch": 0.9724732949876747,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008523659417973898,
      "loss": 4.9686,
      "step": 7101
    },
    {
      "epoch": 0.9726102437688304,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008523475968605709,
      "loss": 4.9051,
      "step": 7102
    },
    {
      "epoch": 0.9727471925499863,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008523292485894016,
      "loss": 4.933,
      "step": 7103
    },
    {
      "epoch": 0.9728841413311422,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000852310896984034,
      "loss": 5.0287,
      "step": 7104
    },
    {
      "epoch": 0.973021090112298,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008522925420446202,
      "loss": 4.9887,
      "step": 7105
    },
    {
      "epoch": 0.9731580388934539,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008522741837713123,
      "loss": 5.0174,
      "step": 7106
    },
    {
      "epoch": 0.9732949876746096,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008522558221642622,
      "loss": 4.9669,
      "step": 7107
    },
    {
      "epoch": 0.9734319364557655,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008522374572236222,
      "loss": 4.9135,
      "step": 7108
    },
    {
      "epoch": 0.9735688852369214,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008522190889495448,
      "loss": 4.9261,
      "step": 7109
    },
    {
      "epoch": 0.9737058340180772,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008522007173421818,
      "loss": 5.075,
      "step": 7110
    },
    {
      "epoch": 0.9738427827992331,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008521823424016856,
      "loss": 4.9679,
      "step": 7111
    },
    {
      "epoch": 0.973979731580389,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008521639641282086,
      "loss": 4.9233,
      "step": 7112
    },
    {
      "epoch": 0.9741166803615448,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008521455825219028,
      "loss": 4.9566,
      "step": 7113
    },
    {
      "epoch": 0.9742536291427006,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008521271975829209,
      "loss": 4.92,
      "step": 7114
    },
    {
      "epoch": 0.9743905779238565,
      "grad_norm": 0.390625,
      "learning_rate": 0.000852108809311415,
      "loss": 4.9664,
      "step": 7115
    },
    {
      "epoch": 0.9745275267050123,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008520904177075377,
      "loss": 4.9314,
      "step": 7116
    },
    {
      "epoch": 0.9746644754861682,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008520720227714413,
      "loss": 4.9417,
      "step": 7117
    },
    {
      "epoch": 0.974801424267324,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000852053624503278,
      "loss": 4.922,
      "step": 7118
    },
    {
      "epoch": 0.9749383730484799,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008520352229032008,
      "loss": 4.9338,
      "step": 7119
    },
    {
      "epoch": 0.9750753218296357,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008520168179713618,
      "loss": 4.982,
      "step": 7120
    },
    {
      "epoch": 0.9752122706107915,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008519984097079136,
      "loss": 5.0597,
      "step": 7121
    },
    {
      "epoch": 0.9753492193919474,
      "grad_norm": 0.375,
      "learning_rate": 0.0008519799981130088,
      "loss": 4.9204,
      "step": 7122
    },
    {
      "epoch": 0.9754861681731033,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008519615831868,
      "loss": 4.9839,
      "step": 7123
    },
    {
      "epoch": 0.9756231169542591,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008519431649294398,
      "loss": 4.9964,
      "step": 7124
    },
    {
      "epoch": 0.975760065735415,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008519247433410806,
      "loss": 4.9105,
      "step": 7125
    },
    {
      "epoch": 0.9758970145165708,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008519063184218755,
      "loss": 4.9741,
      "step": 7126
    },
    {
      "epoch": 0.9760339632977266,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008518878901719771,
      "loss": 4.9265,
      "step": 7127
    },
    {
      "epoch": 0.9761709120788825,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008518694585915378,
      "loss": 4.9521,
      "step": 7128
    },
    {
      "epoch": 0.9763078608600383,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008518510236807105,
      "loss": 4.9208,
      "step": 7129
    },
    {
      "epoch": 0.9764448096411942,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008518325854396479,
      "loss": 4.8867,
      "step": 7130
    },
    {
      "epoch": 0.9765817584223501,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008518141438685032,
      "loss": 4.9182,
      "step": 7131
    },
    {
      "epoch": 0.9767187072035058,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008517956989674289,
      "loss": 4.9863,
      "step": 7132
    },
    {
      "epoch": 0.9768556559846617,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008517772507365777,
      "loss": 5.0046,
      "step": 7133
    },
    {
      "epoch": 0.9769926047658176,
      "grad_norm": 0.375,
      "learning_rate": 0.0008517587991761028,
      "loss": 4.9371,
      "step": 7134
    },
    {
      "epoch": 0.9771295535469734,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008517403442861569,
      "loss": 4.8766,
      "step": 7135
    },
    {
      "epoch": 0.9772665023281293,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008517218860668933,
      "loss": 4.9246,
      "step": 7136
    },
    {
      "epoch": 0.9774034511092852,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008517034245184645,
      "loss": 4.9499,
      "step": 7137
    },
    {
      "epoch": 0.977540399890441,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008516849596410238,
      "loss": 4.9433,
      "step": 7138
    },
    {
      "epoch": 0.9776773486715968,
      "grad_norm": 0.390625,
      "learning_rate": 0.000851666491434724,
      "loss": 4.963,
      "step": 7139
    },
    {
      "epoch": 0.9778142974527527,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008516480198997183,
      "loss": 4.8733,
      "step": 7140
    },
    {
      "epoch": 0.9779512462339085,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008516295450361599,
      "loss": 4.9793,
      "step": 7141
    },
    {
      "epoch": 0.9780881950150644,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008516110668442015,
      "loss": 4.8936,
      "step": 7142
    },
    {
      "epoch": 0.9782251437962202,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008515925853239967,
      "loss": 4.9743,
      "step": 7143
    },
    {
      "epoch": 0.978362092577376,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008515741004756983,
      "loss": 4.9349,
      "step": 7144
    },
    {
      "epoch": 0.9784990413585319,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008515556122994598,
      "loss": 4.9974,
      "step": 7145
    },
    {
      "epoch": 0.9786359901396877,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008515371207954342,
      "loss": 5.0052,
      "step": 7146
    },
    {
      "epoch": 0.9787729389208436,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008515186259637747,
      "loss": 4.8956,
      "step": 7147
    },
    {
      "epoch": 0.9789098877019995,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008515001278046348,
      "loss": 4.9474,
      "step": 7148
    },
    {
      "epoch": 0.9790468364831553,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008514816263181676,
      "loss": 4.9124,
      "step": 7149
    },
    {
      "epoch": 0.9791837852643112,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008514631215045266,
      "loss": 4.9661,
      "step": 7150
    },
    {
      "epoch": 0.979320734045467,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008514446133638649,
      "loss": 5.0257,
      "step": 7151
    },
    {
      "epoch": 0.9794576828266228,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008514261018963361,
      "loss": 4.988,
      "step": 7152
    },
    {
      "epoch": 0.9795946316077787,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008514075871020935,
      "loss": 4.9551,
      "step": 7153
    },
    {
      "epoch": 0.9797315803889345,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008513890689812905,
      "loss": 4.9309,
      "step": 7154
    },
    {
      "epoch": 0.9798685291700904,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008513705475340808,
      "loss": 4.8626,
      "step": 7155
    },
    {
      "epoch": 0.9800054779512463,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008513520227606178,
      "loss": 4.8826,
      "step": 7156
    },
    {
      "epoch": 0.980142426732402,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008513334946610549,
      "loss": 5.0394,
      "step": 7157
    },
    {
      "epoch": 0.9802793755135579,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008513149632355455,
      "loss": 4.8573,
      "step": 7158
    },
    {
      "epoch": 0.9804163242947138,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008512964284842435,
      "loss": 4.9672,
      "step": 7159
    },
    {
      "epoch": 0.9805532730758696,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008512778904073024,
      "loss": 4.8563,
      "step": 7160
    },
    {
      "epoch": 0.9806902218570255,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008512593490048757,
      "loss": 4.9342,
      "step": 7161
    },
    {
      "epoch": 0.9808271706381814,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008512408042771172,
      "loss": 4.9777,
      "step": 7162
    },
    {
      "epoch": 0.9809641194193371,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008512222562241807,
      "loss": 4.9773,
      "step": 7163
    },
    {
      "epoch": 0.981101068200493,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008512037048462196,
      "loss": 4.9603,
      "step": 7164
    },
    {
      "epoch": 0.9812380169816488,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008511851501433878,
      "loss": 4.9685,
      "step": 7165
    },
    {
      "epoch": 0.9813749657628047,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008511665921158391,
      "loss": 4.9107,
      "step": 7166
    },
    {
      "epoch": 0.9815119145439606,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008511480307637272,
      "loss": 4.9447,
      "step": 7167
    },
    {
      "epoch": 0.9816488633251164,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000851129466087206,
      "loss": 4.8545,
      "step": 7168
    },
    {
      "epoch": 0.9817858121062722,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008511108980864294,
      "loss": 4.9472,
      "step": 7169
    },
    {
      "epoch": 0.9819227608874281,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008510923267615511,
      "loss": 4.9231,
      "step": 7170
    },
    {
      "epoch": 0.9820597096685839,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008510737521127253,
      "loss": 4.9054,
      "step": 7171
    },
    {
      "epoch": 0.9821966584497398,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008510551741401056,
      "loss": 5.0264,
      "step": 7172
    },
    {
      "epoch": 0.9823336072308957,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008510365928438462,
      "loss": 4.9713,
      "step": 7173
    },
    {
      "epoch": 0.9824705560120515,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008510180082241008,
      "loss": 4.9769,
      "step": 7174
    },
    {
      "epoch": 0.9826075047932074,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008509994202810238,
      "loss": 4.9044,
      "step": 7175
    },
    {
      "epoch": 0.9827444535743632,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000850980829014769,
      "loss": 4.9726,
      "step": 7176
    },
    {
      "epoch": 0.982881402355519,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008509622344254904,
      "loss": 4.8585,
      "step": 7177
    },
    {
      "epoch": 0.9830183511366749,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008509436365133424,
      "loss": 4.997,
      "step": 7178
    },
    {
      "epoch": 0.9831552999178307,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008509250352784789,
      "loss": 4.9487,
      "step": 7179
    },
    {
      "epoch": 0.9832922486989866,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008509064307210541,
      "loss": 4.8669,
      "step": 7180
    },
    {
      "epoch": 0.9834291974801425,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008508878228412222,
      "loss": 4.9457,
      "step": 7181
    },
    {
      "epoch": 0.9835661462612982,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008508692116391373,
      "loss": 4.8657,
      "step": 7182
    },
    {
      "epoch": 0.9837030950424541,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008508505971149537,
      "loss": 4.857,
      "step": 7183
    },
    {
      "epoch": 0.98384004382361,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008508319792688258,
      "loss": 5.0,
      "step": 7184
    },
    {
      "epoch": 0.9839769926047658,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008508133581009076,
      "loss": 5.0582,
      "step": 7185
    },
    {
      "epoch": 0.9841139413859217,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008507947336113537,
      "loss": 4.9553,
      "step": 7186
    },
    {
      "epoch": 0.9842508901670776,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008507761058003183,
      "loss": 4.9123,
      "step": 7187
    },
    {
      "epoch": 0.9843878389482333,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008507574746679559,
      "loss": 4.9474,
      "step": 7188
    },
    {
      "epoch": 0.9845247877293892,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0008507388402144208,
      "loss": 4.9885,
      "step": 7189
    },
    {
      "epoch": 0.984661736510545,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008507202024398674,
      "loss": 4.9958,
      "step": 7190
    },
    {
      "epoch": 0.9847986852917009,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008507015613444502,
      "loss": 4.8985,
      "step": 7191
    },
    {
      "epoch": 0.9849356340728568,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008506829169283237,
      "loss": 5.0283,
      "step": 7192
    },
    {
      "epoch": 0.9850725828540126,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008506642691916424,
      "loss": 4.8863,
      "step": 7193
    },
    {
      "epoch": 0.9852095316351684,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008506456181345608,
      "loss": 5.003,
      "step": 7194
    },
    {
      "epoch": 0.9853464804163243,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008506269637572335,
      "loss": 5.0447,
      "step": 7195
    },
    {
      "epoch": 0.9854834291974801,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008506083060598151,
      "loss": 4.8516,
      "step": 7196
    },
    {
      "epoch": 0.985620377978636,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008505896450424602,
      "loss": 4.9046,
      "step": 7197
    },
    {
      "epoch": 0.9857573267597919,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008505709807053234,
      "loss": 4.9492,
      "step": 7198
    },
    {
      "epoch": 0.9858942755409477,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008505523130485595,
      "loss": 4.9534,
      "step": 7199
    },
    {
      "epoch": 0.9860312243221035,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008505336420723231,
      "loss": 4.8802,
      "step": 7200
    },
    {
      "epoch": 0.9861681731032593,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000850514967776769,
      "loss": 4.9306,
      "step": 7201
    },
    {
      "epoch": 0.9863051218844152,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008504962901620519,
      "loss": 4.9809,
      "step": 7202
    },
    {
      "epoch": 0.9864420706655711,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008504776092283266,
      "loss": 4.9506,
      "step": 7203
    },
    {
      "epoch": 0.9865790194467269,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008504589249757479,
      "loss": 5.043,
      "step": 7204
    },
    {
      "epoch": 0.9867159682278828,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008504402374044707,
      "loss": 5.0,
      "step": 7205
    },
    {
      "epoch": 0.9868529170090387,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008504215465146498,
      "loss": 5.007,
      "step": 7206
    },
    {
      "epoch": 0.9869898657901944,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008504028523064403,
      "loss": 5.0415,
      "step": 7207
    },
    {
      "epoch": 0.9871268145713503,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008503841547799967,
      "loss": 4.9702,
      "step": 7208
    },
    {
      "epoch": 0.9872637633525062,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008503654539354743,
      "loss": 4.919,
      "step": 7209
    },
    {
      "epoch": 0.987400712133662,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000850346749773028,
      "loss": 4.856,
      "step": 7210
    },
    {
      "epoch": 0.9875376609148179,
      "grad_norm": 0.5,
      "learning_rate": 0.0008503280422928128,
      "loss": 5.0249,
      "step": 7211
    },
    {
      "epoch": 0.9876746096959738,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008503093314949839,
      "loss": 5.0277,
      "step": 7212
    },
    {
      "epoch": 0.9878115584771295,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008502906173796958,
      "loss": 4.9652,
      "step": 7213
    },
    {
      "epoch": 0.9879485072582854,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008502718999471041,
      "loss": 4.9147,
      "step": 7214
    },
    {
      "epoch": 0.9880854560394412,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000850253179197364,
      "loss": 4.9889,
      "step": 7215
    },
    {
      "epoch": 0.9882224048205971,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008502344551306303,
      "loss": 4.9717,
      "step": 7216
    },
    {
      "epoch": 0.988359353601753,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008502157277470582,
      "loss": 4.9574,
      "step": 7217
    },
    {
      "epoch": 0.9884963023829088,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008501969970468032,
      "loss": 5.0469,
      "step": 7218
    },
    {
      "epoch": 0.9886332511640646,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008501782630300201,
      "loss": 4.8831,
      "step": 7219
    },
    {
      "epoch": 0.9887701999452205,
      "grad_norm": 0.375,
      "learning_rate": 0.0008501595256968645,
      "loss": 4.9925,
      "step": 7220
    },
    {
      "epoch": 0.9889071487263763,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008501407850474915,
      "loss": 4.914,
      "step": 7221
    },
    {
      "epoch": 0.9890440975075322,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008501220410820565,
      "loss": 4.9019,
      "step": 7222
    },
    {
      "epoch": 0.9891810462886881,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008501032938007148,
      "loss": 4.9582,
      "step": 7223
    },
    {
      "epoch": 0.9893179950698439,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008500845432036218,
      "loss": 4.9768,
      "step": 7224
    },
    {
      "epoch": 0.9894549438509997,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008500657892909327,
      "loss": 4.9952,
      "step": 7225
    },
    {
      "epoch": 0.9895918926321555,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008500470320628033,
      "loss": 5.0379,
      "step": 7226
    },
    {
      "epoch": 0.9897288414133114,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008500282715193887,
      "loss": 4.9887,
      "step": 7227
    },
    {
      "epoch": 0.9898657901944673,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008500095076608444,
      "loss": 4.9249,
      "step": 7228
    },
    {
      "epoch": 0.9900027389756231,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008499907404873262,
      "loss": 4.8714,
      "step": 7229
    },
    {
      "epoch": 0.990139687756779,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008499719699989893,
      "loss": 4.8911,
      "step": 7230
    },
    {
      "epoch": 0.9902766365379349,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008499531961959894,
      "loss": 5.0529,
      "step": 7231
    },
    {
      "epoch": 0.9904135853190906,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008499344190784821,
      "loss": 4.8906,
      "step": 7232
    },
    {
      "epoch": 0.9905505341002465,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008499156386466229,
      "loss": 4.9256,
      "step": 7233
    },
    {
      "epoch": 0.9906874828814024,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008498968549005675,
      "loss": 4.9192,
      "step": 7234
    },
    {
      "epoch": 0.9908244316625582,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008498780678404716,
      "loss": 4.9233,
      "step": 7235
    },
    {
      "epoch": 0.9909613804437141,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008498592774664908,
      "loss": 4.9808,
      "step": 7236
    },
    {
      "epoch": 0.9910983292248698,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008498404837787811,
      "loss": 4.9883,
      "step": 7237
    },
    {
      "epoch": 0.9912352780060257,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008498216867774979,
      "loss": 4.938,
      "step": 7238
    },
    {
      "epoch": 0.9913722267871816,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000849802886462797,
      "loss": 4.8987,
      "step": 7239
    },
    {
      "epoch": 0.9915091755683374,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008497840828348344,
      "loss": 5.0506,
      "step": 7240
    },
    {
      "epoch": 0.9916461243494933,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008497652758937659,
      "loss": 4.9826,
      "step": 7241
    },
    {
      "epoch": 0.9917830731306492,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008497464656397472,
      "loss": 4.9532,
      "step": 7242
    },
    {
      "epoch": 0.991920021911805,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008497276520729344,
      "loss": 4.962,
      "step": 7243
    },
    {
      "epoch": 0.9920569706929608,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008497088351934832,
      "loss": 4.9772,
      "step": 7244
    },
    {
      "epoch": 0.9921939194741167,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008496900150015498,
      "loss": 4.9443,
      "step": 7245
    },
    {
      "epoch": 0.9923308682552725,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008496711914972898,
      "loss": 4.9241,
      "step": 7246
    },
    {
      "epoch": 0.9924678170364284,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008496523646808595,
      "loss": 4.8979,
      "step": 7247
    },
    {
      "epoch": 0.9926047658175843,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008496335345524147,
      "loss": 5.0824,
      "step": 7248
    },
    {
      "epoch": 0.99274171459874,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0008496147011121118,
      "loss": 4.8962,
      "step": 7249
    },
    {
      "epoch": 0.9928786633798959,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008495958643601065,
      "loss": 4.9472,
      "step": 7250
    },
    {
      "epoch": 0.9930156121610517,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008495770242965549,
      "loss": 4.909,
      "step": 7251
    },
    {
      "epoch": 0.9931525609422076,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008495581809216136,
      "loss": 5.0047,
      "step": 7252
    },
    {
      "epoch": 0.9932895097233635,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008495393342354382,
      "loss": 4.9594,
      "step": 7253
    },
    {
      "epoch": 0.9934264585045193,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008495204842381853,
      "loss": 4.9585,
      "step": 7254
    },
    {
      "epoch": 0.9935634072856752,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008495016309300109,
      "loss": 5.0134,
      "step": 7255
    },
    {
      "epoch": 0.993700356066831,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008494827743110712,
      "loss": 5.052,
      "step": 7256
    },
    {
      "epoch": 0.9938373048479868,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008494639143815227,
      "loss": 5.0014,
      "step": 7257
    },
    {
      "epoch": 0.9939742536291427,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008494450511415216,
      "loss": 4.8584,
      "step": 7258
    },
    {
      "epoch": 0.9941112024102986,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000849426184591224,
      "loss": 4.8601,
      "step": 7259
    },
    {
      "epoch": 0.9942481511914544,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008494073147307865,
      "loss": 4.882,
      "step": 7260
    },
    {
      "epoch": 0.9943850999726103,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008493884415603654,
      "loss": 5.0135,
      "step": 7261
    },
    {
      "epoch": 0.994522048753766,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008493695650801171,
      "loss": 5.0295,
      "step": 7262
    },
    {
      "epoch": 0.9946589975349219,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008493506852901981,
      "loss": 4.9696,
      "step": 7263
    },
    {
      "epoch": 0.9947959463160778,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008493318021907648,
      "loss": 5.0176,
      "step": 7264
    },
    {
      "epoch": 0.9949328950972336,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008493129157819735,
      "loss": 4.978,
      "step": 7265
    },
    {
      "epoch": 0.9950698438783895,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008492940260639812,
      "loss": 4.923,
      "step": 7266
    },
    {
      "epoch": 0.9952067926595454,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000849275133036944,
      "loss": 4.9724,
      "step": 7267
    },
    {
      "epoch": 0.9953437414407011,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008492562367010188,
      "loss": 4.9555,
      "step": 7268
    },
    {
      "epoch": 0.995480690221857,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008492373370563619,
      "loss": 5.004,
      "step": 7269
    },
    {
      "epoch": 0.9956176390030129,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008492184341031301,
      "loss": 4.9598,
      "step": 7270
    },
    {
      "epoch": 0.9957545877841687,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008491995278414799,
      "loss": 4.9224,
      "step": 7271
    },
    {
      "epoch": 0.9958915365653246,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008491806182715681,
      "loss": 4.9853,
      "step": 7272
    },
    {
      "epoch": 0.9960284853464804,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008491617053935515,
      "loss": 4.9726,
      "step": 7273
    },
    {
      "epoch": 0.9961654341276363,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008491427892075868,
      "loss": 4.9453,
      "step": 7274
    },
    {
      "epoch": 0.9963023829087921,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008491238697138306,
      "loss": 4.9258,
      "step": 7275
    },
    {
      "epoch": 0.9964393316899479,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008491049469124398,
      "loss": 5.006,
      "step": 7276
    },
    {
      "epoch": 0.9965762804711038,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000849086020803571,
      "loss": 5.0144,
      "step": 7277
    },
    {
      "epoch": 0.9967132292522597,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008490670913873815,
      "loss": 4.8541,
      "step": 7278
    },
    {
      "epoch": 0.9968501780334155,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008490481586640278,
      "loss": 4.9201,
      "step": 7279
    },
    {
      "epoch": 0.9969871268145714,
      "grad_norm": 0.390625,
      "learning_rate": 0.000849029222633667,
      "loss": 4.9553,
      "step": 7280
    },
    {
      "epoch": 0.9971240755957272,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008490102832964559,
      "loss": 4.9366,
      "step": 7281
    },
    {
      "epoch": 0.997261024376883,
      "grad_norm": 0.375,
      "learning_rate": 0.0008489913406525515,
      "loss": 4.9101,
      "step": 7282
    },
    {
      "epoch": 0.9973979731580389,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008489723947021108,
      "loss": 5.0077,
      "step": 7283
    },
    {
      "epoch": 0.9975349219391948,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008489534454452908,
      "loss": 4.9464,
      "step": 7284
    },
    {
      "epoch": 0.9976718707203506,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008489344928822486,
      "loss": 4.9092,
      "step": 7285
    },
    {
      "epoch": 0.9978088195015065,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008489155370131412,
      "loss": 4.9541,
      "step": 7286
    },
    {
      "epoch": 0.9979457682826622,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008488965778381256,
      "loss": 4.9833,
      "step": 7287
    },
    {
      "epoch": 0.9980827170638181,
      "grad_norm": 0.421875,
      "learning_rate": 0.000848877615357359,
      "loss": 5.0015,
      "step": 7288
    },
    {
      "epoch": 0.998219665844974,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008488586495709985,
      "loss": 4.9943,
      "step": 7289
    },
    {
      "epoch": 0.9983566146261298,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008488396804792015,
      "loss": 4.9134,
      "step": 7290
    },
    {
      "epoch": 0.9984935634072857,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000848820708082125,
      "loss": 4.9228,
      "step": 7291
    },
    {
      "epoch": 0.9986305121884416,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008488017323799261,
      "loss": 5.0117,
      "step": 7292
    },
    {
      "epoch": 0.9987674609695973,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008487827533727624,
      "loss": 4.9037,
      "step": 7293
    },
    {
      "epoch": 0.9989044097507532,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008487637710607908,
      "loss": 4.9724,
      "step": 7294
    },
    {
      "epoch": 0.9990413585319091,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008487447854441689,
      "loss": 4.958,
      "step": 7295
    },
    {
      "epoch": 0.9991783073130649,
      "grad_norm": 0.34375,
      "learning_rate": 0.000848725796523054,
      "loss": 4.9524,
      "step": 7296
    },
    {
      "epoch": 0.9993152560942208,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008487068042976034,
      "loss": 4.9467,
      "step": 7297
    },
    {
      "epoch": 0.9994522048753766,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008486878087679745,
      "loss": 4.943,
      "step": 7298
    },
    {
      "epoch": 0.9995891536565324,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008486688099343246,
      "loss": 4.8913,
      "step": 7299
    },
    {
      "epoch": 0.9997261024376883,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008486498077968115,
      "loss": 4.9688,
      "step": 7300
    },
    {
      "epoch": 0.9998630512188441,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008486308023555923,
      "loss": 4.899,
      "step": 7301
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008486117936108247,
      "loss": 4.9298,
      "step": 7302
    },
    {
      "epoch": 1.0001369487811558,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008485927815626662,
      "loss": 4.9571,
      "step": 7303
    },
    {
      "epoch": 1.0002738975623118,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008485737662112743,
      "loss": 4.9636,
      "step": 7304
    },
    {
      "epoch": 1.0004108463434676,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008485547475568066,
      "loss": 4.8671,
      "step": 7305
    },
    {
      "epoch": 1.0005477951246233,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008485357255994208,
      "loss": 4.9459,
      "step": 7306
    },
    {
      "epoch": 1.0006847439057793,
      "grad_norm": 0.375,
      "learning_rate": 0.0008485167003392744,
      "loss": 5.0305,
      "step": 7307
    },
    {
      "epoch": 1.000821692686935,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008484976717765252,
      "loss": 4.9162,
      "step": 7308
    },
    {
      "epoch": 1.0009586414680909,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008484786399113308,
      "loss": 4.8986,
      "step": 7309
    },
    {
      "epoch": 1.0010955902492469,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000848459604743849,
      "loss": 4.9628,
      "step": 7310
    },
    {
      "epoch": 1.0012325390304027,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008484405662742376,
      "loss": 4.8915,
      "step": 7311
    },
    {
      "epoch": 1.0013694878115584,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008484215245026541,
      "loss": 4.9201,
      "step": 7312
    },
    {
      "epoch": 1.0015064365927144,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008484024794292565,
      "loss": 4.92,
      "step": 7313
    },
    {
      "epoch": 1.0016433853738702,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008483834310542027,
      "loss": 5.0296,
      "step": 7314
    },
    {
      "epoch": 1.001780334155026,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008483643793776505,
      "loss": 4.9524,
      "step": 7315
    },
    {
      "epoch": 1.0019172829361818,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008483453243997577,
      "loss": 4.9603,
      "step": 7316
    },
    {
      "epoch": 1.0020542317173378,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008483262661206823,
      "loss": 4.9958,
      "step": 7317
    },
    {
      "epoch": 1.0021911804984935,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008483072045405822,
      "loss": 4.9677,
      "step": 7318
    },
    {
      "epoch": 1.0023281292796493,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008482881396596154,
      "loss": 4.9523,
      "step": 7319
    },
    {
      "epoch": 1.0024650780608053,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00084826907147794,
      "loss": 5.0238,
      "step": 7320
    },
    {
      "epoch": 1.002602026841961,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008482499999957138,
      "loss": 4.9584,
      "step": 7321
    },
    {
      "epoch": 1.0027389756231169,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008482309252130949,
      "loss": 4.8444,
      "step": 7322
    },
    {
      "epoch": 1.0028759244042729,
      "grad_norm": 0.375,
      "learning_rate": 0.0008482118471302414,
      "loss": 4.9711,
      "step": 7323
    },
    {
      "epoch": 1.0030128731854286,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008481927657473115,
      "loss": 4.9317,
      "step": 7324
    },
    {
      "epoch": 1.0031498219665844,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008481736810644632,
      "loss": 4.9663,
      "step": 7325
    },
    {
      "epoch": 1.0032867707477404,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008481545930818548,
      "loss": 5.0089,
      "step": 7326
    },
    {
      "epoch": 1.0034237195288962,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008481355017996444,
      "loss": 4.9584,
      "step": 7327
    },
    {
      "epoch": 1.003560668310052,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008481164072179903,
      "loss": 4.9658,
      "step": 7328
    },
    {
      "epoch": 1.003697617091208,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008480973093370507,
      "loss": 4.8442,
      "step": 7329
    },
    {
      "epoch": 1.0038345658723637,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008480782081569836,
      "loss": 4.9498,
      "step": 7330
    },
    {
      "epoch": 1.0039715146535195,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008480591036779477,
      "loss": 4.9158,
      "step": 7331
    },
    {
      "epoch": 1.0041084634346755,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008480399959001011,
      "loss": 4.9812,
      "step": 7332
    },
    {
      "epoch": 1.0042454122158313,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008480208848236023,
      "loss": 4.9534,
      "step": 7333
    },
    {
      "epoch": 1.004382360996987,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008480017704486095,
      "loss": 4.9947,
      "step": 7334
    },
    {
      "epoch": 1.004519309778143,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008479826527752811,
      "loss": 4.931,
      "step": 7335
    },
    {
      "epoch": 1.0046562585592989,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008479635318037758,
      "loss": 5.0226,
      "step": 7336
    },
    {
      "epoch": 1.0047932073404546,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008479444075342518,
      "loss": 4.8095,
      "step": 7337
    },
    {
      "epoch": 1.0049301561216104,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008479252799668676,
      "loss": 5.0115,
      "step": 7338
    },
    {
      "epoch": 1.0050671049027664,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000847906149101782,
      "loss": 4.9787,
      "step": 7339
    },
    {
      "epoch": 1.0052040536839222,
      "grad_norm": 0.375,
      "learning_rate": 0.0008478870149391531,
      "loss": 4.9533,
      "step": 7340
    },
    {
      "epoch": 1.005341002465078,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008478678774791397,
      "loss": 4.9496,
      "step": 7341
    },
    {
      "epoch": 1.005477951246234,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008478487367219005,
      "loss": 5.0106,
      "step": 7342
    },
    {
      "epoch": 1.0056149000273897,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008478295926675941,
      "loss": 4.9113,
      "step": 7343
    },
    {
      "epoch": 1.0057518488085455,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000847810445316379,
      "loss": 5.0082,
      "step": 7344
    },
    {
      "epoch": 1.0058887975897015,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008477912946684138,
      "loss": 4.9393,
      "step": 7345
    },
    {
      "epoch": 1.0060257463708573,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008477721407238575,
      "loss": 4.9765,
      "step": 7346
    },
    {
      "epoch": 1.006162695152013,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008477529834828685,
      "loss": 4.9462,
      "step": 7347
    },
    {
      "epoch": 1.006299643933169,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008477338229456059,
      "loss": 4.9919,
      "step": 7348
    },
    {
      "epoch": 1.0064365927143248,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008477146591122284,
      "loss": 4.9496,
      "step": 7349
    },
    {
      "epoch": 1.0065735414954806,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008476954919828946,
      "loss": 4.9926,
      "step": 7350
    },
    {
      "epoch": 1.0067104902766366,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008476763215577637,
      "loss": 4.9244,
      "step": 7351
    },
    {
      "epoch": 1.0068474390577924,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008476571478369942,
      "loss": 4.9445,
      "step": 7352
    },
    {
      "epoch": 1.0069843878389482,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008476379708207452,
      "loss": 4.9838,
      "step": 7353
    },
    {
      "epoch": 1.0071213366201042,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008476187905091755,
      "loss": 4.9339,
      "step": 7354
    },
    {
      "epoch": 1.00725828540126,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008475996069024443,
      "loss": 4.862,
      "step": 7355
    },
    {
      "epoch": 1.0073952341824157,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008475804200007103,
      "loss": 4.9166,
      "step": 7356
    },
    {
      "epoch": 1.0075321829635717,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008475612298041327,
      "loss": 4.8855,
      "step": 7357
    },
    {
      "epoch": 1.0076691317447275,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008475420363128703,
      "loss": 4.9451,
      "step": 7358
    },
    {
      "epoch": 1.0078060805258833,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008475228395270825,
      "loss": 4.9439,
      "step": 7359
    },
    {
      "epoch": 1.0079430293070393,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008475036394469281,
      "loss": 4.9932,
      "step": 7360
    },
    {
      "epoch": 1.008079978088195,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008474844360725663,
      "loss": 4.9864,
      "step": 7361
    },
    {
      "epoch": 1.0082169268693508,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008474652294041563,
      "loss": 4.9278,
      "step": 7362
    },
    {
      "epoch": 1.0083538756505066,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008474460194418573,
      "loss": 4.906,
      "step": 7363
    },
    {
      "epoch": 1.0084908244316626,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008474268061858284,
      "loss": 4.8892,
      "step": 7364
    },
    {
      "epoch": 1.0086277732128184,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008474075896362288,
      "loss": 4.9737,
      "step": 7365
    },
    {
      "epoch": 1.0087647219939742,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008473883697932178,
      "loss": 4.9803,
      "step": 7366
    },
    {
      "epoch": 1.0089016707751302,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008473691466569546,
      "loss": 4.9661,
      "step": 7367
    },
    {
      "epoch": 1.009038619556286,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008473499202275988,
      "loss": 4.8852,
      "step": 7368
    },
    {
      "epoch": 1.0091755683374417,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008473306905053094,
      "loss": 4.9477,
      "step": 7369
    },
    {
      "epoch": 1.0093125171185977,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000847311457490246,
      "loss": 4.9258,
      "step": 7370
    },
    {
      "epoch": 1.0094494658997535,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008472922211825677,
      "loss": 4.9785,
      "step": 7371
    },
    {
      "epoch": 1.0095864146809093,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008472729815824342,
      "loss": 4.9072,
      "step": 7372
    },
    {
      "epoch": 1.0097233634620653,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000847253738690005,
      "loss": 4.951,
      "step": 7373
    },
    {
      "epoch": 1.009860312243221,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008472344925054392,
      "loss": 4.823,
      "step": 7374
    },
    {
      "epoch": 1.0099972610243768,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008472152430288965,
      "loss": 4.9323,
      "step": 7375
    },
    {
      "epoch": 1.0101342098055328,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008471959902605365,
      "loss": 4.9524,
      "step": 7376
    },
    {
      "epoch": 1.0102711585866886,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008471767342005186,
      "loss": 4.9983,
      "step": 7377
    },
    {
      "epoch": 1.0104081073678444,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008471574748490025,
      "loss": 4.8994,
      "step": 7378
    },
    {
      "epoch": 1.0105450561490004,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008471382122061478,
      "loss": 4.9435,
      "step": 7379
    },
    {
      "epoch": 1.0106820049301561,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008471189462721142,
      "loss": 4.9364,
      "step": 7380
    },
    {
      "epoch": 1.010818953711312,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000847099677047061,
      "loss": 4.8999,
      "step": 7381
    },
    {
      "epoch": 1.010955902492468,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008470804045311484,
      "loss": 4.9724,
      "step": 7382
    },
    {
      "epoch": 1.0110928512736237,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008470611287245357,
      "loss": 4.9662,
      "step": 7383
    },
    {
      "epoch": 1.0112298000547795,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000847041849627383,
      "loss": 4.957,
      "step": 7384
    },
    {
      "epoch": 1.0113667488359355,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008470225672398497,
      "loss": 4.9948,
      "step": 7385
    },
    {
      "epoch": 1.0115036976170912,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008470032815620958,
      "loss": 4.9951,
      "step": 7386
    },
    {
      "epoch": 1.011640646398247,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008469839925942811,
      "loss": 5.0122,
      "step": 7387
    },
    {
      "epoch": 1.0117775951794028,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008469647003365656,
      "loss": 4.868,
      "step": 7388
    },
    {
      "epoch": 1.0119145439605588,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008469454047891089,
      "loss": 4.8279,
      "step": 7389
    },
    {
      "epoch": 1.0120514927417146,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008469261059520711,
      "loss": 4.9644,
      "step": 7390
    },
    {
      "epoch": 1.0121884415228704,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008469068038256121,
      "loss": 4.9802,
      "step": 7391
    },
    {
      "epoch": 1.0123253903040264,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008468874984098919,
      "loss": 5.0288,
      "step": 7392
    },
    {
      "epoch": 1.0124623390851821,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008468681897050703,
      "loss": 4.932,
      "step": 7393
    },
    {
      "epoch": 1.012599287866338,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008468488777113075,
      "loss": 5.0681,
      "step": 7394
    },
    {
      "epoch": 1.012736236647494,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008468295624287634,
      "loss": 4.922,
      "step": 7395
    },
    {
      "epoch": 1.0128731854286497,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008468102438575982,
      "loss": 4.9426,
      "step": 7396
    },
    {
      "epoch": 1.0130101342098055,
      "grad_norm": 0.390625,
      "learning_rate": 0.000846790921997972,
      "loss": 4.9236,
      "step": 7397
    },
    {
      "epoch": 1.0131470829909615,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008467715968500448,
      "loss": 4.8549,
      "step": 7398
    },
    {
      "epoch": 1.0132840317721172,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008467522684139769,
      "loss": 4.9075,
      "step": 7399
    },
    {
      "epoch": 1.013420980553273,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008467329366899284,
      "loss": 5.0302,
      "step": 7400
    },
    {
      "epoch": 1.013557929334429,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008467136016780596,
      "loss": 4.9335,
      "step": 7401
    },
    {
      "epoch": 1.0136948781155848,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008466942633785305,
      "loss": 4.9421,
      "step": 7402
    },
    {
      "epoch": 1.0138318268967406,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008466749217915017,
      "loss": 4.9084,
      "step": 7403
    },
    {
      "epoch": 1.0139687756778966,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008466555769171331,
      "loss": 4.9608,
      "step": 7404
    },
    {
      "epoch": 1.0141057244590523,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008466362287555853,
      "loss": 4.9499,
      "step": 7405
    },
    {
      "epoch": 1.0142426732402081,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008466168773070186,
      "loss": 4.8402,
      "step": 7406
    },
    {
      "epoch": 1.0143796220213641,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008465975225715932,
      "loss": 5.0126,
      "step": 7407
    },
    {
      "epoch": 1.01451657080252,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008465781645494698,
      "loss": 4.9894,
      "step": 7408
    },
    {
      "epoch": 1.0146535195836757,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008465588032408085,
      "loss": 4.8625,
      "step": 7409
    },
    {
      "epoch": 1.0147904683648314,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00084653943864577,
      "loss": 5.0691,
      "step": 7410
    },
    {
      "epoch": 1.0149274171459874,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008465200707645146,
      "loss": 4.9245,
      "step": 7411
    },
    {
      "epoch": 1.0150643659271432,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000846500699597203,
      "loss": 4.8977,
      "step": 7412
    },
    {
      "epoch": 1.015201314708299,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008464813251439956,
      "loss": 4.8903,
      "step": 7413
    },
    {
      "epoch": 1.015338263489455,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008464619474050529,
      "loss": 4.9909,
      "step": 7414
    },
    {
      "epoch": 1.0154752122706108,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008464425663805354,
      "loss": 4.9636,
      "step": 7415
    },
    {
      "epoch": 1.0156121610517665,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008464231820706043,
      "loss": 4.9368,
      "step": 7416
    },
    {
      "epoch": 1.0157491098329225,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008464037944754196,
      "loss": 4.9904,
      "step": 7417
    },
    {
      "epoch": 1.0158860586140783,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008463844035951422,
      "loss": 4.9179,
      "step": 7418
    },
    {
      "epoch": 1.016023007395234,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008463650094299327,
      "loss": 4.9315,
      "step": 7419
    },
    {
      "epoch": 1.01615995617639,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008463456119799521,
      "loss": 4.9788,
      "step": 7420
    },
    {
      "epoch": 1.0162969049575459,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008463262112453608,
      "loss": 4.9471,
      "step": 7421
    },
    {
      "epoch": 1.0164338537387017,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008463068072263199,
      "loss": 4.9276,
      "step": 7422
    },
    {
      "epoch": 1.0165708025198577,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008462873999229902,
      "loss": 4.8743,
      "step": 7423
    },
    {
      "epoch": 1.0167077513010134,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008462679893355321,
      "loss": 4.8967,
      "step": 7424
    },
    {
      "epoch": 1.0168447000821692,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008462485754641068,
      "loss": 4.9379,
      "step": 7425
    },
    {
      "epoch": 1.0169816488633252,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008462291583088752,
      "loss": 4.8366,
      "step": 7426
    },
    {
      "epoch": 1.017118597644481,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008462097378699981,
      "loss": 5.0117,
      "step": 7427
    },
    {
      "epoch": 1.0172555464256368,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008461903141476364,
      "loss": 4.9935,
      "step": 7428
    },
    {
      "epoch": 1.0173924952067928,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008461708871419513,
      "loss": 4.9346,
      "step": 7429
    },
    {
      "epoch": 1.0175294439879485,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008461514568531037,
      "loss": 4.9439,
      "step": 7430
    },
    {
      "epoch": 1.0176663927691043,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008461320232812545,
      "loss": 4.9118,
      "step": 7431
    },
    {
      "epoch": 1.0178033415502603,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008461125864265647,
      "loss": 4.927,
      "step": 7432
    },
    {
      "epoch": 1.017940290331416,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008460931462891958,
      "loss": 4.8027,
      "step": 7433
    },
    {
      "epoch": 1.0180772391125719,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008460737028693085,
      "loss": 4.8792,
      "step": 7434
    },
    {
      "epoch": 1.0182141878937276,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000846054256167064,
      "loss": 4.9774,
      "step": 7435
    },
    {
      "epoch": 1.0183511366748836,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008460348061826234,
      "loss": 4.9393,
      "step": 7436
    },
    {
      "epoch": 1.0184880854560394,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008460153529161481,
      "loss": 5.038,
      "step": 7437
    },
    {
      "epoch": 1.0186250342371952,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008459958963677993,
      "loss": 5.0075,
      "step": 7438
    },
    {
      "epoch": 1.0187619830183512,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008459764365377379,
      "loss": 4.8422,
      "step": 7439
    },
    {
      "epoch": 1.018898931799507,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008459569734261257,
      "loss": 4.9887,
      "step": 7440
    },
    {
      "epoch": 1.0190358805806627,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008459375070331236,
      "loss": 4.9444,
      "step": 7441
    },
    {
      "epoch": 1.0191728293618187,
      "grad_norm": 0.390625,
      "learning_rate": 0.000845918037358893,
      "loss": 4.9523,
      "step": 7442
    },
    {
      "epoch": 1.0193097781429745,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008458985644035952,
      "loss": 4.9398,
      "step": 7443
    },
    {
      "epoch": 1.0194467269241303,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008458790881673917,
      "loss": 5.0121,
      "step": 7444
    },
    {
      "epoch": 1.0195836757052863,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000845859608650444,
      "loss": 4.938,
      "step": 7445
    },
    {
      "epoch": 1.019720624486442,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008458401258529133,
      "loss": 4.9595,
      "step": 7446
    },
    {
      "epoch": 1.0198575732675979,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008458206397749611,
      "loss": 4.9211,
      "step": 7447
    },
    {
      "epoch": 1.0199945220487538,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008458011504167491,
      "loss": 4.9684,
      "step": 7448
    },
    {
      "epoch": 1.0201314708299096,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008457816577784384,
      "loss": 4.9592,
      "step": 7449
    },
    {
      "epoch": 1.0202684196110654,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008457621618601911,
      "loss": 4.9492,
      "step": 7450
    },
    {
      "epoch": 1.0204053683922214,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008457426626621682,
      "loss": 4.9567,
      "step": 7451
    },
    {
      "epoch": 1.0205423171733772,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008457231601845317,
      "loss": 4.9219,
      "step": 7452
    },
    {
      "epoch": 1.020679265954533,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008457036544274431,
      "loss": 4.9532,
      "step": 7453
    },
    {
      "epoch": 1.020816214735689,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008456841453910639,
      "loss": 4.9553,
      "step": 7454
    },
    {
      "epoch": 1.0209531635168447,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008456646330755559,
      "loss": 4.9924,
      "step": 7455
    },
    {
      "epoch": 1.0210901122980005,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008456451174810809,
      "loss": 4.836,
      "step": 7456
    },
    {
      "epoch": 1.0212270610791565,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008456255986078004,
      "loss": 4.978,
      "step": 7457
    },
    {
      "epoch": 1.0213640098603123,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008456060764558764,
      "loss": 4.9129,
      "step": 7458
    },
    {
      "epoch": 1.021500958641468,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008455865510254706,
      "loss": 4.9371,
      "step": 7459
    },
    {
      "epoch": 1.0216379074226238,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008455670223167448,
      "loss": 4.9062,
      "step": 7460
    },
    {
      "epoch": 1.0217748562037798,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008455474903298607,
      "loss": 4.9359,
      "step": 7461
    },
    {
      "epoch": 1.0219118049849356,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008455279550649804,
      "loss": 4.9603,
      "step": 7462
    },
    {
      "epoch": 1.0220487537660914,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008455084165222657,
      "loss": 4.9799,
      "step": 7463
    },
    {
      "epoch": 1.0221857025472474,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008454888747018785,
      "loss": 5.0127,
      "step": 7464
    },
    {
      "epoch": 1.0223226513284032,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008454693296039807,
      "loss": 4.9056,
      "step": 7465
    },
    {
      "epoch": 1.022459600109559,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008454497812287343,
      "loss": 4.9661,
      "step": 7466
    },
    {
      "epoch": 1.022596548890715,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008454302295763013,
      "loss": 4.9752,
      "step": 7467
    },
    {
      "epoch": 1.0227334976718707,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008454106746468439,
      "loss": 4.956,
      "step": 7468
    },
    {
      "epoch": 1.0228704464530265,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008453911164405239,
      "loss": 4.9091,
      "step": 7469
    },
    {
      "epoch": 1.0230073952341825,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008453715549575037,
      "loss": 4.9052,
      "step": 7470
    },
    {
      "epoch": 1.0231443440153383,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008453519901979451,
      "loss": 4.9125,
      "step": 7471
    },
    {
      "epoch": 1.023281292796494,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008453324221620103,
      "loss": 4.9561,
      "step": 7472
    },
    {
      "epoch": 1.02341824157765,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008453128508498616,
      "loss": 4.9057,
      "step": 7473
    },
    {
      "epoch": 1.0235551903588058,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000845293276261661,
      "loss": 4.8591,
      "step": 7474
    },
    {
      "epoch": 1.0236921391399616,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008452736983975708,
      "loss": 4.9528,
      "step": 7475
    },
    {
      "epoch": 1.0238290879211176,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008452541172577533,
      "loss": 4.9604,
      "step": 7476
    },
    {
      "epoch": 1.0239660367022734,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000845234532842371,
      "loss": 5.0036,
      "step": 7477
    },
    {
      "epoch": 1.0241029854834292,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008452149451515855,
      "loss": 4.9323,
      "step": 7478
    },
    {
      "epoch": 1.0242399342645852,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00084519535418556,
      "loss": 4.8762,
      "step": 7479
    },
    {
      "epoch": 1.024376883045741,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008451757599444562,
      "loss": 5.0073,
      "step": 7480
    },
    {
      "epoch": 1.0245138318268967,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008451561624284368,
      "loss": 4.9388,
      "step": 7481
    },
    {
      "epoch": 1.0246507806080525,
      "grad_norm": 0.40625,
      "learning_rate": 0.000845136561637664,
      "loss": 4.9191,
      "step": 7482
    },
    {
      "epoch": 1.0247877293892085,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008451169575723005,
      "loss": 4.9137,
      "step": 7483
    },
    {
      "epoch": 1.0249246781703643,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008450973502325086,
      "loss": 4.9265,
      "step": 7484
    },
    {
      "epoch": 1.02506162695152,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008450777396184508,
      "loss": 4.952,
      "step": 7485
    },
    {
      "epoch": 1.025198575732676,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008450581257302896,
      "loss": 5.0279,
      "step": 7486
    },
    {
      "epoch": 1.0253355245138318,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008450385085681875,
      "loss": 4.9502,
      "step": 7487
    },
    {
      "epoch": 1.0254724732949876,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008450188881323072,
      "loss": 4.8967,
      "step": 7488
    },
    {
      "epoch": 1.0256094220761436,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008449992644228113,
      "loss": 4.9143,
      "step": 7489
    },
    {
      "epoch": 1.0257463708572994,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008449796374398623,
      "loss": 4.9665,
      "step": 7490
    },
    {
      "epoch": 1.0258833196384551,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000844960007183623,
      "loss": 4.9646,
      "step": 7491
    },
    {
      "epoch": 1.0260202684196111,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000844940373654256,
      "loss": 4.8852,
      "step": 7492
    },
    {
      "epoch": 1.026157217200767,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000844920736851924,
      "loss": 4.9471,
      "step": 7493
    },
    {
      "epoch": 1.0262941659819227,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008449010967767897,
      "loss": 4.9498,
      "step": 7494
    },
    {
      "epoch": 1.0264311147630787,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008448814534290159,
      "loss": 4.9121,
      "step": 7495
    },
    {
      "epoch": 1.0265680635442345,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008448618068087654,
      "loss": 5.0143,
      "step": 7496
    },
    {
      "epoch": 1.0267050123253902,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008448421569162011,
      "loss": 5.0275,
      "step": 7497
    },
    {
      "epoch": 1.0268419611065462,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008448225037514857,
      "loss": 4.859,
      "step": 7498
    },
    {
      "epoch": 1.026978909887702,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008448028473147822,
      "loss": 5.0037,
      "step": 7499
    },
    {
      "epoch": 1.0271158586688578,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008447831876062534,
      "loss": 4.9367,
      "step": 7500
    },
    {
      "epoch": 1.0272528074500138,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008447635246260623,
      "loss": 4.9407,
      "step": 7501
    },
    {
      "epoch": 1.0273897562311696,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008447438583743717,
      "loss": 4.9727,
      "step": 7502
    },
    {
      "epoch": 1.0275267050123253,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008447241888513448,
      "loss": 4.969,
      "step": 7503
    },
    {
      "epoch": 1.0276636537934813,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008447045160571444,
      "loss": 4.9169,
      "step": 7504
    },
    {
      "epoch": 1.0278006025746371,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008446848399919336,
      "loss": 4.9074,
      "step": 7505
    },
    {
      "epoch": 1.027937551355793,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008446651606558756,
      "loss": 4.9569,
      "step": 7506
    },
    {
      "epoch": 1.0280745001369487,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008446454780491333,
      "loss": 4.9753,
      "step": 7507
    },
    {
      "epoch": 1.0282114489181047,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008446257921718698,
      "loss": 4.9134,
      "step": 7508
    },
    {
      "epoch": 1.0283483976992605,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008446061030242484,
      "loss": 4.9222,
      "step": 7509
    },
    {
      "epoch": 1.0284853464804162,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008445864106064322,
      "loss": 4.9522,
      "step": 7510
    },
    {
      "epoch": 1.0286222952615722,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008445667149185844,
      "loss": 4.9387,
      "step": 7511
    },
    {
      "epoch": 1.028759244042728,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008445470159608682,
      "loss": 4.9848,
      "step": 7512
    },
    {
      "epoch": 1.0288961928238838,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008445273137334469,
      "loss": 5.0028,
      "step": 7513
    },
    {
      "epoch": 1.0290331416050398,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008445076082364836,
      "loss": 4.9065,
      "step": 7514
    },
    {
      "epoch": 1.0291700903861956,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008444878994701419,
      "loss": 4.974,
      "step": 7515
    },
    {
      "epoch": 1.0293070391673513,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008444681874345849,
      "loss": 4.996,
      "step": 7516
    },
    {
      "epoch": 1.0294439879485073,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008444484721299761,
      "loss": 4.9575,
      "step": 7517
    },
    {
      "epoch": 1.029580936729663,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008444287535564788,
      "loss": 4.8914,
      "step": 7518
    },
    {
      "epoch": 1.0297178855108189,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008444090317142563,
      "loss": 4.8518,
      "step": 7519
    },
    {
      "epoch": 1.0298548342919749,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008443893066034724,
      "loss": 4.8842,
      "step": 7520
    },
    {
      "epoch": 1.0299917830731307,
      "grad_norm": 0.359375,
      "learning_rate": 0.00084436957822429,
      "loss": 4.943,
      "step": 7521
    },
    {
      "epoch": 1.0301287318542864,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008443498465768733,
      "loss": 5.0118,
      "step": 7522
    },
    {
      "epoch": 1.0302656806354424,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008443301116613853,
      "loss": 4.8569,
      "step": 7523
    },
    {
      "epoch": 1.0304026294165982,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008443103734779898,
      "loss": 4.9226,
      "step": 7524
    },
    {
      "epoch": 1.030539578197754,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008442906320268502,
      "loss": 4.8814,
      "step": 7525
    },
    {
      "epoch": 1.03067652697891,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00084427088730813,
      "loss": 4.9815,
      "step": 7526
    },
    {
      "epoch": 1.0308134757600658,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008442511393219933,
      "loss": 4.9269,
      "step": 7527
    },
    {
      "epoch": 1.0309504245412215,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008442313880686033,
      "loss": 4.8953,
      "step": 7528
    },
    {
      "epoch": 1.0310873733223773,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008442116335481239,
      "loss": 4.8583,
      "step": 7529
    },
    {
      "epoch": 1.0312243221035333,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008441918757607186,
      "loss": 4.9619,
      "step": 7530
    },
    {
      "epoch": 1.031361270884689,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008441721147065515,
      "loss": 4.8916,
      "step": 7531
    },
    {
      "epoch": 1.0314982196658449,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008441523503857861,
      "loss": 4.9996,
      "step": 7532
    },
    {
      "epoch": 1.0316351684470009,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008441325827985862,
      "loss": 4.9138,
      "step": 7533
    },
    {
      "epoch": 1.0317721172281566,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008441128119451157,
      "loss": 4.8669,
      "step": 7534
    },
    {
      "epoch": 1.0319090660093124,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008440930378255386,
      "loss": 4.9831,
      "step": 7535
    },
    {
      "epoch": 1.0320460147904684,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008440732604400183,
      "loss": 4.9402,
      "step": 7536
    },
    {
      "epoch": 1.0321829635716242,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008440534797887192,
      "loss": 4.9585,
      "step": 7537
    },
    {
      "epoch": 1.03231991235278,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008440336958718048,
      "loss": 4.8801,
      "step": 7538
    },
    {
      "epoch": 1.032456861133936,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008440139086894394,
      "loss": 4.972,
      "step": 7539
    },
    {
      "epoch": 1.0325938099150918,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008439941182417867,
      "loss": 4.9623,
      "step": 7540
    },
    {
      "epoch": 1.0327307586962475,
      "grad_norm": 0.5,
      "learning_rate": 0.0008439743245290111,
      "loss": 4.8801,
      "step": 7541
    },
    {
      "epoch": 1.0328677074774035,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008439545275512763,
      "loss": 4.8843,
      "step": 7542
    },
    {
      "epoch": 1.0330046562585593,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008439347273087464,
      "loss": 4.9217,
      "step": 7543
    },
    {
      "epoch": 1.033141605039715,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008439149238015857,
      "loss": 4.9727,
      "step": 7544
    },
    {
      "epoch": 1.033278553820871,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008438951170299581,
      "loss": 5.0185,
      "step": 7545
    },
    {
      "epoch": 1.0334155026020269,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008438753069940276,
      "loss": 5.0252,
      "step": 7546
    },
    {
      "epoch": 1.0335524513831826,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008438554936939588,
      "loss": 4.9965,
      "step": 7547
    },
    {
      "epoch": 1.0336894001643386,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008438356771299155,
      "loss": 5.0026,
      "step": 7548
    },
    {
      "epoch": 1.0338263489454944,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008438158573020623,
      "loss": 4.9451,
      "step": 7549
    },
    {
      "epoch": 1.0339632977266502,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000843796034210563,
      "loss": 4.9379,
      "step": 7550
    },
    {
      "epoch": 1.0341002465078062,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008437762078555824,
      "loss": 4.961,
      "step": 7551
    },
    {
      "epoch": 1.034237195288962,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008437563782372845,
      "loss": 4.9324,
      "step": 7552
    },
    {
      "epoch": 1.0343741440701177,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008437365453558335,
      "loss": 4.9524,
      "step": 7553
    },
    {
      "epoch": 1.0345110928512735,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008437167092113939,
      "loss": 4.879,
      "step": 7554
    },
    {
      "epoch": 1.0346480416324295,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008436968698041303,
      "loss": 4.9845,
      "step": 7555
    },
    {
      "epoch": 1.0347849904135853,
      "grad_norm": 0.375,
      "learning_rate": 0.0008436770271342069,
      "loss": 4.9771,
      "step": 7556
    },
    {
      "epoch": 1.034921939194741,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0008436571812017881,
      "loss": 4.9755,
      "step": 7557
    },
    {
      "epoch": 1.035058887975897,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008436373320070385,
      "loss": 4.909,
      "step": 7558
    },
    {
      "epoch": 1.0351958367570528,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008436174795501224,
      "loss": 4.9441,
      "step": 7559
    },
    {
      "epoch": 1.0353327855382086,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008435976238312044,
      "loss": 4.9057,
      "step": 7560
    },
    {
      "epoch": 1.0354697343193646,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008435777648504494,
      "loss": 4.9648,
      "step": 7561
    },
    {
      "epoch": 1.0356066831005204,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008435579026080214,
      "loss": 4.8775,
      "step": 7562
    },
    {
      "epoch": 1.0357436318816762,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008435380371040854,
      "loss": 4.9389,
      "step": 7563
    },
    {
      "epoch": 1.0358805806628322,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008435181683388059,
      "loss": 4.9203,
      "step": 7564
    },
    {
      "epoch": 1.036017529443988,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008434982963123476,
      "loss": 4.926,
      "step": 7565
    },
    {
      "epoch": 1.0361544782251437,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008434784210248752,
      "loss": 4.971,
      "step": 7566
    },
    {
      "epoch": 1.0362914270062997,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008434585424765533,
      "loss": 4.9223,
      "step": 7567
    },
    {
      "epoch": 1.0364283757874555,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008434386606675467,
      "loss": 4.9602,
      "step": 7568
    },
    {
      "epoch": 1.0365653245686113,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008434187755980202,
      "loss": 4.9011,
      "step": 7569
    },
    {
      "epoch": 1.0367022733497673,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008433988872681384,
      "loss": 4.8968,
      "step": 7570
    },
    {
      "epoch": 1.036839222130923,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008433789956780666,
      "loss": 4.9226,
      "step": 7571
    },
    {
      "epoch": 1.0369761709120788,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000843359100827969,
      "loss": 4.8964,
      "step": 7572
    },
    {
      "epoch": 1.0371131196932348,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008433392027180109,
      "loss": 4.8465,
      "step": 7573
    },
    {
      "epoch": 1.0372500684743906,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008433193013483572,
      "loss": 4.9216,
      "step": 7574
    },
    {
      "epoch": 1.0373870172555464,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008432993967191726,
      "loss": 4.8619,
      "step": 7575
    },
    {
      "epoch": 1.0375239660367024,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008432794888306223,
      "loss": 4.8832,
      "step": 7576
    },
    {
      "epoch": 1.0376609148178582,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008432595776828711,
      "loss": 4.8379,
      "step": 7577
    },
    {
      "epoch": 1.037797863599014,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000843239663276084,
      "loss": 4.9938,
      "step": 7578
    },
    {
      "epoch": 1.0379348123801697,
      "grad_norm": 0.375,
      "learning_rate": 0.0008432197456104262,
      "loss": 4.8889,
      "step": 7579
    },
    {
      "epoch": 1.0380717611613257,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008431998246860626,
      "loss": 4.9707,
      "step": 7580
    },
    {
      "epoch": 1.0382087099424815,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008431799005031585,
      "loss": 4.8905,
      "step": 7581
    },
    {
      "epoch": 1.0383456587236373,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008431599730618788,
      "loss": 4.9204,
      "step": 7582
    },
    {
      "epoch": 1.0384826075047933,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008431400423623886,
      "loss": 5.0127,
      "step": 7583
    },
    {
      "epoch": 1.038619556285949,
      "grad_norm": 0.375,
      "learning_rate": 0.0008431201084048534,
      "loss": 4.956,
      "step": 7584
    },
    {
      "epoch": 1.0387565050671048,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008431001711894379,
      "loss": 4.9669,
      "step": 7585
    },
    {
      "epoch": 1.0388934538482608,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008430802307163079,
      "loss": 4.9794,
      "step": 7586
    },
    {
      "epoch": 1.0390304026294166,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008430602869856282,
      "loss": 4.9437,
      "step": 7587
    },
    {
      "epoch": 1.0391673514105724,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008430403399975643,
      "loss": 4.9813,
      "step": 7588
    },
    {
      "epoch": 1.0393043001917284,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008430203897522814,
      "loss": 4.9975,
      "step": 7589
    },
    {
      "epoch": 1.0394412489728841,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008430004362499449,
      "loss": 5.0047,
      "step": 7590
    },
    {
      "epoch": 1.03957819775404,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008429804794907201,
      "loss": 4.8514,
      "step": 7591
    },
    {
      "epoch": 1.039715146535196,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008429605194747726,
      "loss": 4.9462,
      "step": 7592
    },
    {
      "epoch": 1.0398520953163517,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008429405562022675,
      "loss": 5.0146,
      "step": 7593
    },
    {
      "epoch": 1.0399890440975075,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008429205896733703,
      "loss": 4.8984,
      "step": 7594
    },
    {
      "epoch": 1.0401259928786635,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008429006198882467,
      "loss": 5.0055,
      "step": 7595
    },
    {
      "epoch": 1.0402629416598193,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000842880646847062,
      "loss": 4.9676,
      "step": 7596
    },
    {
      "epoch": 1.040399890440975,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008428606705499817,
      "loss": 4.9309,
      "step": 7597
    },
    {
      "epoch": 1.040536839222131,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008428406909971713,
      "loss": 4.9307,
      "step": 7598
    },
    {
      "epoch": 1.0406737880032868,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008428207081887967,
      "loss": 4.9748,
      "step": 7599
    },
    {
      "epoch": 1.0408107367844426,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008428007221250232,
      "loss": 4.8979,
      "step": 7600
    },
    {
      "epoch": 1.0409476855655986,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008427807328060166,
      "loss": 4.9233,
      "step": 7601
    },
    {
      "epoch": 1.0410846343467544,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008427607402319424,
      "loss": 4.9688,
      "step": 7602
    },
    {
      "epoch": 1.0412215831279101,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008427407444029664,
      "loss": 5.0065,
      "step": 7603
    },
    {
      "epoch": 1.041358531909066,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008427207453192542,
      "loss": 4.9328,
      "step": 7604
    },
    {
      "epoch": 1.041495480690222,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008427007429809716,
      "loss": 4.9725,
      "step": 7605
    },
    {
      "epoch": 1.0416324294713777,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008426807373882844,
      "loss": 4.8898,
      "step": 7606
    },
    {
      "epoch": 1.0417693782525335,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008426607285413583,
      "loss": 4.9162,
      "step": 7607
    },
    {
      "epoch": 1.0419063270336895,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008426407164403594,
      "loss": 5.0281,
      "step": 7608
    },
    {
      "epoch": 1.0420432758148452,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000842620701085453,
      "loss": 4.9657,
      "step": 7609
    },
    {
      "epoch": 1.042180224596001,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008426006824768055,
      "loss": 4.9071,
      "step": 7610
    },
    {
      "epoch": 1.042317173377157,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008425806606145826,
      "loss": 4.8992,
      "step": 7611
    },
    {
      "epoch": 1.0424541221583128,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00084256063549895,
      "loss": 4.8625,
      "step": 7612
    },
    {
      "epoch": 1.0425910709394686,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008425406071300741,
      "loss": 4.9412,
      "step": 7613
    },
    {
      "epoch": 1.0427280197206246,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008425205755081206,
      "loss": 4.9143,
      "step": 7614
    },
    {
      "epoch": 1.0428649685017803,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008425005406332555,
      "loss": 4.9388,
      "step": 7615
    },
    {
      "epoch": 1.0430019172829361,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008424805025056449,
      "loss": 4.9486,
      "step": 7616
    },
    {
      "epoch": 1.0431388660640921,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008424604611254549,
      "loss": 4.9584,
      "step": 7617
    },
    {
      "epoch": 1.043275814845248,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008424404164928516,
      "loss": 4.8788,
      "step": 7618
    },
    {
      "epoch": 1.0434127636264037,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000842420368608001,
      "loss": 4.9088,
      "step": 7619
    },
    {
      "epoch": 1.0435497124075597,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008424003174710692,
      "loss": 4.8797,
      "step": 7620
    },
    {
      "epoch": 1.0436866611887154,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008423802630822226,
      "loss": 4.9376,
      "step": 7621
    },
    {
      "epoch": 1.0438236099698712,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008423602054416271,
      "loss": 4.9324,
      "step": 7622
    },
    {
      "epoch": 1.0439605587510272,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008423401445494492,
      "loss": 4.9964,
      "step": 7623
    },
    {
      "epoch": 1.044097507532183,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008423200804058551,
      "loss": 4.9802,
      "step": 7624
    },
    {
      "epoch": 1.0442344563133388,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008423000130110109,
      "loss": 4.9365,
      "step": 7625
    },
    {
      "epoch": 1.0443714050944946,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000842279942365083,
      "loss": 4.9113,
      "step": 7626
    },
    {
      "epoch": 1.0445083538756506,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008422598684682377,
      "loss": 4.9888,
      "step": 7627
    },
    {
      "epoch": 1.0446453026568063,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008422397913206416,
      "loss": 4.8453,
      "step": 7628
    },
    {
      "epoch": 1.044782251437962,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008422197109224607,
      "loss": 4.9664,
      "step": 7629
    },
    {
      "epoch": 1.044919200219118,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008421996272738616,
      "loss": 4.8911,
      "step": 7630
    },
    {
      "epoch": 1.0450561490002739,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0008421795403750109,
      "loss": 5.0314,
      "step": 7631
    },
    {
      "epoch": 1.0451930977814297,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008421594502260747,
      "loss": 4.8935,
      "step": 7632
    },
    {
      "epoch": 1.0453300465625857,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008421393568272198,
      "loss": 4.9915,
      "step": 7633
    },
    {
      "epoch": 1.0454669953437414,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008421192601786126,
      "loss": 5.0196,
      "step": 7634
    },
    {
      "epoch": 1.0456039441248972,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008420991602804197,
      "loss": 4.9742,
      "step": 7635
    },
    {
      "epoch": 1.0457408929060532,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008420790571328075,
      "loss": 4.8277,
      "step": 7636
    },
    {
      "epoch": 1.045877841687209,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008420589507359429,
      "loss": 4.8916,
      "step": 7637
    },
    {
      "epoch": 1.0460147904683648,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008420388410899922,
      "loss": 4.9419,
      "step": 7638
    },
    {
      "epoch": 1.0461517392495208,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008420187281951222,
      "loss": 4.9156,
      "step": 7639
    },
    {
      "epoch": 1.0462886880306765,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008419986120514998,
      "loss": 5.0002,
      "step": 7640
    },
    {
      "epoch": 1.0464256368118323,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008419784926592912,
      "loss": 4.9852,
      "step": 7641
    },
    {
      "epoch": 1.0465625855929883,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008419583700186636,
      "loss": 4.9409,
      "step": 7642
    },
    {
      "epoch": 1.046699534374144,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008419382441297835,
      "loss": 4.957,
      "step": 7643
    },
    {
      "epoch": 1.0468364831552999,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008419181149928179,
      "loss": 4.9951,
      "step": 7644
    },
    {
      "epoch": 1.0469734319364559,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008418979826079335,
      "loss": 4.9714,
      "step": 7645
    },
    {
      "epoch": 1.0471103807176116,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000841877846975297,
      "loss": 5.0174,
      "step": 7646
    },
    {
      "epoch": 1.0472473294987674,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008418577080950753,
      "loss": 4.9522,
      "step": 7647
    },
    {
      "epoch": 1.0473842782799232,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008418375659674357,
      "loss": 4.8671,
      "step": 7648
    },
    {
      "epoch": 1.0475212270610792,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008418174205925446,
      "loss": 4.93,
      "step": 7649
    },
    {
      "epoch": 1.047658175842235,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008417972719705693,
      "loss": 4.9712,
      "step": 7650
    },
    {
      "epoch": 1.0477951246233908,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008417771201016767,
      "loss": 4.9606,
      "step": 7651
    },
    {
      "epoch": 1.0479320734045467,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008417569649860335,
      "loss": 4.8915,
      "step": 7652
    },
    {
      "epoch": 1.0480690221857025,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008417368066238071,
      "loss": 4.8519,
      "step": 7653
    },
    {
      "epoch": 1.0482059709668583,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008417166450151645,
      "loss": 4.9301,
      "step": 7654
    },
    {
      "epoch": 1.0483429197480143,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008416964801602726,
      "loss": 4.9697,
      "step": 7655
    },
    {
      "epoch": 1.04847986852917,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008416763120592987,
      "loss": 4.8956,
      "step": 7656
    },
    {
      "epoch": 1.0486168173103259,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00084165614071241,
      "loss": 4.95,
      "step": 7657
    },
    {
      "epoch": 1.0487537660914819,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008416359661197733,
      "loss": 4.9126,
      "step": 7658
    },
    {
      "epoch": 1.0488907148726376,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008416157882815562,
      "loss": 4.9722,
      "step": 7659
    },
    {
      "epoch": 1.0490276636537934,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008415956071979258,
      "loss": 4.8102,
      "step": 7660
    },
    {
      "epoch": 1.0491646124349494,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008415754228690491,
      "loss": 4.9615,
      "step": 7661
    },
    {
      "epoch": 1.0493015612161052,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008415552352950936,
      "loss": 4.909,
      "step": 7662
    },
    {
      "epoch": 1.049438509997261,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008415350444762268,
      "loss": 4.9636,
      "step": 7663
    },
    {
      "epoch": 1.049575458778417,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008415148504126155,
      "loss": 4.973,
      "step": 7664
    },
    {
      "epoch": 1.0497124075595727,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008414946531044274,
      "loss": 5.0207,
      "step": 7665
    },
    {
      "epoch": 1.0498493563407285,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008414744525518298,
      "loss": 4.9465,
      "step": 7666
    },
    {
      "epoch": 1.0499863051218845,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008414542487549903,
      "loss": 4.9578,
      "step": 7667
    },
    {
      "epoch": 1.0501232539030403,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008414340417140759,
      "loss": 4.854,
      "step": 7668
    },
    {
      "epoch": 1.050260202684196,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008414138314292547,
      "loss": 4.9013,
      "step": 7669
    },
    {
      "epoch": 1.050397151465352,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008413936179006936,
      "loss": 4.9765,
      "step": 7670
    },
    {
      "epoch": 1.0505341002465078,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008413734011285601,
      "loss": 4.9126,
      "step": 7671
    },
    {
      "epoch": 1.0506710490276636,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008413531811130222,
      "loss": 4.947,
      "step": 7672
    },
    {
      "epoch": 1.0508079978088194,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008413329578542473,
      "loss": 5.001,
      "step": 7673
    },
    {
      "epoch": 1.0509449465899754,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008413127313524029,
      "loss": 4.9277,
      "step": 7674
    },
    {
      "epoch": 1.0510818953711312,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008412925016076564,
      "loss": 4.8929,
      "step": 7675
    },
    {
      "epoch": 1.051218844152287,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000841272268620176,
      "loss": 4.8853,
      "step": 7676
    },
    {
      "epoch": 1.051355792933443,
      "grad_norm": 0.390625,
      "learning_rate": 0.000841252032390129,
      "loss": 4.8959,
      "step": 7677
    },
    {
      "epoch": 1.0514927417145987,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008412317929176831,
      "loss": 4.9648,
      "step": 7678
    },
    {
      "epoch": 1.0516296904957545,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008412115502030062,
      "loss": 4.9565,
      "step": 7679
    },
    {
      "epoch": 1.0517666392769105,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008411913042462659,
      "loss": 4.8993,
      "step": 7680
    },
    {
      "epoch": 1.0519035880580663,
      "grad_norm": 0.359375,
      "learning_rate": 0.00084117105504763,
      "loss": 4.8832,
      "step": 7681
    },
    {
      "epoch": 1.052040536839222,
      "grad_norm": 0.375,
      "learning_rate": 0.0008411508026072664,
      "loss": 4.9599,
      "step": 7682
    },
    {
      "epoch": 1.052177485620378,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008411305469253431,
      "loss": 4.8646,
      "step": 7683
    },
    {
      "epoch": 1.0523144344015338,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008411102880020274,
      "loss": 4.8941,
      "step": 7684
    },
    {
      "epoch": 1.0524513831826896,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008410900258374877,
      "loss": 5.0337,
      "step": 7685
    },
    {
      "epoch": 1.0525883319638456,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008410697604318919,
      "loss": 4.8804,
      "step": 7686
    },
    {
      "epoch": 1.0527252807450014,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008410494917854076,
      "loss": 4.9516,
      "step": 7687
    },
    {
      "epoch": 1.0528622295261572,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000841029219898203,
      "loss": 5.0182,
      "step": 7688
    },
    {
      "epoch": 1.0529991783073132,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008410089447704462,
      "loss": 4.8903,
      "step": 7689
    },
    {
      "epoch": 1.053136127088469,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000840988666402305,
      "loss": 5.017,
      "step": 7690
    },
    {
      "epoch": 1.0532730758696247,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008409683847939477,
      "loss": 4.9562,
      "step": 7691
    },
    {
      "epoch": 1.0534100246507807,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008409480999455421,
      "loss": 4.9327,
      "step": 7692
    },
    {
      "epoch": 1.0535469734319365,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008409278118572564,
      "loss": 4.9799,
      "step": 7693
    },
    {
      "epoch": 1.0536839222130923,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008409075205292588,
      "loss": 4.8311,
      "step": 7694
    },
    {
      "epoch": 1.0538208709942483,
      "grad_norm": 0.375,
      "learning_rate": 0.0008408872259617176,
      "loss": 4.9166,
      "step": 7695
    },
    {
      "epoch": 1.053957819775404,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008408669281548006,
      "loss": 4.9981,
      "step": 7696
    },
    {
      "epoch": 1.0540947685565598,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008408466271086763,
      "loss": 4.966,
      "step": 7697
    },
    {
      "epoch": 1.0542317173377156,
      "grad_norm": 0.421875,
      "learning_rate": 0.000840826322823513,
      "loss": 4.8917,
      "step": 7698
    },
    {
      "epoch": 1.0543686661188716,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008408060152994785,
      "loss": 4.914,
      "step": 7699
    },
    {
      "epoch": 1.0545056149000274,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008407857045367417,
      "loss": 4.882,
      "step": 7700
    },
    {
      "epoch": 1.0546425636811831,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008407653905354705,
      "loss": 4.8946,
      "step": 7701
    },
    {
      "epoch": 1.0547795124623391,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008407450732958335,
      "loss": 4.9419,
      "step": 7702
    },
    {
      "epoch": 1.054916461243495,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008407247528179991,
      "loss": 4.8611,
      "step": 7703
    },
    {
      "epoch": 1.0550534100246507,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008407044291021353,
      "loss": 4.929,
      "step": 7704
    },
    {
      "epoch": 1.0551903588058067,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000840684102148411,
      "loss": 4.9696,
      "step": 7705
    },
    {
      "epoch": 1.0553273075869625,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008406637719569943,
      "loss": 4.9259,
      "step": 7706
    },
    {
      "epoch": 1.0554642563681182,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008406434385280539,
      "loss": 4.8352,
      "step": 7707
    },
    {
      "epoch": 1.0556012051492742,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008406231018617582,
      "loss": 4.9359,
      "step": 7708
    },
    {
      "epoch": 1.05573815393043,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008406027619582759,
      "loss": 4.9083,
      "step": 7709
    },
    {
      "epoch": 1.0558751027115858,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008405824188177753,
      "loss": 4.8913,
      "step": 7710
    },
    {
      "epoch": 1.0560120514927418,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008405620724404252,
      "loss": 4.9706,
      "step": 7711
    },
    {
      "epoch": 1.0561490002738976,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008405417228263941,
      "loss": 4.9548,
      "step": 7712
    },
    {
      "epoch": 1.0562859490550534,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0008405213699758505,
      "loss": 4.9776,
      "step": 7713
    },
    {
      "epoch": 1.0564228978362094,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008405010138889636,
      "loss": 4.946,
      "step": 7714
    },
    {
      "epoch": 1.0565598466173651,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008404806545659016,
      "loss": 4.896,
      "step": 7715
    },
    {
      "epoch": 1.056696795398521,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008404602920068333,
      "loss": 4.8598,
      "step": 7716
    },
    {
      "epoch": 1.056833744179677,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008404399262119276,
      "loss": 4.9126,
      "step": 7717
    },
    {
      "epoch": 1.0569706929608327,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008404195571813532,
      "loss": 4.9397,
      "step": 7718
    },
    {
      "epoch": 1.0571076417419885,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008403991849152788,
      "loss": 4.929,
      "step": 7719
    },
    {
      "epoch": 1.0572445905231445,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008403788094138734,
      "loss": 4.9229,
      "step": 7720
    },
    {
      "epoch": 1.0573815393043002,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008403584306773058,
      "loss": 4.9815,
      "step": 7721
    },
    {
      "epoch": 1.057518488085456,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008403380487057449,
      "loss": 4.8935,
      "step": 7722
    },
    {
      "epoch": 1.0576554368666118,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008403176634993596,
      "loss": 4.9389,
      "step": 7723
    },
    {
      "epoch": 1.0577923856477678,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008402972750583186,
      "loss": 4.8989,
      "step": 7724
    },
    {
      "epoch": 1.0579293344289236,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008402768833827912,
      "loss": 4.9977,
      "step": 7725
    },
    {
      "epoch": 1.0580662832100793,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008402564884729464,
      "loss": 4.8913,
      "step": 7726
    },
    {
      "epoch": 1.0582032319912353,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008402360903289528,
      "loss": 4.9479,
      "step": 7727
    },
    {
      "epoch": 1.0583401807723911,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00084021568895098,
      "loss": 4.8929,
      "step": 7728
    },
    {
      "epoch": 1.058477129553547,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008401952843391967,
      "loss": 4.9331,
      "step": 7729
    },
    {
      "epoch": 1.058614078334703,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000840174876493772,
      "loss": 4.9405,
      "step": 7730
    },
    {
      "epoch": 1.0587510271158587,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008401544654148753,
      "loss": 4.8672,
      "step": 7731
    },
    {
      "epoch": 1.0588879758970144,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008401340511026756,
      "loss": 4.9785,
      "step": 7732
    },
    {
      "epoch": 1.0590249246781704,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008401136335573419,
      "loss": 4.8451,
      "step": 7733
    },
    {
      "epoch": 1.0591618734593262,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008400932127790436,
      "loss": 4.8524,
      "step": 7734
    },
    {
      "epoch": 1.059298822240482,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008400727887679499,
      "loss": 5.0161,
      "step": 7735
    },
    {
      "epoch": 1.059435771021638,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008400523615242301,
      "loss": 4.9286,
      "step": 7736
    },
    {
      "epoch": 1.0595727198027938,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008400319310480534,
      "loss": 5.0262,
      "step": 7737
    },
    {
      "epoch": 1.0597096685839495,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008400114973395892,
      "loss": 4.8968,
      "step": 7738
    },
    {
      "epoch": 1.0598466173651055,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008399910603990068,
      "loss": 4.9276,
      "step": 7739
    },
    {
      "epoch": 1.0599835661462613,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008399706202264753,
      "loss": 4.8619,
      "step": 7740
    },
    {
      "epoch": 1.060120514927417,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008399501768221646,
      "loss": 4.8757,
      "step": 7741
    },
    {
      "epoch": 1.060257463708573,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008399297301862438,
      "loss": 4.9956,
      "step": 7742
    },
    {
      "epoch": 1.0603944124897289,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008399092803188824,
      "loss": 4.9485,
      "step": 7743
    },
    {
      "epoch": 1.0605313612708847,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00083988882722025,
      "loss": 4.9225,
      "step": 7744
    },
    {
      "epoch": 1.0606683100520404,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008398683708905158,
      "loss": 4.987,
      "step": 7745
    },
    {
      "epoch": 1.0608052588331964,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008398479113298496,
      "loss": 4.9074,
      "step": 7746
    },
    {
      "epoch": 1.0609422076143522,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008398274485384208,
      "loss": 4.939,
      "step": 7747
    },
    {
      "epoch": 1.061079156395508,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008398069825163992,
      "loss": 4.894,
      "step": 7748
    },
    {
      "epoch": 1.061216105176664,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000839786513263954,
      "loss": 5.0068,
      "step": 7749
    },
    {
      "epoch": 1.0613530539578198,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008397660407812553,
      "loss": 4.8367,
      "step": 7750
    },
    {
      "epoch": 1.0614900027389755,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008397455650684724,
      "loss": 5.0139,
      "step": 7751
    },
    {
      "epoch": 1.0616269515201315,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008397250861257753,
      "loss": 4.8698,
      "step": 7752
    },
    {
      "epoch": 1.0617639003012873,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008397046039533333,
      "loss": 5.0266,
      "step": 7753
    },
    {
      "epoch": 1.061900849082443,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008396841185513167,
      "loss": 4.9786,
      "step": 7754
    },
    {
      "epoch": 1.062037797863599,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008396636299198946,
      "loss": 4.9651,
      "step": 7755
    },
    {
      "epoch": 1.0621747466447549,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008396431380592374,
      "loss": 4.899,
      "step": 7756
    },
    {
      "epoch": 1.0623116954259106,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008396226429695146,
      "loss": 4.9188,
      "step": 7757
    },
    {
      "epoch": 1.0624486442070666,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008396021446508961,
      "loss": 4.9848,
      "step": 7758
    },
    {
      "epoch": 1.0625855929882224,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008395816431035519,
      "loss": 4.9978,
      "step": 7759
    },
    {
      "epoch": 1.0627225417693782,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008395611383276517,
      "loss": 4.908,
      "step": 7760
    },
    {
      "epoch": 1.0628594905505342,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008395406303233655,
      "loss": 5.0205,
      "step": 7761
    },
    {
      "epoch": 1.06299643933169,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008395201190908632,
      "loss": 4.8844,
      "step": 7762
    },
    {
      "epoch": 1.0631333881128457,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000839499604630315,
      "loss": 4.9503,
      "step": 7763
    },
    {
      "epoch": 1.0632703368940017,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008394790869418906,
      "loss": 4.8937,
      "step": 7764
    },
    {
      "epoch": 1.0634072856751575,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008394585660257604,
      "loss": 4.8689,
      "step": 7765
    },
    {
      "epoch": 1.0635442344563133,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008394380418820941,
      "loss": 4.9779,
      "step": 7766
    },
    {
      "epoch": 1.063681183237469,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008394175145110621,
      "loss": 4.9644,
      "step": 7767
    },
    {
      "epoch": 1.063818132018625,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008393969839128342,
      "loss": 4.9128,
      "step": 7768
    },
    {
      "epoch": 1.0639550807997809,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008393764500875809,
      "loss": 4.9406,
      "step": 7769
    },
    {
      "epoch": 1.0640920295809366,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000839355913035472,
      "loss": 4.8749,
      "step": 7770
    },
    {
      "epoch": 1.0642289783620926,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008393353727566779,
      "loss": 4.955,
      "step": 7771
    },
    {
      "epoch": 1.0643659271432484,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008393148292513688,
      "loss": 4.9038,
      "step": 7772
    },
    {
      "epoch": 1.0645028759244042,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000839294282519715,
      "loss": 4.9461,
      "step": 7773
    },
    {
      "epoch": 1.0646398247055602,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008392737325618867,
      "loss": 4.9312,
      "step": 7774
    },
    {
      "epoch": 1.064776773486716,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008392531793780542,
      "loss": 4.9007,
      "step": 7775
    },
    {
      "epoch": 1.0649137222678717,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008392326229683878,
      "loss": 4.9817,
      "step": 7776
    },
    {
      "epoch": 1.0650506710490277,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000839212063333058,
      "loss": 4.8903,
      "step": 7777
    },
    {
      "epoch": 1.0651876198301835,
      "grad_norm": 0.375,
      "learning_rate": 0.000839191500472235,
      "loss": 4.9353,
      "step": 7778
    },
    {
      "epoch": 1.0653245686113393,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008391709343860894,
      "loss": 4.8406,
      "step": 7779
    },
    {
      "epoch": 1.0654615173924953,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008391503650747915,
      "loss": 4.8085,
      "step": 7780
    },
    {
      "epoch": 1.065598466173651,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008391297925385117,
      "loss": 4.9479,
      "step": 7781
    },
    {
      "epoch": 1.0657354149548068,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008391092167774207,
      "loss": 4.944,
      "step": 7782
    },
    {
      "epoch": 1.0658723637359628,
      "grad_norm": 0.375,
      "learning_rate": 0.0008390886377916889,
      "loss": 5.0214,
      "step": 7783
    },
    {
      "epoch": 1.0660093125171186,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008390680555814867,
      "loss": 4.9777,
      "step": 7784
    },
    {
      "epoch": 1.0661462612982744,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008390474701469849,
      "loss": 4.9184,
      "step": 7785
    },
    {
      "epoch": 1.0662832100794304,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008390268814883541,
      "loss": 4.7938,
      "step": 7786
    },
    {
      "epoch": 1.0664201588605862,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008390062896057646,
      "loss": 4.9055,
      "step": 7787
    },
    {
      "epoch": 1.066557107641742,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008389856944993876,
      "loss": 4.9478,
      "step": 7788
    },
    {
      "epoch": 1.066694056422898,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008389650961693933,
      "loss": 4.8586,
      "step": 7789
    },
    {
      "epoch": 1.0668310052040537,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008389444946159525,
      "loss": 4.9535,
      "step": 7790
    },
    {
      "epoch": 1.0669679539852095,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000838923889839236,
      "loss": 4.9405,
      "step": 7791
    },
    {
      "epoch": 1.0671049027663653,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008389032818394147,
      "loss": 4.9498,
      "step": 7792
    },
    {
      "epoch": 1.0672418515475213,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008388826706166591,
      "loss": 5.0421,
      "step": 7793
    },
    {
      "epoch": 1.067378800328677,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008388620561711401,
      "loss": 4.9073,
      "step": 7794
    },
    {
      "epoch": 1.0675157491098328,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008388414385030287,
      "loss": 4.9135,
      "step": 7795
    },
    {
      "epoch": 1.0676526978909888,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008388208176124956,
      "loss": 4.9337,
      "step": 7796
    },
    {
      "epoch": 1.0677896466721446,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008388001934997118,
      "loss": 4.8893,
      "step": 7797
    },
    {
      "epoch": 1.0679265954533004,
      "grad_norm": 0.375,
      "learning_rate": 0.000838779566164848,
      "loss": 4.9006,
      "step": 7798
    },
    {
      "epoch": 1.0680635442344564,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008387589356080754,
      "loss": 4.9641,
      "step": 7799
    },
    {
      "epoch": 1.0682004930156122,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008387383018295648,
      "loss": 4.9245,
      "step": 7800
    },
    {
      "epoch": 1.068337441796768,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008387176648294873,
      "loss": 4.9454,
      "step": 7801
    },
    {
      "epoch": 1.068474390577924,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008386970246080139,
      "loss": 5.0006,
      "step": 7802
    },
    {
      "epoch": 1.0686113393590797,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008386763811653156,
      "loss": 4.9676,
      "step": 7803
    },
    {
      "epoch": 1.0687482881402355,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008386557345015635,
      "loss": 4.9744,
      "step": 7804
    },
    {
      "epoch": 1.0688852369213915,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008386350846169288,
      "loss": 4.9537,
      "step": 7805
    },
    {
      "epoch": 1.0690221857025473,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008386144315115824,
      "loss": 4.9872,
      "step": 7806
    },
    {
      "epoch": 1.069159134483703,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008385937751856958,
      "loss": 4.8788,
      "step": 7807
    },
    {
      "epoch": 1.069296083264859,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008385731156394399,
      "loss": 4.9744,
      "step": 7808
    },
    {
      "epoch": 1.0694330320460148,
      "grad_norm": 0.390625,
      "learning_rate": 0.000838552452872986,
      "loss": 4.9212,
      "step": 7809
    },
    {
      "epoch": 1.0695699808271706,
      "grad_norm": 0.75,
      "learning_rate": 0.0008385317868865053,
      "loss": 4.8593,
      "step": 7810
    },
    {
      "epoch": 1.0697069296083266,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008385111176801691,
      "loss": 4.9004,
      "step": 7811
    },
    {
      "epoch": 1.0698438783894824,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008384904452541488,
      "loss": 4.9288,
      "step": 7812
    },
    {
      "epoch": 1.0699808271706381,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008384697696086154,
      "loss": 4.9085,
      "step": 7813
    },
    {
      "epoch": 1.0701177759517941,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008384490907437407,
      "loss": 4.9422,
      "step": 7814
    },
    {
      "epoch": 1.07025472473295,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008384284086596956,
      "loss": 4.9794,
      "step": 7815
    },
    {
      "epoch": 1.0703916735141057,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008384077233566519,
      "loss": 4.9384,
      "step": 7816
    },
    {
      "epoch": 1.0705286222952615,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008383870348347807,
      "loss": 4.8685,
      "step": 7817
    },
    {
      "epoch": 1.0706655710764175,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008383663430942536,
      "loss": 4.9256,
      "step": 7818
    },
    {
      "epoch": 1.0708025198575732,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008383456481352422,
      "loss": 4.9462,
      "step": 7819
    },
    {
      "epoch": 1.070939468638729,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008383249499579178,
      "loss": 4.9571,
      "step": 7820
    },
    {
      "epoch": 1.071076417419885,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000838304248562452,
      "loss": 4.9159,
      "step": 7821
    },
    {
      "epoch": 1.0712133662010408,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008382835439490164,
      "loss": 5.0123,
      "step": 7822
    },
    {
      "epoch": 1.0713503149821966,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008382628361177825,
      "loss": 4.9325,
      "step": 7823
    },
    {
      "epoch": 1.0714872637633526,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000838242125068922,
      "loss": 4.9263,
      "step": 7824
    },
    {
      "epoch": 1.0716242125445083,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008382214108026064,
      "loss": 4.8648,
      "step": 7825
    },
    {
      "epoch": 1.0717611613256641,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008382006933190076,
      "loss": 4.9462,
      "step": 7826
    },
    {
      "epoch": 1.0718981101068201,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008381799726182972,
      "loss": 4.9385,
      "step": 7827
    },
    {
      "epoch": 1.072035058887976,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008381592487006467,
      "loss": 4.9453,
      "step": 7828
    },
    {
      "epoch": 1.0721720076691317,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000838138521566228,
      "loss": 4.9638,
      "step": 7829
    },
    {
      "epoch": 1.0723089564502877,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000838117791215213,
      "loss": 4.9722,
      "step": 7830
    },
    {
      "epoch": 1.0724459052314435,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008380970576477733,
      "loss": 4.9562,
      "step": 7831
    },
    {
      "epoch": 1.0725828540125992,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008380763208640808,
      "loss": 4.9478,
      "step": 7832
    },
    {
      "epoch": 1.0727198027937552,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008380555808643075,
      "loss": 4.9319,
      "step": 7833
    },
    {
      "epoch": 1.072856751574911,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008380348376486249,
      "loss": 4.9238,
      "step": 7834
    },
    {
      "epoch": 1.0729937003560668,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008380140912172053,
      "loss": 4.9501,
      "step": 7835
    },
    {
      "epoch": 1.0731306491372228,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008379933415702206,
      "loss": 4.9332,
      "step": 7836
    },
    {
      "epoch": 1.0732675979183786,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008379725887078423,
      "loss": 4.9368,
      "step": 7837
    },
    {
      "epoch": 1.0734045466995343,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008379518326302429,
      "loss": 4.9238,
      "step": 7838
    },
    {
      "epoch": 1.0735414954806903,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008379310733375942,
      "loss": 4.952,
      "step": 7839
    },
    {
      "epoch": 1.073678444261846,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008379103108300683,
      "loss": 4.9572,
      "step": 7840
    },
    {
      "epoch": 1.0738153930430019,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008378895451078371,
      "loss": 4.9459,
      "step": 7841
    },
    {
      "epoch": 1.0739523418241577,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008378687761710729,
      "loss": 4.8822,
      "step": 7842
    },
    {
      "epoch": 1.0740892906053137,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008378480040199477,
      "loss": 4.851,
      "step": 7843
    },
    {
      "epoch": 1.0742262393864694,
      "grad_norm": 0.375,
      "learning_rate": 0.0008378272286546337,
      "loss": 4.9333,
      "step": 7844
    },
    {
      "epoch": 1.0743631881676252,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008378064500753031,
      "loss": 5.0212,
      "step": 7845
    },
    {
      "epoch": 1.0745001369487812,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000837785668282128,
      "loss": 4.8917,
      "step": 7846
    },
    {
      "epoch": 1.074637085729937,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008377648832752805,
      "loss": 4.9185,
      "step": 7847
    },
    {
      "epoch": 1.0747740345110928,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008377440950549331,
      "loss": 4.9035,
      "step": 7848
    },
    {
      "epoch": 1.0749109832922488,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000837723303621258,
      "loss": 4.883,
      "step": 7849
    },
    {
      "epoch": 1.0750479320734045,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008377025089744276,
      "loss": 4.9774,
      "step": 7850
    },
    {
      "epoch": 1.0751848808545603,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000837681711114614,
      "loss": 4.9105,
      "step": 7851
    },
    {
      "epoch": 1.0753218296357163,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008376609100419896,
      "loss": 4.9173,
      "step": 7852
    },
    {
      "epoch": 1.075458778416872,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008376401057567271,
      "loss": 4.833,
      "step": 7853
    },
    {
      "epoch": 1.0755957271980279,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008376192982589985,
      "loss": 4.9599,
      "step": 7854
    },
    {
      "epoch": 1.0757326759791839,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008375984875489764,
      "loss": 4.9924,
      "step": 7855
    },
    {
      "epoch": 1.0758696247603396,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008375776736268334,
      "loss": 4.8879,
      "step": 7856
    },
    {
      "epoch": 1.0760065735414954,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008375568564927419,
      "loss": 4.946,
      "step": 7857
    },
    {
      "epoch": 1.0761435223226514,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008375360361468742,
      "loss": 4.9895,
      "step": 7858
    },
    {
      "epoch": 1.0762804711038072,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000837515212589403,
      "loss": 4.9239,
      "step": 7859
    },
    {
      "epoch": 1.076417419884963,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008374943858205011,
      "loss": 4.8964,
      "step": 7860
    },
    {
      "epoch": 1.0765543686661188,
      "grad_norm": 0.375,
      "learning_rate": 0.0008374735558403408,
      "loss": 4.9186,
      "step": 7861
    },
    {
      "epoch": 1.0766913174472748,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008374527226490949,
      "loss": 4.8917,
      "step": 7862
    },
    {
      "epoch": 1.0768282662284305,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008374318862469358,
      "loss": 4.858,
      "step": 7863
    },
    {
      "epoch": 1.0769652150095865,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008374110466340364,
      "loss": 5.0443,
      "step": 7864
    },
    {
      "epoch": 1.0771021637907423,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008373902038105693,
      "loss": 4.9556,
      "step": 7865
    },
    {
      "epoch": 1.077239112571898,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008373693577767074,
      "loss": 4.8546,
      "step": 7866
    },
    {
      "epoch": 1.0773760613530539,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008373485085326233,
      "loss": 5.0009,
      "step": 7867
    },
    {
      "epoch": 1.0775130101342099,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008373276560784896,
      "loss": 4.9165,
      "step": 7868
    },
    {
      "epoch": 1.0776499589153656,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008373068004144795,
      "loss": 4.9505,
      "step": 7869
    },
    {
      "epoch": 1.0777869076965214,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008372859415407656,
      "loss": 4.9641,
      "step": 7870
    },
    {
      "epoch": 1.0779238564776774,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008372650794575209,
      "loss": 4.9496,
      "step": 7871
    },
    {
      "epoch": 1.0780608052588332,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008372442141649181,
      "loss": 4.9702,
      "step": 7872
    },
    {
      "epoch": 1.078197754039989,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008372233456631303,
      "loss": 4.9179,
      "step": 7873
    },
    {
      "epoch": 1.078334702821145,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008372024739523304,
      "loss": 4.8743,
      "step": 7874
    },
    {
      "epoch": 1.0784716516023007,
      "grad_norm": 0.375,
      "learning_rate": 0.0008371815990326912,
      "loss": 4.9142,
      "step": 7875
    },
    {
      "epoch": 1.0786086003834565,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000837160720904386,
      "loss": 4.9305,
      "step": 7876
    },
    {
      "epoch": 1.0787455491646125,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008371398395675875,
      "loss": 4.8771,
      "step": 7877
    },
    {
      "epoch": 1.0788824979457683,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008371189550224689,
      "loss": 4.9002,
      "step": 7878
    },
    {
      "epoch": 1.079019446726924,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008370980672692033,
      "loss": 4.9115,
      "step": 7879
    },
    {
      "epoch": 1.07915639550808,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008370771763079639,
      "loss": 4.9307,
      "step": 7880
    },
    {
      "epoch": 1.0792933442892358,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008370562821389235,
      "loss": 4.965,
      "step": 7881
    },
    {
      "epoch": 1.0794302930703916,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008370353847622556,
      "loss": 4.9324,
      "step": 7882
    },
    {
      "epoch": 1.0795672418515476,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008370144841781332,
      "loss": 4.9686,
      "step": 7883
    },
    {
      "epoch": 1.0797041906327034,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008369935803867295,
      "loss": 4.8961,
      "step": 7884
    },
    {
      "epoch": 1.0798411394138592,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000836972673388218,
      "loss": 4.9271,
      "step": 7885
    },
    {
      "epoch": 1.079978088195015,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008369517631827715,
      "loss": 4.8815,
      "step": 7886
    },
    {
      "epoch": 1.080115036976171,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008369308497705635,
      "loss": 4.9472,
      "step": 7887
    },
    {
      "epoch": 1.0802519857573267,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008369099331517676,
      "loss": 4.8816,
      "step": 7888
    },
    {
      "epoch": 1.0803889345384825,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008368890133265566,
      "loss": 4.9384,
      "step": 7889
    },
    {
      "epoch": 1.0805258833196385,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008368680902951045,
      "loss": 4.9584,
      "step": 7890
    },
    {
      "epoch": 1.0806628321007943,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008368471640575841,
      "loss": 4.9551,
      "step": 7891
    },
    {
      "epoch": 1.08079978088195,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008368262346141691,
      "loss": 4.9613,
      "step": 7892
    },
    {
      "epoch": 1.080936729663106,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000836805301965033,
      "loss": 4.9215,
      "step": 7893
    },
    {
      "epoch": 1.0810736784442618,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008367843661103491,
      "loss": 4.9365,
      "step": 7894
    },
    {
      "epoch": 1.0812106272254176,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000836763427050291,
      "loss": 4.9207,
      "step": 7895
    },
    {
      "epoch": 1.0813475760065736,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008367424847850322,
      "loss": 4.8894,
      "step": 7896
    },
    {
      "epoch": 1.0814845247877294,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008367215393147463,
      "loss": 4.8633,
      "step": 7897
    },
    {
      "epoch": 1.0816214735688852,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000836700590639607,
      "loss": 4.8477,
      "step": 7898
    },
    {
      "epoch": 1.0817584223500412,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008366796387597876,
      "loss": 4.9795,
      "step": 7899
    },
    {
      "epoch": 1.081895371131197,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008366586836754619,
      "loss": 4.9688,
      "step": 7900
    },
    {
      "epoch": 1.0820323199123527,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008366377253868036,
      "loss": 4.9427,
      "step": 7901
    },
    {
      "epoch": 1.0821692686935087,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008366167638939862,
      "loss": 4.9619,
      "step": 7902
    },
    {
      "epoch": 1.0823062174746645,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008365957991971838,
      "loss": 4.8931,
      "step": 7903
    },
    {
      "epoch": 1.0824431662558203,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008365748312965698,
      "loss": 4.9687,
      "step": 7904
    },
    {
      "epoch": 1.0825801150369763,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000836553860192318,
      "loss": 4.9308,
      "step": 7905
    },
    {
      "epoch": 1.082717063818132,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008365328858846022,
      "loss": 4.9685,
      "step": 7906
    },
    {
      "epoch": 1.0828540125992878,
      "grad_norm": 0.375,
      "learning_rate": 0.0008365119083735965,
      "loss": 4.9551,
      "step": 7907
    },
    {
      "epoch": 1.0829909613804438,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008364909276594743,
      "loss": 4.9494,
      "step": 7908
    },
    {
      "epoch": 1.0831279101615996,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008364699437424097,
      "loss": 4.9096,
      "step": 7909
    },
    {
      "epoch": 1.0832648589427554,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008364489566225767,
      "loss": 4.9397,
      "step": 7910
    },
    {
      "epoch": 1.0834018077239111,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000836427966300149,
      "loss": 4.9271,
      "step": 7911
    },
    {
      "epoch": 1.0835387565050671,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008364069727753007,
      "loss": 4.9394,
      "step": 7912
    },
    {
      "epoch": 1.083675705286223,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008363859760482056,
      "loss": 4.9681,
      "step": 7913
    },
    {
      "epoch": 1.0838126540673787,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008363649761190381,
      "loss": 4.9272,
      "step": 7914
    },
    {
      "epoch": 1.0839496028485347,
      "grad_norm": 0.375,
      "learning_rate": 0.0008363439729879718,
      "loss": 4.9373,
      "step": 7915
    },
    {
      "epoch": 1.0840865516296905,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008363229666551809,
      "loss": 4.9586,
      "step": 7916
    },
    {
      "epoch": 1.0842235004108463,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008363019571208397,
      "loss": 4.9733,
      "step": 7917
    },
    {
      "epoch": 1.0843604491920023,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008362809443851221,
      "loss": 4.9081,
      "step": 7918
    },
    {
      "epoch": 1.084497397973158,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008362599284482022,
      "loss": 4.891,
      "step": 7919
    },
    {
      "epoch": 1.0846343467543138,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008362389093102543,
      "loss": 4.9551,
      "step": 7920
    },
    {
      "epoch": 1.0847712955354698,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008362178869714524,
      "loss": 4.9586,
      "step": 7921
    },
    {
      "epoch": 1.0849082443166256,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000836196861431971,
      "loss": 4.9289,
      "step": 7922
    },
    {
      "epoch": 1.0850451930977814,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000836175832691984,
      "loss": 4.9453,
      "step": 7923
    },
    {
      "epoch": 1.0851821418789374,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008361548007516659,
      "loss": 4.84,
      "step": 7924
    },
    {
      "epoch": 1.0853190906600931,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008361337656111911,
      "loss": 4.969,
      "step": 7925
    },
    {
      "epoch": 1.085456039441249,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008361127272707336,
      "loss": 4.8669,
      "step": 7926
    },
    {
      "epoch": 1.085592988222405,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008360916857304679,
      "loss": 4.897,
      "step": 7927
    },
    {
      "epoch": 1.0857299370035607,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008360706409905686,
      "loss": 4.9683,
      "step": 7928
    },
    {
      "epoch": 1.0858668857847165,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008360495930512099,
      "loss": 4.8274,
      "step": 7929
    },
    {
      "epoch": 1.0860038345658725,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008360285419125661,
      "loss": 4.9145,
      "step": 7930
    },
    {
      "epoch": 1.0861407833470282,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008360074875748117,
      "loss": 4.9627,
      "step": 7931
    },
    {
      "epoch": 1.086277732128184,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008359864300381214,
      "loss": 4.8766,
      "step": 7932
    },
    {
      "epoch": 1.08641468090934,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008359653693026695,
      "loss": 4.978,
      "step": 7933
    },
    {
      "epoch": 1.0865516296904958,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008359443053686308,
      "loss": 4.9034,
      "step": 7934
    },
    {
      "epoch": 1.0866885784716516,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008359232382361795,
      "loss": 4.9419,
      "step": 7935
    },
    {
      "epoch": 1.0868255272528073,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008359021679054904,
      "loss": 4.9319,
      "step": 7936
    },
    {
      "epoch": 1.0869624760339633,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008358810943767382,
      "loss": 4.8306,
      "step": 7937
    },
    {
      "epoch": 1.0870994248151191,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008358600176500972,
      "loss": 4.9211,
      "step": 7938
    },
    {
      "epoch": 1.087236373596275,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008358389377257425,
      "loss": 4.9273,
      "step": 7939
    },
    {
      "epoch": 1.087373322377431,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008358178546038485,
      "loss": 4.8631,
      "step": 7940
    },
    {
      "epoch": 1.0875102711585867,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00083579676828459,
      "loss": 4.9383,
      "step": 7941
    },
    {
      "epoch": 1.0876472199397424,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008357756787681418,
      "loss": 4.8755,
      "step": 7942
    },
    {
      "epoch": 1.0877841687208984,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008357545860546785,
      "loss": 4.8765,
      "step": 7943
    },
    {
      "epoch": 1.0879211175020542,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008357334901443751,
      "loss": 4.9317,
      "step": 7944
    },
    {
      "epoch": 1.08805806628321,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008357123910374063,
      "loss": 4.8755,
      "step": 7945
    },
    {
      "epoch": 1.088195015064366,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008356912887339471,
      "loss": 4.8639,
      "step": 7946
    },
    {
      "epoch": 1.0883319638455218,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008356701832341722,
      "loss": 4.8955,
      "step": 7947
    },
    {
      "epoch": 1.0884689126266776,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008356490745382566,
      "loss": 4.9508,
      "step": 7948
    },
    {
      "epoch": 1.0886058614078336,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008356279626463753,
      "loss": 4.8818,
      "step": 7949
    },
    {
      "epoch": 1.0887428101889893,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008356068475587033,
      "loss": 4.9474,
      "step": 7950
    },
    {
      "epoch": 1.088879758970145,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008355857292754151,
      "loss": 4.8781,
      "step": 7951
    },
    {
      "epoch": 1.089016707751301,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008355646077966863,
      "loss": 4.831,
      "step": 7952
    },
    {
      "epoch": 1.0891536565324569,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008355434831226917,
      "loss": 4.9538,
      "step": 7953
    },
    {
      "epoch": 1.0892906053136127,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008355223552536064,
      "loss": 4.8943,
      "step": 7954
    },
    {
      "epoch": 1.0894275540947687,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008355012241896056,
      "loss": 4.9063,
      "step": 7955
    },
    {
      "epoch": 1.0895645028759244,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008354800899308641,
      "loss": 4.9065,
      "step": 7956
    },
    {
      "epoch": 1.0897014516570802,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008354589524775572,
      "loss": 4.901,
      "step": 7957
    },
    {
      "epoch": 1.0898384004382362,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008354378118298604,
      "loss": 4.8728,
      "step": 7958
    },
    {
      "epoch": 1.089975349219392,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008354166679879483,
      "loss": 4.9268,
      "step": 7959
    },
    {
      "epoch": 1.0901122980005478,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008353955209519966,
      "loss": 4.9188,
      "step": 7960
    },
    {
      "epoch": 1.0902492467817035,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008353743707221803,
      "loss": 4.9126,
      "step": 7961
    },
    {
      "epoch": 1.0903861955628595,
      "grad_norm": 0.375,
      "learning_rate": 0.0008353532172986748,
      "loss": 4.9793,
      "step": 7962
    },
    {
      "epoch": 1.0905231443440153,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008353320606816553,
      "loss": 4.9476,
      "step": 7963
    },
    {
      "epoch": 1.090660093125171,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008353109008712973,
      "loss": 4.9752,
      "step": 7964
    },
    {
      "epoch": 1.090797041906327,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000835289737867776,
      "loss": 4.9794,
      "step": 7965
    },
    {
      "epoch": 1.0909339906874829,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008352685716712669,
      "loss": 5.0172,
      "step": 7966
    },
    {
      "epoch": 1.0910709394686386,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008352474022819452,
      "loss": 4.8971,
      "step": 7967
    },
    {
      "epoch": 1.0912078882497946,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008352262296999866,
      "loss": 4.96,
      "step": 7968
    },
    {
      "epoch": 1.0913448370309504,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008352050539255664,
      "loss": 5.0089,
      "step": 7969
    },
    {
      "epoch": 1.0914817858121062,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008351838749588602,
      "loss": 4.909,
      "step": 7970
    },
    {
      "epoch": 1.0916187345932622,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008351626928000433,
      "loss": 4.9595,
      "step": 7971
    },
    {
      "epoch": 1.091755683374418,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008351415074492915,
      "loss": 4.8896,
      "step": 7972
    },
    {
      "epoch": 1.0918926321555738,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008351203189067802,
      "loss": 4.954,
      "step": 7973
    },
    {
      "epoch": 1.0920295809367297,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000835099127172685,
      "loss": 4.9092,
      "step": 7974
    },
    {
      "epoch": 1.0921665297178855,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008350779322471815,
      "loss": 5.0033,
      "step": 7975
    },
    {
      "epoch": 1.0923034784990413,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008350567341304456,
      "loss": 5.0087,
      "step": 7976
    },
    {
      "epoch": 1.0924404272801973,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008350355328226528,
      "loss": 4.9437,
      "step": 7977
    },
    {
      "epoch": 1.092577376061353,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008350143283239787,
      "loss": 4.9128,
      "step": 7978
    },
    {
      "epoch": 1.0927143248425089,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008349931206345991,
      "loss": 4.8611,
      "step": 7979
    },
    {
      "epoch": 1.0928512736236646,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008349719097546898,
      "loss": 4.9498,
      "step": 7980
    },
    {
      "epoch": 1.0929882224048206,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008349506956844266,
      "loss": 4.8528,
      "step": 7981
    },
    {
      "epoch": 1.0931251711859764,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008349294784239853,
      "loss": 4.9088,
      "step": 7982
    },
    {
      "epoch": 1.0932621199671324,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008349082579735415,
      "loss": 4.9713,
      "step": 7983
    },
    {
      "epoch": 1.0933990687482882,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008348870343332714,
      "loss": 4.925,
      "step": 7984
    },
    {
      "epoch": 1.093536017529444,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008348658075033507,
      "loss": 5.0322,
      "step": 7985
    },
    {
      "epoch": 1.0936729663105997,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008348445774839554,
      "loss": 4.8434,
      "step": 7986
    },
    {
      "epoch": 1.0938099150917557,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008348233442752612,
      "loss": 5.0142,
      "step": 7987
    },
    {
      "epoch": 1.0939468638729115,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008348021078774444,
      "loss": 4.8333,
      "step": 7988
    },
    {
      "epoch": 1.0940838126540673,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008347808682906808,
      "loss": 4.8898,
      "step": 7989
    },
    {
      "epoch": 1.0942207614352233,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008347596255151465,
      "loss": 4.8884,
      "step": 7990
    },
    {
      "epoch": 1.094357710216379,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008347383795510174,
      "loss": 4.9195,
      "step": 7991
    },
    {
      "epoch": 1.0944946589975348,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008347171303984698,
      "loss": 4.9803,
      "step": 7992
    },
    {
      "epoch": 1.0946316077786908,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008346958780576796,
      "loss": 4.9214,
      "step": 7993
    },
    {
      "epoch": 1.0947685565598466,
      "grad_norm": 0.4921875,
      "learning_rate": 0.000834674622528823,
      "loss": 4.9805,
      "step": 7994
    },
    {
      "epoch": 1.0949055053410024,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008346533638120761,
      "loss": 4.9505,
      "step": 7995
    },
    {
      "epoch": 1.0950424541221584,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008346321019076151,
      "loss": 4.8604,
      "step": 7996
    },
    {
      "epoch": 1.0951794029033142,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008346108368156162,
      "loss": 4.8874,
      "step": 7997
    },
    {
      "epoch": 1.09531635168447,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008345895685362558,
      "loss": 4.9851,
      "step": 7998
    },
    {
      "epoch": 1.095453300465626,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008345682970697099,
      "loss": 4.9307,
      "step": 7999
    },
    {
      "epoch": 1.0955902492467817,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008345470224161549,
      "loss": 4.907,
      "step": 8000
    },
    {
      "epoch": 1.0957271980279375,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008345257445757672,
      "loss": 4.8429,
      "step": 8001
    },
    {
      "epoch": 1.0958641468090935,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008345044635487228,
      "loss": 4.9594,
      "step": 8002
    },
    {
      "epoch": 1.0960010955902493,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008344831793351984,
      "loss": 4.8963,
      "step": 8003
    },
    {
      "epoch": 1.096138044371405,
      "grad_norm": 0.5,
      "learning_rate": 0.0008344618919353703,
      "loss": 5.0046,
      "step": 8004
    },
    {
      "epoch": 1.0962749931525608,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008344406013494148,
      "loss": 4.9319,
      "step": 8005
    },
    {
      "epoch": 1.0964119419337168,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008344193075775086,
      "loss": 4.9604,
      "step": 8006
    },
    {
      "epoch": 1.0965488907148726,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008343980106198279,
      "loss": 4.9172,
      "step": 8007
    },
    {
      "epoch": 1.0966858394960286,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008343767104765492,
      "loss": 4.9339,
      "step": 8008
    },
    {
      "epoch": 1.0968227882771844,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008343554071478492,
      "loss": 4.96,
      "step": 8009
    },
    {
      "epoch": 1.0969597370583402,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008343341006339044,
      "loss": 4.9748,
      "step": 8010
    },
    {
      "epoch": 1.097096685839496,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008343127909348913,
      "loss": 4.9362,
      "step": 8011
    },
    {
      "epoch": 1.097233634620652,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008342914780509864,
      "loss": 4.9117,
      "step": 8012
    },
    {
      "epoch": 1.0973705834018077,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008342701619823666,
      "loss": 4.8647,
      "step": 8013
    },
    {
      "epoch": 1.0975075321829635,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008342488427292082,
      "loss": 4.877,
      "step": 8014
    },
    {
      "epoch": 1.0976444809641195,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008342275202916883,
      "loss": 4.9575,
      "step": 8015
    },
    {
      "epoch": 1.0977814297452753,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008342061946699834,
      "loss": 4.9449,
      "step": 8016
    },
    {
      "epoch": 1.097918378526431,
      "grad_norm": 0.390625,
      "learning_rate": 0.00083418486586427,
      "loss": 4.9408,
      "step": 8017
    },
    {
      "epoch": 1.098055327307587,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008341635338747252,
      "loss": 4.9075,
      "step": 8018
    },
    {
      "epoch": 1.0981922760887428,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008341421987015257,
      "loss": 4.975,
      "step": 8019
    },
    {
      "epoch": 1.0983292248698986,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008341208603448483,
      "loss": 4.8909,
      "step": 8020
    },
    {
      "epoch": 1.0984661736510546,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008340995188048695,
      "loss": 4.9048,
      "step": 8021
    },
    {
      "epoch": 1.0986031224322104,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008340781740817667,
      "loss": 4.9523,
      "step": 8022
    },
    {
      "epoch": 1.0987400712133661,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0008340568261757166,
      "loss": 4.9011,
      "step": 8023
    },
    {
      "epoch": 1.0988770199945221,
      "grad_norm": 0.390625,
      "learning_rate": 0.000834035475086896,
      "loss": 4.9739,
      "step": 8024
    },
    {
      "epoch": 1.099013968775678,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008340141208154819,
      "loss": 4.9043,
      "step": 8025
    },
    {
      "epoch": 1.0991509175568337,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008339927633616513,
      "loss": 4.9183,
      "step": 8026
    },
    {
      "epoch": 1.0992878663379897,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008339714027255812,
      "loss": 4.943,
      "step": 8027
    },
    {
      "epoch": 1.0994248151191455,
      "grad_norm": 0.375,
      "learning_rate": 0.0008339500389074484,
      "loss": 4.971,
      "step": 8028
    },
    {
      "epoch": 1.0995617639003012,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008339286719074302,
      "loss": 4.9221,
      "step": 8029
    },
    {
      "epoch": 1.099698712681457,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008339073017257037,
      "loss": 4.9292,
      "step": 8030
    },
    {
      "epoch": 1.099835661462613,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000833885928362446,
      "loss": 4.9827,
      "step": 8031
    },
    {
      "epoch": 1.0999726102437688,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000833864551817834,
      "loss": 4.9439,
      "step": 8032
    },
    {
      "epoch": 1.1001095590249246,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008338431720920451,
      "loss": 4.942,
      "step": 8033
    },
    {
      "epoch": 1.1002465078060806,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008338217891852562,
      "loss": 4.984,
      "step": 8034
    },
    {
      "epoch": 1.1003834565872364,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008338004030976447,
      "loss": 4.8904,
      "step": 8035
    },
    {
      "epoch": 1.1005204053683921,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008337790138293878,
      "loss": 4.9504,
      "step": 8036
    },
    {
      "epoch": 1.1006573541495481,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008337576213806629,
      "loss": 4.8257,
      "step": 8037
    },
    {
      "epoch": 1.100794302930704,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000833736225751647,
      "loss": 4.9199,
      "step": 8038
    },
    {
      "epoch": 1.1009312517118597,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008337148269425177,
      "loss": 4.9176,
      "step": 8039
    },
    {
      "epoch": 1.1010682004930157,
      "grad_norm": 0.359375,
      "learning_rate": 0.000833693424953452,
      "loss": 5.0061,
      "step": 8040
    },
    {
      "epoch": 1.1012051492741715,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008336720197846277,
      "loss": 4.8773,
      "step": 8041
    },
    {
      "epoch": 1.1013420980553272,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008336506114362218,
      "loss": 4.9091,
      "step": 8042
    },
    {
      "epoch": 1.1014790468364832,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000833629199908412,
      "loss": 4.921,
      "step": 8043
    },
    {
      "epoch": 1.101615995617639,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008336077852013754,
      "loss": 5.0066,
      "step": 8044
    },
    {
      "epoch": 1.1017529443987948,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008335863673152897,
      "loss": 4.9663,
      "step": 8045
    },
    {
      "epoch": 1.1018898931799508,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008335649462503325,
      "loss": 4.8656,
      "step": 8046
    },
    {
      "epoch": 1.1020268419611066,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008335435220066811,
      "loss": 5.0119,
      "step": 8047
    },
    {
      "epoch": 1.1021637907422623,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008335220945845132,
      "loss": 5.0132,
      "step": 8048
    },
    {
      "epoch": 1.1023007395234183,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008335006639840064,
      "loss": 4.9562,
      "step": 8049
    },
    {
      "epoch": 1.1024376883045741,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000833479230205338,
      "loss": 4.8605,
      "step": 8050
    },
    {
      "epoch": 1.10257463708573,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000833457793248686,
      "loss": 4.9387,
      "step": 8051
    },
    {
      "epoch": 1.102711585866886,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008334363531142279,
      "loss": 4.9708,
      "step": 8052
    },
    {
      "epoch": 1.1028485346480417,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008334149098021413,
      "loss": 4.9228,
      "step": 8053
    },
    {
      "epoch": 1.1029854834291974,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000833393463312604,
      "loss": 4.9269,
      "step": 8054
    },
    {
      "epoch": 1.1031224322103532,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008333720136457937,
      "loss": 4.8775,
      "step": 8055
    },
    {
      "epoch": 1.1032593809915092,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008333505608018882,
      "loss": 4.8743,
      "step": 8056
    },
    {
      "epoch": 1.103396329772665,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008333291047810651,
      "loss": 4.8478,
      "step": 8057
    },
    {
      "epoch": 1.1035332785538208,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008333076455835026,
      "loss": 4.9096,
      "step": 8058
    },
    {
      "epoch": 1.1036702273349768,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008332861832093781,
      "loss": 4.8922,
      "step": 8059
    },
    {
      "epoch": 1.1038071761161325,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008332647176588697,
      "loss": 4.9137,
      "step": 8060
    },
    {
      "epoch": 1.1039441248972883,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008332432489321554,
      "loss": 4.9492,
      "step": 8061
    },
    {
      "epoch": 1.1040810736784443,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008332217770294129,
      "loss": 4.9738,
      "step": 8062
    },
    {
      "epoch": 1.1042180224596,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008332003019508202,
      "loss": 4.9096,
      "step": 8063
    },
    {
      "epoch": 1.1043549712407559,
      "grad_norm": 0.375,
      "learning_rate": 0.0008331788236965554,
      "loss": 4.9181,
      "step": 8064
    },
    {
      "epoch": 1.1044919200219119,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008331573422667962,
      "loss": 4.9743,
      "step": 8065
    },
    {
      "epoch": 1.1046288688030677,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008331358576617209,
      "loss": 4.8967,
      "step": 8066
    },
    {
      "epoch": 1.1047658175842234,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008331143698815075,
      "loss": 4.8972,
      "step": 8067
    },
    {
      "epoch": 1.1049027663653794,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000833092878926334,
      "loss": 4.9344,
      "step": 8068
    },
    {
      "epoch": 1.1050397151465352,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008330713847963785,
      "loss": 4.8468,
      "step": 8069
    },
    {
      "epoch": 1.105176663927691,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008330498874918191,
      "loss": 5.0026,
      "step": 8070
    },
    {
      "epoch": 1.105313612708847,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008330283870128341,
      "loss": 4.9356,
      "step": 8071
    },
    {
      "epoch": 1.1054505614900028,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008330068833596014,
      "loss": 4.9369,
      "step": 8072
    },
    {
      "epoch": 1.1055875102711585,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008329853765322995,
      "loss": 4.9317,
      "step": 8073
    },
    {
      "epoch": 1.1057244590523145,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008329638665311067,
      "loss": 4.9487,
      "step": 8074
    },
    {
      "epoch": 1.1058614078334703,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008329423533562008,
      "loss": 4.8604,
      "step": 8075
    },
    {
      "epoch": 1.105998356614626,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008329208370077605,
      "loss": 4.929,
      "step": 8076
    },
    {
      "epoch": 1.106135305395782,
      "grad_norm": 0.375,
      "learning_rate": 0.0008328993174859641,
      "loss": 4.8805,
      "step": 8077
    },
    {
      "epoch": 1.1062722541769379,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008328777947909897,
      "loss": 4.9283,
      "step": 8078
    },
    {
      "epoch": 1.1064092029580936,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008328562689230158,
      "loss": 4.9079,
      "step": 8079
    },
    {
      "epoch": 1.1065461517392494,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008328347398822207,
      "loss": 4.9007,
      "step": 8080
    },
    {
      "epoch": 1.1066831005204054,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008328132076687828,
      "loss": 4.9636,
      "step": 8081
    },
    {
      "epoch": 1.1068200493015612,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008327916722828809,
      "loss": 4.9119,
      "step": 8082
    },
    {
      "epoch": 1.106956998082717,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000832770133724693,
      "loss": 4.9715,
      "step": 8083
    },
    {
      "epoch": 1.107093946863873,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000832748591994398,
      "loss": 4.8685,
      "step": 8084
    },
    {
      "epoch": 1.1072308956450287,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008327270470921739,
      "loss": 4.9168,
      "step": 8085
    },
    {
      "epoch": 1.1073678444261845,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008327054990181997,
      "loss": 4.9107,
      "step": 8086
    },
    {
      "epoch": 1.1075047932073405,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000832683947772654,
      "loss": 4.9,
      "step": 8087
    },
    {
      "epoch": 1.1076417419884963,
      "grad_norm": 0.453125,
      "learning_rate": 0.000832662393355715,
      "loss": 4.9019,
      "step": 8088
    },
    {
      "epoch": 1.107778690769652,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008326408357675617,
      "loss": 4.925,
      "step": 8089
    },
    {
      "epoch": 1.107915639550808,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008326192750083726,
      "loss": 4.8927,
      "step": 8090
    },
    {
      "epoch": 1.1080525883319639,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008325977110783263,
      "loss": 5.0356,
      "step": 8091
    },
    {
      "epoch": 1.1081895371131196,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008325761439776018,
      "loss": 4.9258,
      "step": 8092
    },
    {
      "epoch": 1.1083264858942756,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008325545737063773,
      "loss": 4.8791,
      "step": 8093
    },
    {
      "epoch": 1.1084634346754314,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008325330002648322,
      "loss": 4.9647,
      "step": 8094
    },
    {
      "epoch": 1.1086003834565872,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008325114236531449,
      "loss": 4.9639,
      "step": 8095
    },
    {
      "epoch": 1.1087373322377432,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008324898438714942,
      "loss": 4.9671,
      "step": 8096
    },
    {
      "epoch": 1.108874281018899,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008324682609200592,
      "loss": 4.8853,
      "step": 8097
    },
    {
      "epoch": 1.1090112298000547,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008324466747990183,
      "loss": 4.967,
      "step": 8098
    },
    {
      "epoch": 1.1091481785812107,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000832425085508551,
      "loss": 4.937,
      "step": 8099
    },
    {
      "epoch": 1.1092851273623665,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008324034930488356,
      "loss": 4.9482,
      "step": 8100
    },
    {
      "epoch": 1.1094220761435223,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008323818974200516,
      "loss": 4.9736,
      "step": 8101
    },
    {
      "epoch": 1.1095590249246783,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008323602986223775,
      "loss": 4.9389,
      "step": 8102
    },
    {
      "epoch": 1.109695973705834,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008323386966559924,
      "loss": 4.9624,
      "step": 8103
    },
    {
      "epoch": 1.1098329224869898,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008323170915210756,
      "loss": 4.8803,
      "step": 8104
    },
    {
      "epoch": 1.1099698712681456,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008322954832178061,
      "loss": 4.9583,
      "step": 8105
    },
    {
      "epoch": 1.1101068200493016,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008322738717463625,
      "loss": 4.9215,
      "step": 8106
    },
    {
      "epoch": 1.1102437688304574,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008322522571069244,
      "loss": 4.9463,
      "step": 8107
    },
    {
      "epoch": 1.1103807176116132,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008322306392996708,
      "loss": 5.0277,
      "step": 8108
    },
    {
      "epoch": 1.1105176663927692,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008322090183247808,
      "loss": 4.919,
      "step": 8109
    },
    {
      "epoch": 1.110654615173925,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008321873941824335,
      "loss": 4.9,
      "step": 8110
    },
    {
      "epoch": 1.1107915639550807,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008321657668728083,
      "loss": 4.9428,
      "step": 8111
    },
    {
      "epoch": 1.1109285127362367,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008321441363960843,
      "loss": 4.9232,
      "step": 8112
    },
    {
      "epoch": 1.1110654615173925,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008321225027524406,
      "loss": 4.8973,
      "step": 8113
    },
    {
      "epoch": 1.1112024102985483,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008321008659420569,
      "loss": 4.9336,
      "step": 8114
    },
    {
      "epoch": 1.1113393590797043,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008320792259651122,
      "loss": 4.8335,
      "step": 8115
    },
    {
      "epoch": 1.11147630786086,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008320575828217859,
      "loss": 4.9632,
      "step": 8116
    },
    {
      "epoch": 1.1116132566420158,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008320359365122573,
      "loss": 4.9183,
      "step": 8117
    },
    {
      "epoch": 1.1117502054231718,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000832014287036706,
      "loss": 4.9739,
      "step": 8118
    },
    {
      "epoch": 1.1118871542043276,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008319926343953112,
      "loss": 4.8706,
      "step": 8119
    },
    {
      "epoch": 1.1120241029854834,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008319709785882523,
      "loss": 4.9636,
      "step": 8120
    },
    {
      "epoch": 1.1121610517666394,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008319493196157091,
      "loss": 4.8741,
      "step": 8121
    },
    {
      "epoch": 1.1122980005477952,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008319276574778608,
      "loss": 4.8944,
      "step": 8122
    },
    {
      "epoch": 1.112434949328951,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008319059921748869,
      "loss": 4.9253,
      "step": 8123
    },
    {
      "epoch": 1.1125718981101067,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008318843237069672,
      "loss": 4.956,
      "step": 8124
    },
    {
      "epoch": 1.1127088468912627,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000831862652074281,
      "loss": 4.8805,
      "step": 8125
    },
    {
      "epoch": 1.1128457956724185,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000831840977277008,
      "loss": 4.8883,
      "step": 8126
    },
    {
      "epoch": 1.1129827444535745,
      "grad_norm": 0.5,
      "learning_rate": 0.0008318192993153279,
      "loss": 4.9517,
      "step": 8127
    },
    {
      "epoch": 1.1131196932347303,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008317976181894202,
      "loss": 4.8904,
      "step": 8128
    },
    {
      "epoch": 1.113256642015886,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008317759338994648,
      "loss": 4.9129,
      "step": 8129
    },
    {
      "epoch": 1.1133935907970418,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008317542464456412,
      "loss": 5.0333,
      "step": 8130
    },
    {
      "epoch": 1.1135305395781978,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008317325558281291,
      "loss": 4.9446,
      "step": 8131
    },
    {
      "epoch": 1.1136674883593536,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008317108620471085,
      "loss": 4.9743,
      "step": 8132
    },
    {
      "epoch": 1.1138044371405094,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000831689165102759,
      "loss": 4.9617,
      "step": 8133
    },
    {
      "epoch": 1.1139413859216654,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008316674649952603,
      "loss": 4.886,
      "step": 8134
    },
    {
      "epoch": 1.1140783347028211,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008316457617247925,
      "loss": 4.9638,
      "step": 8135
    },
    {
      "epoch": 1.114215283483977,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008316240552915352,
      "loss": 4.9071,
      "step": 8136
    },
    {
      "epoch": 1.114352232265133,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008316023456956686,
      "loss": 4.9129,
      "step": 8137
    },
    {
      "epoch": 1.1144891810462887,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008315806329373723,
      "loss": 4.9002,
      "step": 8138
    },
    {
      "epoch": 1.1146261298274445,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008315589170168263,
      "loss": 4.9143,
      "step": 8139
    },
    {
      "epoch": 1.1147630786086005,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008315371979342108,
      "loss": 4.8639,
      "step": 8140
    },
    {
      "epoch": 1.1149000273897562,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008315154756897056,
      "loss": 4.9074,
      "step": 8141
    },
    {
      "epoch": 1.115036976170912,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008314937502834907,
      "loss": 4.8752,
      "step": 8142
    },
    {
      "epoch": 1.115173924952068,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008314720217157462,
      "loss": 5.0206,
      "step": 8143
    },
    {
      "epoch": 1.1153108737332238,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008314502899866522,
      "loss": 4.8126,
      "step": 8144
    },
    {
      "epoch": 1.1154478225143796,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008314285550963886,
      "loss": 4.9577,
      "step": 8145
    },
    {
      "epoch": 1.1155847712955356,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008314068170451358,
      "loss": 4.9617,
      "step": 8146
    },
    {
      "epoch": 1.1157217200766913,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008313850758330737,
      "loss": 4.8681,
      "step": 8147
    },
    {
      "epoch": 1.1158586688578471,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008313633314603828,
      "loss": 4.8157,
      "step": 8148
    },
    {
      "epoch": 1.115995617639003,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000831341583927243,
      "loss": 5.0173,
      "step": 8149
    },
    {
      "epoch": 1.116132566420159,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008313198332338346,
      "loss": 4.9153,
      "step": 8150
    },
    {
      "epoch": 1.1162695152013147,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008312980793803379,
      "loss": 4.897,
      "step": 8151
    },
    {
      "epoch": 1.1164064639824705,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008312763223669331,
      "loss": 4.8731,
      "step": 8152
    },
    {
      "epoch": 1.1165434127636265,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008312545621938006,
      "loss": 4.8812,
      "step": 8153
    },
    {
      "epoch": 1.1166803615447822,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008312327988611208,
      "loss": 4.8259,
      "step": 8154
    },
    {
      "epoch": 1.116817310325938,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008312110323690739,
      "loss": 4.8576,
      "step": 8155
    },
    {
      "epoch": 1.116954259107094,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008311892627178403,
      "loss": 5.0107,
      "step": 8156
    },
    {
      "epoch": 1.1170912078882498,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008311674899076003,
      "loss": 4.9146,
      "step": 8157
    },
    {
      "epoch": 1.1172281566694056,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008311457139385347,
      "loss": 4.9002,
      "step": 8158
    },
    {
      "epoch": 1.1173651054505616,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008311239348108237,
      "loss": 4.9059,
      "step": 8159
    },
    {
      "epoch": 1.1175020542317173,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008311021525246479,
      "loss": 4.8657,
      "step": 8160
    },
    {
      "epoch": 1.1176390030128731,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008310803670801875,
      "loss": 4.9644,
      "step": 8161
    },
    {
      "epoch": 1.117775951794029,
      "grad_norm": 0.375,
      "learning_rate": 0.0008310585784776235,
      "loss": 5.0447,
      "step": 8162
    },
    {
      "epoch": 1.1179129005751849,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008310367867171363,
      "loss": 4.9153,
      "step": 8163
    },
    {
      "epoch": 1.1180498493563407,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008310149917989064,
      "loss": 4.9957,
      "step": 8164
    },
    {
      "epoch": 1.1181867981374967,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008309931937231144,
      "loss": 4.9324,
      "step": 8165
    },
    {
      "epoch": 1.1183237469186524,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008309713924899411,
      "loss": 4.9431,
      "step": 8166
    },
    {
      "epoch": 1.1184606956998082,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000830949588099567,
      "loss": 4.8839,
      "step": 8167
    },
    {
      "epoch": 1.1185976444809642,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000830927780552173,
      "loss": 4.8955,
      "step": 8168
    },
    {
      "epoch": 1.11873459326212,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008309059698479395,
      "loss": 4.8511,
      "step": 8169
    },
    {
      "epoch": 1.1188715420432758,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008308841559870477,
      "loss": 4.959,
      "step": 8170
    },
    {
      "epoch": 1.1190084908244318,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000830862338969678,
      "loss": 4.9746,
      "step": 8171
    },
    {
      "epoch": 1.1191454396055875,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008308405187960115,
      "loss": 4.8721,
      "step": 8172
    },
    {
      "epoch": 1.1192823883867433,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008308186954662288,
      "loss": 4.9442,
      "step": 8173
    },
    {
      "epoch": 1.119419337167899,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008307968689805108,
      "loss": 4.8907,
      "step": 8174
    },
    {
      "epoch": 1.119556285949055,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008307750393390384,
      "loss": 4.9726,
      "step": 8175
    },
    {
      "epoch": 1.1196932347302109,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008307532065419924,
      "loss": 4.9686,
      "step": 8176
    },
    {
      "epoch": 1.1198301835113667,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008307313705895541,
      "loss": 4.9688,
      "step": 8177
    },
    {
      "epoch": 1.1199671322925226,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000830709531481904,
      "loss": 4.8758,
      "step": 8178
    },
    {
      "epoch": 1.1201040810736784,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008306876892192233,
      "loss": 4.8989,
      "step": 8179
    },
    {
      "epoch": 1.1202410298548342,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008306658438016931,
      "loss": 4.9466,
      "step": 8180
    },
    {
      "epoch": 1.1203779786359902,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008306439952294944,
      "loss": 4.9511,
      "step": 8181
    },
    {
      "epoch": 1.120514927417146,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000830622143502808,
      "loss": 4.9501,
      "step": 8182
    },
    {
      "epoch": 1.1206518761983018,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008306002886218154,
      "loss": 4.9252,
      "step": 8183
    },
    {
      "epoch": 1.1207888249794578,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008305784305866974,
      "loss": 4.9417,
      "step": 8184
    },
    {
      "epoch": 1.1209257737606135,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008305565693976354,
      "loss": 4.9502,
      "step": 8185
    },
    {
      "epoch": 1.1210627225417693,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008305347050548103,
      "loss": 4.8738,
      "step": 8186
    },
    {
      "epoch": 1.1211996713229253,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008305128375584034,
      "loss": 4.9567,
      "step": 8187
    },
    {
      "epoch": 1.121336620104081,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008304909669085961,
      "loss": 4.9431,
      "step": 8188
    },
    {
      "epoch": 1.1214735688852369,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008304690931055694,
      "loss": 4.8823,
      "step": 8189
    },
    {
      "epoch": 1.1216105176663929,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008304472161495045,
      "loss": 4.998,
      "step": 8190
    },
    {
      "epoch": 1.1217474664475486,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000830425336040583,
      "loss": 4.9347,
      "step": 8191
    },
    {
      "epoch": 1.1218844152287044,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008304034527789861,
      "loss": 4.9301,
      "step": 8192
    },
    {
      "epoch": 1.1220213640098604,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008303815663648952,
      "loss": 4.9071,
      "step": 8193
    },
    {
      "epoch": 1.1221583127910162,
      "grad_norm": 0.5,
      "learning_rate": 0.0008303596767984916,
      "loss": 4.8781,
      "step": 8194
    },
    {
      "epoch": 1.122295261572172,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008303377840799566,
      "loss": 4.9146,
      "step": 8195
    },
    {
      "epoch": 1.122432210353328,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008303158882094719,
      "loss": 4.972,
      "step": 8196
    },
    {
      "epoch": 1.1225691591344837,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008302939891872187,
      "loss": 4.9539,
      "step": 8197
    },
    {
      "epoch": 1.1227061079156395,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008302720870133787,
      "loss": 4.9686,
      "step": 8198
    },
    {
      "epoch": 1.1228430566967953,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008302501816881333,
      "loss": 4.9638,
      "step": 8199
    },
    {
      "epoch": 1.1229800054779513,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000830228273211664,
      "loss": 4.9494,
      "step": 8200
    },
    {
      "epoch": 1.123116954259107,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008302063615841524,
      "loss": 4.9438,
      "step": 8201
    },
    {
      "epoch": 1.1232539030402628,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00083018444680578,
      "loss": 4.9026,
      "step": 8202
    },
    {
      "epoch": 1.1233908518214188,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008301625288767285,
      "loss": 4.9381,
      "step": 8203
    },
    {
      "epoch": 1.1235278006025746,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008301406077971796,
      "loss": 4.9627,
      "step": 8204
    },
    {
      "epoch": 1.1236647493837304,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000830118683567315,
      "loss": 4.9309,
      "step": 8205
    },
    {
      "epoch": 1.1238016981648864,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008300967561873161,
      "loss": 4.9223,
      "step": 8206
    },
    {
      "epoch": 1.1239386469460422,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008300748256573649,
      "loss": 4.9122,
      "step": 8207
    },
    {
      "epoch": 1.124075595727198,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008300528919776431,
      "loss": 4.8872,
      "step": 8208
    },
    {
      "epoch": 1.124212544508354,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008300309551483323,
      "loss": 4.9339,
      "step": 8209
    },
    {
      "epoch": 1.1243494932895097,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008300090151696145,
      "loss": 4.9742,
      "step": 8210
    },
    {
      "epoch": 1.1244864420706655,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008299870720416715,
      "loss": 4.9123,
      "step": 8211
    },
    {
      "epoch": 1.1246233908518215,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000829965125764685,
      "loss": 4.91,
      "step": 8212
    },
    {
      "epoch": 1.1247603396329773,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000829943176338837,
      "loss": 4.9658,
      "step": 8213
    },
    {
      "epoch": 1.124897288414133,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008299212237643093,
      "loss": 4.9476,
      "step": 8214
    },
    {
      "epoch": 1.125034237195289,
      "grad_norm": 0.375,
      "learning_rate": 0.000829899268041284,
      "loss": 4.863,
      "step": 8215
    },
    {
      "epoch": 1.1251711859764448,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008298773091699428,
      "loss": 4.8392,
      "step": 8216
    },
    {
      "epoch": 1.1253081347576006,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008298553471504679,
      "loss": 4.939,
      "step": 8217
    },
    {
      "epoch": 1.1254450835387564,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008298333819830412,
      "loss": 4.9427,
      "step": 8218
    },
    {
      "epoch": 1.1255820323199124,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008298114136678447,
      "loss": 4.9489,
      "step": 8219
    },
    {
      "epoch": 1.1257189811010682,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008297894422050606,
      "loss": 4.9148,
      "step": 8220
    },
    {
      "epoch": 1.1258559298822242,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008297674675948709,
      "loss": 5.0088,
      "step": 8221
    },
    {
      "epoch": 1.12599287866338,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008297454898374577,
      "loss": 4.9516,
      "step": 8222
    },
    {
      "epoch": 1.1261298274445357,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008297235089330031,
      "loss": 4.8932,
      "step": 8223
    },
    {
      "epoch": 1.1262667762256915,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008297015248816894,
      "loss": 4.9301,
      "step": 8224
    },
    {
      "epoch": 1.1264037250068475,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008296795376836985,
      "loss": 4.9389,
      "step": 8225
    },
    {
      "epoch": 1.1265406737880033,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008296575473392129,
      "loss": 4.8077,
      "step": 8226
    },
    {
      "epoch": 1.126677622569159,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008296355538484147,
      "loss": 4.8916,
      "step": 8227
    },
    {
      "epoch": 1.126814571350315,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008296135572114862,
      "loss": 4.9472,
      "step": 8228
    },
    {
      "epoch": 1.1269515201314708,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008295915574286099,
      "loss": 4.9631,
      "step": 8229
    },
    {
      "epoch": 1.1270884689126266,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008295695544999677,
      "loss": 4.9853,
      "step": 8230
    },
    {
      "epoch": 1.1272254176937826,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008295475484257421,
      "loss": 4.782,
      "step": 8231
    },
    {
      "epoch": 1.1273623664749384,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008295255392061156,
      "loss": 4.8168,
      "step": 8232
    },
    {
      "epoch": 1.1274993152560941,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008295035268412706,
      "loss": 5.0202,
      "step": 8233
    },
    {
      "epoch": 1.1276362640372501,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008294815113313894,
      "loss": 4.9182,
      "step": 8234
    },
    {
      "epoch": 1.127773212818406,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008294594926766545,
      "loss": 4.9315,
      "step": 8235
    },
    {
      "epoch": 1.1279101615995617,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008294374708772484,
      "loss": 4.876,
      "step": 8236
    },
    {
      "epoch": 1.1280471103807177,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008294154459333534,
      "loss": 4.8751,
      "step": 8237
    },
    {
      "epoch": 1.1281840591618735,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008293934178451524,
      "loss": 4.944,
      "step": 8238
    },
    {
      "epoch": 1.1283210079430293,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008293713866128277,
      "loss": 4.9225,
      "step": 8239
    },
    {
      "epoch": 1.1284579567241853,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008293493522365617,
      "loss": 4.8256,
      "step": 8240
    },
    {
      "epoch": 1.128594905505341,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008293273147165375,
      "loss": 4.893,
      "step": 8241
    },
    {
      "epoch": 1.1287318542864968,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008293052740529373,
      "loss": 4.8644,
      "step": 8242
    },
    {
      "epoch": 1.1288688030676526,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008292832302459438,
      "loss": 5.0348,
      "step": 8243
    },
    {
      "epoch": 1.1290057518488086,
      "grad_norm": 0.375,
      "learning_rate": 0.00082926118329574,
      "loss": 4.9122,
      "step": 8244
    },
    {
      "epoch": 1.1291427006299644,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008292391332025083,
      "loss": 4.8909,
      "step": 8245
    },
    {
      "epoch": 1.1292796494111204,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008292170799664315,
      "loss": 5.0287,
      "step": 8246
    },
    {
      "epoch": 1.1294165981922761,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008291950235876926,
      "loss": 4.9188,
      "step": 8247
    },
    {
      "epoch": 1.129553546973432,
      "grad_norm": 0.375,
      "learning_rate": 0.0008291729640664741,
      "loss": 4.9647,
      "step": 8248
    },
    {
      "epoch": 1.1296904957545877,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008291509014029587,
      "loss": 4.8783,
      "step": 8249
    },
    {
      "epoch": 1.1298274445357437,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008291288355973297,
      "loss": 4.9325,
      "step": 8250
    },
    {
      "epoch": 1.1299643933168995,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008291067666497695,
      "loss": 4.9034,
      "step": 8251
    },
    {
      "epoch": 1.1301013420980552,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008290846945604613,
      "loss": 4.9153,
      "step": 8252
    },
    {
      "epoch": 1.1302382908792112,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008290626193295879,
      "loss": 4.8209,
      "step": 8253
    },
    {
      "epoch": 1.130375239660367,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008290405409573323,
      "loss": 4.9272,
      "step": 8254
    },
    {
      "epoch": 1.1305121884415228,
      "grad_norm": 0.375,
      "learning_rate": 0.0008290184594438772,
      "loss": 4.8323,
      "step": 8255
    },
    {
      "epoch": 1.1306491372226788,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008289963747894059,
      "loss": 4.9206,
      "step": 8256
    },
    {
      "epoch": 1.1307860860038346,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008289742869941015,
      "loss": 5.0114,
      "step": 8257
    },
    {
      "epoch": 1.1309230347849903,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008289521960581467,
      "loss": 4.8548,
      "step": 8258
    },
    {
      "epoch": 1.1310599835661463,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008289301019817249,
      "loss": 4.8354,
      "step": 8259
    },
    {
      "epoch": 1.1311969323473021,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008289080047650188,
      "loss": 4.905,
      "step": 8260
    },
    {
      "epoch": 1.131333881128458,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008288859044082121,
      "loss": 4.9905,
      "step": 8261
    },
    {
      "epoch": 1.131470829909614,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008288638009114874,
      "loss": 4.8459,
      "step": 8262
    },
    {
      "epoch": 1.1316077786907697,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008288416942750281,
      "loss": 4.9493,
      "step": 8263
    },
    {
      "epoch": 1.1317447274719254,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008288195844990175,
      "loss": 4.9342,
      "step": 8264
    },
    {
      "epoch": 1.1318816762530814,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008287974715836387,
      "loss": 4.9565,
      "step": 8265
    },
    {
      "epoch": 1.1320186250342372,
      "grad_norm": 0.46875,
      "learning_rate": 0.000828775355529075,
      "loss": 4.8601,
      "step": 8266
    },
    {
      "epoch": 1.132155573815393,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008287532363355098,
      "loss": 4.8412,
      "step": 8267
    },
    {
      "epoch": 1.1322925225965488,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000828731114003126,
      "loss": 4.8495,
      "step": 8268
    },
    {
      "epoch": 1.1324294713777048,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008287089885321073,
      "loss": 4.9608,
      "step": 8269
    },
    {
      "epoch": 1.1325664201588606,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008286868599226371,
      "loss": 4.9962,
      "step": 8270
    },
    {
      "epoch": 1.1327033689400166,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008286647281748986,
      "loss": 4.9276,
      "step": 8271
    },
    {
      "epoch": 1.1328403177211723,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008286425932890753,
      "loss": 4.8983,
      "step": 8272
    },
    {
      "epoch": 1.132977266502328,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008286204552653506,
      "loss": 4.9144,
      "step": 8273
    },
    {
      "epoch": 1.1331142152834839,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008285983141039079,
      "loss": 4.9704,
      "step": 8274
    },
    {
      "epoch": 1.1332511640646399,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008285761698049308,
      "loss": 4.9492,
      "step": 8275
    },
    {
      "epoch": 1.1333881128457957,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008285540223686027,
      "loss": 4.8287,
      "step": 8276
    },
    {
      "epoch": 1.1335250616269514,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008285318717951073,
      "loss": 5.0031,
      "step": 8277
    },
    {
      "epoch": 1.1336620104081074,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008285097180846281,
      "loss": 4.952,
      "step": 8278
    },
    {
      "epoch": 1.1337989591892632,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008284875612373485,
      "loss": 4.8819,
      "step": 8279
    },
    {
      "epoch": 1.133935907970419,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008284654012534525,
      "loss": 4.9363,
      "step": 8280
    },
    {
      "epoch": 1.134072856751575,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008284432381331234,
      "loss": 4.916,
      "step": 8281
    },
    {
      "epoch": 1.1342098055327308,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000828421071876545,
      "loss": 4.9873,
      "step": 8282
    },
    {
      "epoch": 1.1343467543138865,
      "grad_norm": 0.390625,
      "learning_rate": 0.000828398902483901,
      "loss": 4.9439,
      "step": 8283
    },
    {
      "epoch": 1.1344837030950425,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008283767299553753,
      "loss": 4.9097,
      "step": 8284
    },
    {
      "epoch": 1.1346206518761983,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008283545542911513,
      "loss": 4.9707,
      "step": 8285
    },
    {
      "epoch": 1.134757600657354,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000828332375491413,
      "loss": 4.8655,
      "step": 8286
    },
    {
      "epoch": 1.13489454943851,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008283101935563441,
      "loss": 4.8834,
      "step": 8287
    },
    {
      "epoch": 1.1350314982196659,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008282880084861286,
      "loss": 4.9418,
      "step": 8288
    },
    {
      "epoch": 1.1351684470008216,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008282658202809502,
      "loss": 4.9419,
      "step": 8289
    },
    {
      "epoch": 1.1353053957819776,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008282436289409928,
      "loss": 4.8571,
      "step": 8290
    },
    {
      "epoch": 1.1354423445631334,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008282214344664403,
      "loss": 4.8255,
      "step": 8291
    },
    {
      "epoch": 1.1355792933442892,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008281992368574766,
      "loss": 4.905,
      "step": 8292
    },
    {
      "epoch": 1.135716242125445,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008281770361142859,
      "loss": 4.9255,
      "step": 8293
    },
    {
      "epoch": 1.135853190906601,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008281548322370517,
      "loss": 4.9228,
      "step": 8294
    },
    {
      "epoch": 1.1359901396877568,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008281326252259583,
      "loss": 4.9518,
      "step": 8295
    },
    {
      "epoch": 1.1361270884689127,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008281104150811901,
      "loss": 4.8587,
      "step": 8296
    },
    {
      "epoch": 1.1362640372500685,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008280882018029304,
      "loss": 4.9486,
      "step": 8297
    },
    {
      "epoch": 1.1364009860312243,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008280659853913639,
      "loss": 4.9204,
      "step": 8298
    },
    {
      "epoch": 1.13653793481238,
      "grad_norm": 0.375,
      "learning_rate": 0.0008280437658466745,
      "loss": 4.8597,
      "step": 8299
    },
    {
      "epoch": 1.136674883593536,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008280215431690462,
      "loss": 4.9287,
      "step": 8300
    },
    {
      "epoch": 1.1368118323746919,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008279993173586634,
      "loss": 4.8404,
      "step": 8301
    },
    {
      "epoch": 1.1369487811558476,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008279770884157101,
      "loss": 4.9408,
      "step": 8302
    },
    {
      "epoch": 1.1370857299370036,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008279548563403707,
      "loss": 4.9545,
      "step": 8303
    },
    {
      "epoch": 1.1372226787181594,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008279326211328293,
      "loss": 4.9391,
      "step": 8304
    },
    {
      "epoch": 1.1373596274993152,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00082791038279327,
      "loss": 4.8653,
      "step": 8305
    },
    {
      "epoch": 1.1374965762804712,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008278881413218775,
      "loss": 4.9649,
      "step": 8306
    },
    {
      "epoch": 1.137633525061627,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000827865896718836,
      "loss": 4.8872,
      "step": 8307
    },
    {
      "epoch": 1.1377704738427827,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008278436489843298,
      "loss": 4.8727,
      "step": 8308
    },
    {
      "epoch": 1.1379074226239387,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008278213981185431,
      "loss": 4.8134,
      "step": 8309
    },
    {
      "epoch": 1.1380443714050945,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008277991441216604,
      "loss": 4.8729,
      "step": 8310
    },
    {
      "epoch": 1.1381813201862503,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008277768869938663,
      "loss": 4.9445,
      "step": 8311
    },
    {
      "epoch": 1.138318268967406,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008277546267353451,
      "loss": 4.8986,
      "step": 8312
    },
    {
      "epoch": 1.138455217748562,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008277323633462812,
      "loss": 4.9674,
      "step": 8313
    },
    {
      "epoch": 1.1385921665297178,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008277100968268593,
      "loss": 4.9337,
      "step": 8314
    },
    {
      "epoch": 1.1387291153108738,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008276878271772638,
      "loss": 4.9444,
      "step": 8315
    },
    {
      "epoch": 1.1388660640920296,
      "grad_norm": 0.375,
      "learning_rate": 0.0008276655543976792,
      "loss": 5.0525,
      "step": 8316
    },
    {
      "epoch": 1.1390030128731854,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008276432784882903,
      "loss": 4.9437,
      "step": 8317
    },
    {
      "epoch": 1.1391399616543412,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008276209994492815,
      "loss": 4.8974,
      "step": 8318
    },
    {
      "epoch": 1.1392769104354972,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008275987172808375,
      "loss": 4.9566,
      "step": 8319
    },
    {
      "epoch": 1.139413859216653,
      "grad_norm": 0.421875,
      "learning_rate": 0.000827576431983143,
      "loss": 4.9067,
      "step": 8320
    },
    {
      "epoch": 1.1395508079978087,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008275541435563825,
      "loss": 4.9556,
      "step": 8321
    },
    {
      "epoch": 1.1396877567789647,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008275318520007408,
      "loss": 4.8746,
      "step": 8322
    },
    {
      "epoch": 1.1398247055601205,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008275095573164029,
      "loss": 4.8903,
      "step": 8323
    },
    {
      "epoch": 1.1399616543412763,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008274872595035532,
      "loss": 4.939,
      "step": 8324
    },
    {
      "epoch": 1.1400986031224323,
      "grad_norm": 0.5,
      "learning_rate": 0.0008274649585623767,
      "loss": 4.9237,
      "step": 8325
    },
    {
      "epoch": 1.140235551903588,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008274426544930582,
      "loss": 4.9326,
      "step": 8326
    },
    {
      "epoch": 1.1403725006847438,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008274203472957823,
      "loss": 4.8637,
      "step": 8327
    },
    {
      "epoch": 1.1405094494658998,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008273980369707342,
      "loss": 4.8739,
      "step": 8328
    },
    {
      "epoch": 1.1406463982470556,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008273757235180986,
      "loss": 4.8486,
      "step": 8329
    },
    {
      "epoch": 1.1407833470282114,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008273534069380605,
      "loss": 4.904,
      "step": 8330
    },
    {
      "epoch": 1.1409202958093674,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008273310872308047,
      "loss": 4.9393,
      "step": 8331
    },
    {
      "epoch": 1.1410572445905232,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008273087643965163,
      "loss": 4.8811,
      "step": 8332
    },
    {
      "epoch": 1.141194193371679,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008272864384353802,
      "loss": 4.8536,
      "step": 8333
    },
    {
      "epoch": 1.141331142152835,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008272641093475815,
      "loss": 4.8549,
      "step": 8334
    },
    {
      "epoch": 1.1414680909339907,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008272417771333055,
      "loss": 4.8967,
      "step": 8335
    },
    {
      "epoch": 1.1416050397151465,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008272194417927366,
      "loss": 4.9373,
      "step": 8336
    },
    {
      "epoch": 1.1417419884963023,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008271971033260606,
      "loss": 4.9304,
      "step": 8337
    },
    {
      "epoch": 1.1418789372774583,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008271747617334622,
      "loss": 4.9441,
      "step": 8338
    },
    {
      "epoch": 1.142015886058614,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008271524170151266,
      "loss": 4.8828,
      "step": 8339
    },
    {
      "epoch": 1.14215283483977,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008271300691712392,
      "loss": 4.9275,
      "step": 8340
    },
    {
      "epoch": 1.1422897836209258,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000827107718201985,
      "loss": 4.9373,
      "step": 8341
    },
    {
      "epoch": 1.1424267324020816,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008270853641075492,
      "loss": 4.8659,
      "step": 8342
    },
    {
      "epoch": 1.1425636811832374,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008270630068881172,
      "loss": 4.8872,
      "step": 8343
    },
    {
      "epoch": 1.1427006299643934,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008270406465438742,
      "loss": 4.888,
      "step": 8344
    },
    {
      "epoch": 1.1428375787455491,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008270182830750055,
      "loss": 4.944,
      "step": 8345
    },
    {
      "epoch": 1.142974527526705,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008269959164816965,
      "loss": 4.8864,
      "step": 8346
    },
    {
      "epoch": 1.143111476307861,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008269735467641325,
      "loss": 4.9863,
      "step": 8347
    },
    {
      "epoch": 1.1432484250890167,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008269511739224988,
      "loss": 4.924,
      "step": 8348
    },
    {
      "epoch": 1.1433853738701725,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000826928797956981,
      "loss": 4.9452,
      "step": 8349
    },
    {
      "epoch": 1.1435223226513285,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0008269064188677644,
      "loss": 4.9585,
      "step": 8350
    },
    {
      "epoch": 1.1436592714324842,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008268840366550345,
      "loss": 4.919,
      "step": 8351
    },
    {
      "epoch": 1.14379622021364,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008268616513189767,
      "loss": 4.8722,
      "step": 8352
    },
    {
      "epoch": 1.143933168994796,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008268392628597767,
      "loss": 4.9172,
      "step": 8353
    },
    {
      "epoch": 1.1440701177759518,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00082681687127762,
      "loss": 4.8667,
      "step": 8354
    },
    {
      "epoch": 1.1442070665571076,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000826794476572692,
      "loss": 4.8874,
      "step": 8355
    },
    {
      "epoch": 1.1443440153382636,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008267720787451785,
      "loss": 4.9213,
      "step": 8356
    },
    {
      "epoch": 1.1444809641194194,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008267496777952648,
      "loss": 4.8953,
      "step": 8357
    },
    {
      "epoch": 1.1446179129005751,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008267272737231368,
      "loss": 4.9262,
      "step": 8358
    },
    {
      "epoch": 1.1447548616817311,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008267048665289802,
      "loss": 4.9438,
      "step": 8359
    },
    {
      "epoch": 1.144891810462887,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008266824562129805,
      "loss": 4.937,
      "step": 8360
    },
    {
      "epoch": 1.1450287592440427,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008266600427753237,
      "loss": 4.9083,
      "step": 8361
    },
    {
      "epoch": 1.1451657080251985,
      "grad_norm": 0.375,
      "learning_rate": 0.0008266376262161952,
      "loss": 4.9756,
      "step": 8362
    },
    {
      "epoch": 1.1453026568063545,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008266152065357809,
      "loss": 4.8727,
      "step": 8363
    },
    {
      "epoch": 1.1454396055875102,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008265927837342667,
      "loss": 4.935,
      "step": 8364
    },
    {
      "epoch": 1.1455765543686662,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008265703578118385,
      "loss": 4.9604,
      "step": 8365
    },
    {
      "epoch": 1.145713503149822,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008265479287686819,
      "loss": 4.922,
      "step": 8366
    },
    {
      "epoch": 1.1458504519309778,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000826525496604983,
      "loss": 4.9216,
      "step": 8367
    },
    {
      "epoch": 1.1459874007121336,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008265030613209274,
      "loss": 4.8882,
      "step": 8368
    },
    {
      "epoch": 1.1461243494932896,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008264806229167014,
      "loss": 4.9783,
      "step": 8369
    },
    {
      "epoch": 1.1462612982744453,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008264581813924906,
      "loss": 4.874,
      "step": 8370
    },
    {
      "epoch": 1.1463982470556011,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008264357367484813,
      "loss": 4.9159,
      "step": 8371
    },
    {
      "epoch": 1.1465351958367571,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008264132889848593,
      "loss": 4.9228,
      "step": 8372
    },
    {
      "epoch": 1.146672144617913,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008263908381018106,
      "loss": 4.8757,
      "step": 8373
    },
    {
      "epoch": 1.1468090933990687,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008263683840995214,
      "loss": 4.9272,
      "step": 8374
    },
    {
      "epoch": 1.1469460421802247,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008263459269781777,
      "loss": 4.9389,
      "step": 8375
    },
    {
      "epoch": 1.1470829909613804,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008263234667379656,
      "loss": 4.9747,
      "step": 8376
    },
    {
      "epoch": 1.1472199397425362,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008263010033790714,
      "loss": 4.9908,
      "step": 8377
    },
    {
      "epoch": 1.1473568885236922,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008262785369016809,
      "loss": 4.9893,
      "step": 8378
    },
    {
      "epoch": 1.147493837304848,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008262560673059807,
      "loss": 4.951,
      "step": 8379
    },
    {
      "epoch": 1.1476307860860038,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008262335945921567,
      "loss": 4.9503,
      "step": 8380
    },
    {
      "epoch": 1.1477677348671598,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008262111187603953,
      "loss": 4.9633,
      "step": 8381
    },
    {
      "epoch": 1.1479046836483155,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008261886398108827,
      "loss": 4.8965,
      "step": 8382
    },
    {
      "epoch": 1.1480416324294713,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008261661577438051,
      "loss": 4.8543,
      "step": 8383
    },
    {
      "epoch": 1.1481785812106273,
      "grad_norm": 0.359375,
      "learning_rate": 0.000826143672559349,
      "loss": 4.9652,
      "step": 8384
    },
    {
      "epoch": 1.148315529991783,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008261211842577007,
      "loss": 4.8313,
      "step": 8385
    },
    {
      "epoch": 1.1484524787729389,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008260986928390465,
      "loss": 4.9741,
      "step": 8386
    },
    {
      "epoch": 1.1485894275540947,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008260761983035728,
      "loss": 4.8684,
      "step": 8387
    },
    {
      "epoch": 1.1487263763352507,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000826053700651466,
      "loss": 4.895,
      "step": 8388
    },
    {
      "epoch": 1.1488633251164064,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008260311998829126,
      "loss": 4.8619,
      "step": 8389
    },
    {
      "epoch": 1.1490002738975624,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000826008695998099,
      "loss": 4.8905,
      "step": 8390
    },
    {
      "epoch": 1.1491372226787182,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008259861889972118,
      "loss": 4.9164,
      "step": 8391
    },
    {
      "epoch": 1.149274171459874,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008259636788804374,
      "loss": 4.9745,
      "step": 8392
    },
    {
      "epoch": 1.1494111202410298,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008259411656479625,
      "loss": 4.8977,
      "step": 8393
    },
    {
      "epoch": 1.1495480690221858,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008259186492999734,
      "loss": 4.8951,
      "step": 8394
    },
    {
      "epoch": 1.1496850178033415,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008258961298366572,
      "loss": 4.7961,
      "step": 8395
    },
    {
      "epoch": 1.1498219665844973,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008258736072581999,
      "loss": 4.902,
      "step": 8396
    },
    {
      "epoch": 1.1499589153656533,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008258510815647885,
      "loss": 4.9109,
      "step": 8397
    },
    {
      "epoch": 1.150095864146809,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008258285527566097,
      "loss": 5.0086,
      "step": 8398
    },
    {
      "epoch": 1.1502328129279649,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008258060208338502,
      "loss": 4.8973,
      "step": 8399
    },
    {
      "epoch": 1.1503697617091209,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008257834857966965,
      "loss": 4.8983,
      "step": 8400
    },
    {
      "epoch": 1.1505067104902766,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008257609476453356,
      "loss": 4.8253,
      "step": 8401
    },
    {
      "epoch": 1.1506436592714324,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008257384063799541,
      "loss": 4.8831,
      "step": 8402
    },
    {
      "epoch": 1.1507806080525884,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000825715862000739,
      "loss": 4.8863,
      "step": 8403
    },
    {
      "epoch": 1.1509175568337442,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000825693314507877,
      "loss": 4.8854,
      "step": 8404
    },
    {
      "epoch": 1.1510545056149,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000825670763901555,
      "loss": 4.8758,
      "step": 8405
    },
    {
      "epoch": 1.151191454396056,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008256482101819597,
      "loss": 4.9523,
      "step": 8406
    },
    {
      "epoch": 1.1513284031772117,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008256256533492784,
      "loss": 4.8461,
      "step": 8407
    },
    {
      "epoch": 1.1514653519583675,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008256030934036976,
      "loss": 4.9915,
      "step": 8408
    },
    {
      "epoch": 1.1516023007395235,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008255805303454047,
      "loss": 4.8288,
      "step": 8409
    },
    {
      "epoch": 1.1517392495206793,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008255579641745862,
      "loss": 4.8939,
      "step": 8410
    },
    {
      "epoch": 1.151876198301835,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008255353948914295,
      "loss": 4.9615,
      "step": 8411
    },
    {
      "epoch": 1.1520131470829909,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008255128224961214,
      "loss": 4.8731,
      "step": 8412
    },
    {
      "epoch": 1.1521500958641469,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008254902469888491,
      "loss": 4.8937,
      "step": 8413
    },
    {
      "epoch": 1.1522870446453026,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008254676683697997,
      "loss": 4.9357,
      "step": 8414
    },
    {
      "epoch": 1.1524239934264586,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008254450866391602,
      "loss": 4.898,
      "step": 8415
    },
    {
      "epoch": 1.1525609422076144,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008254225017971179,
      "loss": 5.0068,
      "step": 8416
    },
    {
      "epoch": 1.1526978909887702,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008253999138438597,
      "loss": 4.8529,
      "step": 8417
    },
    {
      "epoch": 1.152834839769926,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008253773227795732,
      "loss": 4.9147,
      "step": 8418
    },
    {
      "epoch": 1.152971788551082,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008253547286044452,
      "loss": 4.886,
      "step": 8419
    },
    {
      "epoch": 1.1531087373322377,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008253321313186632,
      "loss": 4.9776,
      "step": 8420
    },
    {
      "epoch": 1.1532456861133935,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008253095309224143,
      "loss": 4.9539,
      "step": 8421
    },
    {
      "epoch": 1.1533826348945495,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000825286927415886,
      "loss": 4.9183,
      "step": 8422
    },
    {
      "epoch": 1.1535195836757053,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008252643207992654,
      "loss": 4.886,
      "step": 8423
    },
    {
      "epoch": 1.153656532456861,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00082524171107274,
      "loss": 4.9083,
      "step": 8424
    },
    {
      "epoch": 1.153793481238017,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008252190982364973,
      "loss": 4.9031,
      "step": 8425
    },
    {
      "epoch": 1.1539304300191728,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008251964822907242,
      "loss": 4.9065,
      "step": 8426
    },
    {
      "epoch": 1.1540673788003286,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008251738632356085,
      "loss": 4.9423,
      "step": 8427
    },
    {
      "epoch": 1.1542043275814846,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008251512410713377,
      "loss": 4.9275,
      "step": 8428
    },
    {
      "epoch": 1.1543412763626404,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008251286157980992,
      "loss": 4.938,
      "step": 8429
    },
    {
      "epoch": 1.1544782251437962,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008251059874160805,
      "loss": 4.8973,
      "step": 8430
    },
    {
      "epoch": 1.1546151739249522,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008250833559254691,
      "loss": 4.8682,
      "step": 8431
    },
    {
      "epoch": 1.154752122706108,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008250607213264524,
      "loss": 4.9172,
      "step": 8432
    },
    {
      "epoch": 1.1548890714872637,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008250380836192183,
      "loss": 4.9217,
      "step": 8433
    },
    {
      "epoch": 1.1550260202684197,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000825015442803954,
      "loss": 4.9143,
      "step": 8434
    },
    {
      "epoch": 1.1551629690495755,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008249927988808476,
      "loss": 4.9092,
      "step": 8435
    },
    {
      "epoch": 1.1552999178307313,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008249701518500865,
      "loss": 5.0095,
      "step": 8436
    },
    {
      "epoch": 1.155436866611887,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008249475017118584,
      "loss": 4.9723,
      "step": 8437
    },
    {
      "epoch": 1.155573815393043,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008249248484663508,
      "loss": 4.9175,
      "step": 8438
    },
    {
      "epoch": 1.1557107641741988,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008249021921137519,
      "loss": 4.905,
      "step": 8439
    },
    {
      "epoch": 1.1558477129553546,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000824879532654249,
      "loss": 4.9164,
      "step": 8440
    },
    {
      "epoch": 1.1559846617365106,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008248568700880302,
      "loss": 4.8473,
      "step": 8441
    },
    {
      "epoch": 1.1561216105176664,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008248342044152832,
      "loss": 4.9078,
      "step": 8442
    },
    {
      "epoch": 1.1562585592988222,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008248115356361958,
      "loss": 4.9743,
      "step": 8443
    },
    {
      "epoch": 1.1563955080799782,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008247888637509558,
      "loss": 4.94,
      "step": 8444
    },
    {
      "epoch": 1.156532456861134,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008247661887597513,
      "loss": 4.8738,
      "step": 8445
    },
    {
      "epoch": 1.1566694056422897,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008247435106627701,
      "loss": 4.8554,
      "step": 8446
    },
    {
      "epoch": 1.1568063544234457,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008247208294602,
      "loss": 4.9629,
      "step": 8447
    },
    {
      "epoch": 1.1569433032046015,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008246981451522292,
      "loss": 4.9005,
      "step": 8448
    },
    {
      "epoch": 1.1570802519857573,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008246754577390455,
      "loss": 4.9243,
      "step": 8449
    },
    {
      "epoch": 1.1572172007669133,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008246527672208373,
      "loss": 4.9589,
      "step": 8450
    },
    {
      "epoch": 1.157354149548069,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000824630073597792,
      "loss": 4.9252,
      "step": 8451
    },
    {
      "epoch": 1.1574910983292248,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000824607376870098,
      "loss": 4.9565,
      "step": 8452
    },
    {
      "epoch": 1.1576280471103808,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008245846770379437,
      "loss": 4.9396,
      "step": 8453
    },
    {
      "epoch": 1.1577649958915366,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008245619741015167,
      "loss": 4.9175,
      "step": 8454
    },
    {
      "epoch": 1.1579019446726924,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008245392680610055,
      "loss": 4.8773,
      "step": 8455
    },
    {
      "epoch": 1.1580388934538481,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008245165589165979,
      "loss": 4.9827,
      "step": 8456
    },
    {
      "epoch": 1.1581758422350041,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008244938466684825,
      "loss": 4.9284,
      "step": 8457
    },
    {
      "epoch": 1.15831279101616,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008244711313168474,
      "loss": 4.9913,
      "step": 8458
    },
    {
      "epoch": 1.158449739797316,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008244484128618807,
      "loss": 4.9883,
      "step": 8459
    },
    {
      "epoch": 1.1585866885784717,
      "grad_norm": 0.375,
      "learning_rate": 0.0008244256913037709,
      "loss": 4.9491,
      "step": 8460
    },
    {
      "epoch": 1.1587236373596275,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000824402966642706,
      "loss": 4.8168,
      "step": 8461
    },
    {
      "epoch": 1.1588605861407832,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008243802388788747,
      "loss": 4.8843,
      "step": 8462
    },
    {
      "epoch": 1.1589975349219392,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008243575080124651,
      "loss": 4.904,
      "step": 8463
    },
    {
      "epoch": 1.159134483703095,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008243347740436655,
      "loss": 4.8739,
      "step": 8464
    },
    {
      "epoch": 1.1592714324842508,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008243120369726644,
      "loss": 4.9104,
      "step": 8465
    },
    {
      "epoch": 1.1594083812654068,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008242892967996505,
      "loss": 4.882,
      "step": 8466
    },
    {
      "epoch": 1.1595453300465626,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0008242665535248119,
      "loss": 4.899,
      "step": 8467
    },
    {
      "epoch": 1.1596822788277183,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008242438071483371,
      "loss": 4.9866,
      "step": 8468
    },
    {
      "epoch": 1.1598192276088743,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008242210576704148,
      "loss": 4.8621,
      "step": 8469
    },
    {
      "epoch": 1.1599561763900301,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008241983050912333,
      "loss": 4.9735,
      "step": 8470
    },
    {
      "epoch": 1.160093125171186,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008241755494109814,
      "loss": 4.9429,
      "step": 8471
    },
    {
      "epoch": 1.160230073952342,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008241527906298475,
      "loss": 4.9443,
      "step": 8472
    },
    {
      "epoch": 1.1603670227334977,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008241300287480203,
      "loss": 4.8231,
      "step": 8473
    },
    {
      "epoch": 1.1605039715146535,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008241072637656882,
      "loss": 4.8865,
      "step": 8474
    },
    {
      "epoch": 1.1606409202958095,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008240844956830405,
      "loss": 4.9725,
      "step": 8475
    },
    {
      "epoch": 1.1607778690769652,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008240617245002651,
      "loss": 4.9114,
      "step": 8476
    },
    {
      "epoch": 1.160914817858121,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008240389502175512,
      "loss": 4.8882,
      "step": 8477
    },
    {
      "epoch": 1.161051766639277,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008240161728350873,
      "loss": 4.9076,
      "step": 8478
    },
    {
      "epoch": 1.1611887154204328,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008239933923530623,
      "loss": 4.8299,
      "step": 8479
    },
    {
      "epoch": 1.1613256642015886,
      "grad_norm": 0.40625,
      "learning_rate": 0.000823970608771665,
      "loss": 4.9104,
      "step": 8480
    },
    {
      "epoch": 1.1614626129827443,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008239478220910839,
      "loss": 4.8883,
      "step": 8481
    },
    {
      "epoch": 1.1615995617639003,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008239250323115084,
      "loss": 4.8538,
      "step": 8482
    },
    {
      "epoch": 1.1617365105450561,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008239022394331269,
      "loss": 4.8689,
      "step": 8483
    },
    {
      "epoch": 1.161873459326212,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008238794434561285,
      "loss": 4.9194,
      "step": 8484
    },
    {
      "epoch": 1.1620104081073679,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000823856644380702,
      "loss": 4.9107,
      "step": 8485
    },
    {
      "epoch": 1.1621473568885237,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008238338422070364,
      "loss": 4.8489,
      "step": 8486
    },
    {
      "epoch": 1.1622843056696794,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008238110369353207,
      "loss": 4.8843,
      "step": 8487
    },
    {
      "epoch": 1.1624212544508354,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008237882285657439,
      "loss": 4.9501,
      "step": 8488
    },
    {
      "epoch": 1.1625582032319912,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000823765417098495,
      "loss": 4.9324,
      "step": 8489
    },
    {
      "epoch": 1.162695152013147,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008237426025337629,
      "loss": 4.9988,
      "step": 8490
    },
    {
      "epoch": 1.162832100794303,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008237197848717368,
      "loss": 4.931,
      "step": 8491
    },
    {
      "epoch": 1.1629690495754588,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008236969641126059,
      "loss": 4.8862,
      "step": 8492
    },
    {
      "epoch": 1.1631059983566145,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008236741402565592,
      "loss": 4.8587,
      "step": 8493
    },
    {
      "epoch": 1.1632429471377705,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008236513133037859,
      "loss": 4.9618,
      "step": 8494
    },
    {
      "epoch": 1.1633798959189263,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000823628483254475,
      "loss": 4.8447,
      "step": 8495
    },
    {
      "epoch": 1.163516844700082,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008236056501088159,
      "loss": 4.9046,
      "step": 8496
    },
    {
      "epoch": 1.163653793481238,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008235828138669978,
      "loss": 4.9553,
      "step": 8497
    },
    {
      "epoch": 1.1637907422623939,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008235599745292098,
      "loss": 4.9113,
      "step": 8498
    },
    {
      "epoch": 1.1639276910435497,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008235371320956412,
      "loss": 4.9662,
      "step": 8499
    },
    {
      "epoch": 1.1640646398247056,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008235142865664816,
      "loss": 4.9415,
      "step": 8500
    },
    {
      "epoch": 1.1642015886058614,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008234914379419201,
      "loss": 4.8948,
      "step": 8501
    },
    {
      "epoch": 1.1643385373870172,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008234685862221459,
      "loss": 4.9793,
      "step": 8502
    },
    {
      "epoch": 1.1644754861681732,
      "grad_norm": 0.375,
      "learning_rate": 0.0008234457314073485,
      "loss": 4.8941,
      "step": 8503
    },
    {
      "epoch": 1.164612434949329,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008234228734977175,
      "loss": 4.947,
      "step": 8504
    },
    {
      "epoch": 1.1647493837304848,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000823400012493442,
      "loss": 4.9181,
      "step": 8505
    },
    {
      "epoch": 1.1648863325116405,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008233771483947118,
      "loss": 4.9088,
      "step": 8506
    },
    {
      "epoch": 1.1650232812927965,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008233542812017162,
      "loss": 4.9488,
      "step": 8507
    },
    {
      "epoch": 1.1651602300739523,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008233314109146446,
      "loss": 4.9674,
      "step": 8508
    },
    {
      "epoch": 1.1652971788551083,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008233085375336866,
      "loss": 4.8711,
      "step": 8509
    },
    {
      "epoch": 1.165434127636264,
      "grad_norm": 0.375,
      "learning_rate": 0.0008232856610590319,
      "loss": 4.9241,
      "step": 8510
    },
    {
      "epoch": 1.1655710764174199,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008232627814908699,
      "loss": 4.852,
      "step": 8511
    },
    {
      "epoch": 1.1657080251985756,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008232398988293902,
      "loss": 4.9096,
      "step": 8512
    },
    {
      "epoch": 1.1658449739797316,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008232170130747827,
      "loss": 4.9124,
      "step": 8513
    },
    {
      "epoch": 1.1659819227608874,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008231941242272369,
      "loss": 4.938,
      "step": 8514
    },
    {
      "epoch": 1.1661188715420432,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008231712322869423,
      "loss": 4.9328,
      "step": 8515
    },
    {
      "epoch": 1.1662558203231992,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008231483372540889,
      "loss": 4.854,
      "step": 8516
    },
    {
      "epoch": 1.166392769104355,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008231254391288663,
      "loss": 4.8979,
      "step": 8517
    },
    {
      "epoch": 1.1665297178855107,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008231025379114642,
      "loss": 4.8655,
      "step": 8518
    },
    {
      "epoch": 1.1666666666666667,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008230796336020725,
      "loss": 4.8375,
      "step": 8519
    },
    {
      "epoch": 1.1668036154478225,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000823056726200881,
      "loss": 4.8243,
      "step": 8520
    },
    {
      "epoch": 1.1669405642289783,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008230338157080795,
      "loss": 4.911,
      "step": 8521
    },
    {
      "epoch": 1.1670775130101343,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008230109021238578,
      "loss": 4.9569,
      "step": 8522
    },
    {
      "epoch": 1.16721446179129,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000822987985448406,
      "loss": 4.8892,
      "step": 8523
    },
    {
      "epoch": 1.1673514105724458,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008229650656819138,
      "loss": 4.9336,
      "step": 8524
    },
    {
      "epoch": 1.1674883593536018,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008229421428245713,
      "loss": 4.9214,
      "step": 8525
    },
    {
      "epoch": 1.1676253081347576,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008229192168765684,
      "loss": 4.8302,
      "step": 8526
    },
    {
      "epoch": 1.1677622569159134,
      "grad_norm": 0.390625,
      "learning_rate": 0.000822896287838095,
      "loss": 4.8813,
      "step": 8527
    },
    {
      "epoch": 1.1678992056970694,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008228733557093413,
      "loss": 4.9,
      "step": 8528
    },
    {
      "epoch": 1.1680361544782252,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008228504204904972,
      "loss": 4.9312,
      "step": 8529
    },
    {
      "epoch": 1.168173103259381,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008228274821817527,
      "loss": 4.8449,
      "step": 8530
    },
    {
      "epoch": 1.1683100520405367,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008228045407832981,
      "loss": 4.9597,
      "step": 8531
    },
    {
      "epoch": 1.1684470008216927,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008227815962953236,
      "loss": 4.9585,
      "step": 8532
    },
    {
      "epoch": 1.1685839496028485,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000822758648718019,
      "loss": 4.8765,
      "step": 8533
    },
    {
      "epoch": 1.1687208983840045,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008227356980515748,
      "loss": 4.8919,
      "step": 8534
    },
    {
      "epoch": 1.1688578471651603,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000822712744296181,
      "loss": 4.8445,
      "step": 8535
    },
    {
      "epoch": 1.168994795946316,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008226897874520278,
      "loss": 4.8717,
      "step": 8536
    },
    {
      "epoch": 1.1691317447274718,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008226668275193056,
      "loss": 4.916,
      "step": 8537
    },
    {
      "epoch": 1.1692686935086278,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008226438644982045,
      "loss": 4.9528,
      "step": 8538
    },
    {
      "epoch": 1.1694056422897836,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008226208983889149,
      "loss": 4.9199,
      "step": 8539
    },
    {
      "epoch": 1.1695425910709394,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008225979291916272,
      "loss": 4.8573,
      "step": 8540
    },
    {
      "epoch": 1.1696795398520954,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008225749569065317,
      "loss": 4.855,
      "step": 8541
    },
    {
      "epoch": 1.1698164886332512,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008225519815338187,
      "loss": 4.8867,
      "step": 8542
    },
    {
      "epoch": 1.169953437414407,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008225290030736788,
      "loss": 4.9339,
      "step": 8543
    },
    {
      "epoch": 1.170090386195563,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000822506021526302,
      "loss": 4.9617,
      "step": 8544
    },
    {
      "epoch": 1.1702273349767187,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008224830368918792,
      "loss": 4.8849,
      "step": 8545
    },
    {
      "epoch": 1.1703642837578745,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008224600491706008,
      "loss": 4.8439,
      "step": 8546
    },
    {
      "epoch": 1.1705012325390305,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008224370583626571,
      "loss": 4.8636,
      "step": 8547
    },
    {
      "epoch": 1.1706381813201863,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008224140644682387,
      "loss": 4.8815,
      "step": 8548
    },
    {
      "epoch": 1.170775130101342,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008223910674875363,
      "loss": 4.8658,
      "step": 8549
    },
    {
      "epoch": 1.170912078882498,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008223680674207404,
      "loss": 4.9025,
      "step": 8550
    },
    {
      "epoch": 1.1710490276636538,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008223450642680416,
      "loss": 4.9758,
      "step": 8551
    },
    {
      "epoch": 1.1711859764448096,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008223220580296304,
      "loss": 4.9295,
      "step": 8552
    },
    {
      "epoch": 1.1713229252259656,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008222990487056978,
      "loss": 4.9059,
      "step": 8553
    },
    {
      "epoch": 1.1714598740071214,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000822276036296434,
      "loss": 4.839,
      "step": 8554
    },
    {
      "epoch": 1.1715968227882771,
      "grad_norm": 0.375,
      "learning_rate": 0.0008222530208020301,
      "loss": 4.9356,
      "step": 8555
    },
    {
      "epoch": 1.171733771569433,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008222300022226768,
      "loss": 4.8161,
      "step": 8556
    },
    {
      "epoch": 1.171870720350589,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008222069805585647,
      "loss": 4.9538,
      "step": 8557
    },
    {
      "epoch": 1.1720076691317447,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008221839558098846,
      "loss": 4.8843,
      "step": 8558
    },
    {
      "epoch": 1.1721446179129007,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008221609279768276,
      "loss": 4.9527,
      "step": 8559
    },
    {
      "epoch": 1.1722815666940565,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008221378970595841,
      "loss": 4.8809,
      "step": 8560
    },
    {
      "epoch": 1.1724185154752123,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008221148630583453,
      "loss": 4.8831,
      "step": 8561
    },
    {
      "epoch": 1.172555464256368,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008220918259733017,
      "loss": 4.9531,
      "step": 8562
    },
    {
      "epoch": 1.172692413037524,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008220687858046448,
      "loss": 4.8696,
      "step": 8563
    },
    {
      "epoch": 1.1728293618186798,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000822045742552565,
      "loss": 4.9246,
      "step": 8564
    },
    {
      "epoch": 1.1729663105998356,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008220226962172536,
      "loss": 4.9722,
      "step": 8565
    },
    {
      "epoch": 1.1731032593809916,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008219996467989015,
      "loss": 4.8436,
      "step": 8566
    },
    {
      "epoch": 1.1732402081621474,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008219765942976995,
      "loss": 4.8673,
      "step": 8567
    },
    {
      "epoch": 1.1733771569433031,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008219535387138388,
      "loss": 4.8288,
      "step": 8568
    },
    {
      "epoch": 1.1735141057244591,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008219304800475107,
      "loss": 4.9377,
      "step": 8569
    },
    {
      "epoch": 1.173651054505615,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000821907418298906,
      "loss": 4.9671,
      "step": 8570
    },
    {
      "epoch": 1.1737880032867707,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008218843534682159,
      "loss": 4.89,
      "step": 8571
    },
    {
      "epoch": 1.1739249520679267,
      "grad_norm": 0.375,
      "learning_rate": 0.0008218612855556316,
      "loss": 4.9407,
      "step": 8572
    },
    {
      "epoch": 1.1740619008490825,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008218382145613443,
      "loss": 4.9169,
      "step": 8573
    },
    {
      "epoch": 1.1741988496302382,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008218151404855448,
      "loss": 4.8876,
      "step": 8574
    },
    {
      "epoch": 1.174335798411394,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008217920633284248,
      "loss": 4.7958,
      "step": 8575
    },
    {
      "epoch": 1.17447274719255,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008217689830901754,
      "loss": 4.8613,
      "step": 8576
    },
    {
      "epoch": 1.1746096959737058,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008217458997709878,
      "loss": 4.8741,
      "step": 8577
    },
    {
      "epoch": 1.1747466447548618,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008217228133710534,
      "loss": 4.8993,
      "step": 8578
    },
    {
      "epoch": 1.1748835935360176,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008216997238905634,
      "loss": 4.9119,
      "step": 8579
    },
    {
      "epoch": 1.1750205423171733,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008216766313297093,
      "loss": 4.8575,
      "step": 8580
    },
    {
      "epoch": 1.1751574910983291,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008216535356886825,
      "loss": 4.9879,
      "step": 8581
    },
    {
      "epoch": 1.1752944398794851,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008216304369676741,
      "loss": 4.8099,
      "step": 8582
    },
    {
      "epoch": 1.175431388660641,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008216073351668758,
      "loss": 4.8908,
      "step": 8583
    },
    {
      "epoch": 1.1755683374417967,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008215842302864789,
      "loss": 4.9341,
      "step": 8584
    },
    {
      "epoch": 1.1757052862229527,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008215611223266749,
      "loss": 4.9308,
      "step": 8585
    },
    {
      "epoch": 1.1758422350041084,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008215380112876555,
      "loss": 4.9425,
      "step": 8586
    },
    {
      "epoch": 1.1759791837852642,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000821514897169612,
      "loss": 4.9442,
      "step": 8587
    },
    {
      "epoch": 1.1761161325664202,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008214917799727361,
      "loss": 4.9787,
      "step": 8588
    },
    {
      "epoch": 1.176253081347576,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008214686596972192,
      "loss": 4.887,
      "step": 8589
    },
    {
      "epoch": 1.1763900301287318,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000821445536343253,
      "loss": 4.7978,
      "step": 8590
    },
    {
      "epoch": 1.1765269789098878,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000821422409911029,
      "loss": 4.8464,
      "step": 8591
    },
    {
      "epoch": 1.1766639276910436,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008213992804007391,
      "loss": 4.9194,
      "step": 8592
    },
    {
      "epoch": 1.1768008764721993,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008213761478125749,
      "loss": 4.994,
      "step": 8593
    },
    {
      "epoch": 1.1769378252533553,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008213530121467281,
      "loss": 4.8567,
      "step": 8594
    },
    {
      "epoch": 1.177074774034511,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008213298734033903,
      "loss": 4.9195,
      "step": 8595
    },
    {
      "epoch": 1.1772117228156669,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008213067315827534,
      "loss": 4.887,
      "step": 8596
    },
    {
      "epoch": 1.1773486715968229,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008212835866850092,
      "loss": 4.9127,
      "step": 8597
    },
    {
      "epoch": 1.1774856203779787,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008212604387103494,
      "loss": 4.8674,
      "step": 8598
    },
    {
      "epoch": 1.1776225691591344,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008212372876589658,
      "loss": 4.8356,
      "step": 8599
    },
    {
      "epoch": 1.1777595179402902,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008212141335310505,
      "loss": 4.8697,
      "step": 8600
    },
    {
      "epoch": 1.1778964667214462,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008211909763267951,
      "loss": 4.8739,
      "step": 8601
    },
    {
      "epoch": 1.178033415502602,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008211678160463916,
      "loss": 4.942,
      "step": 8602
    },
    {
      "epoch": 1.178170364283758,
      "grad_norm": 0.375,
      "learning_rate": 0.000821144652690032,
      "loss": 4.9493,
      "step": 8603
    },
    {
      "epoch": 1.1783073130649138,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008211214862579081,
      "loss": 5.0218,
      "step": 8604
    },
    {
      "epoch": 1.1784442618460695,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008210983167502122,
      "loss": 4.9128,
      "step": 8605
    },
    {
      "epoch": 1.1785812106272253,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008210751441671361,
      "loss": 4.8976,
      "step": 8606
    },
    {
      "epoch": 1.1787181594083813,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008210519685088719,
      "loss": 4.8965,
      "step": 8607
    },
    {
      "epoch": 1.178855108189537,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008210287897756115,
      "loss": 4.8819,
      "step": 8608
    },
    {
      "epoch": 1.1789920569706929,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008210056079675472,
      "loss": 4.9455,
      "step": 8609
    },
    {
      "epoch": 1.1791290057518489,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008209824230848708,
      "loss": 4.8781,
      "step": 8610
    },
    {
      "epoch": 1.1792659545330046,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000820959235127775,
      "loss": 4.898,
      "step": 8611
    },
    {
      "epoch": 1.1794029033141604,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008209360440964514,
      "loss": 4.9558,
      "step": 8612
    },
    {
      "epoch": 1.1795398520953164,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008209128499910924,
      "loss": 4.9028,
      "step": 8613
    },
    {
      "epoch": 1.1796768008764722,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008208896528118904,
      "loss": 4.9031,
      "step": 8614
    },
    {
      "epoch": 1.179813749657628,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008208664525590374,
      "loss": 4.8669,
      "step": 8615
    },
    {
      "epoch": 1.179950698438784,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008208432492327256,
      "loss": 4.8811,
      "step": 8616
    },
    {
      "epoch": 1.1800876472199398,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008208200428331476,
      "loss": 4.8331,
      "step": 8617
    },
    {
      "epoch": 1.1802245960010955,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008207968333604955,
      "loss": 4.9309,
      "step": 8618
    },
    {
      "epoch": 1.1803615447822515,
      "grad_norm": 0.375,
      "learning_rate": 0.0008207736208149615,
      "loss": 4.9308,
      "step": 8619
    },
    {
      "epoch": 1.1804984935634073,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008207504051967384,
      "loss": 4.9263,
      "step": 8620
    },
    {
      "epoch": 1.180635442344563,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008207271865060182,
      "loss": 4.9291,
      "step": 8621
    },
    {
      "epoch": 1.180772391125719,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008207039647429936,
      "loss": 4.8456,
      "step": 8622
    },
    {
      "epoch": 1.1809093399068749,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008206807399078568,
      "loss": 4.8815,
      "step": 8623
    },
    {
      "epoch": 1.1810462886880306,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008206575120008004,
      "loss": 4.9174,
      "step": 8624
    },
    {
      "epoch": 1.1811832374691864,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008206342810220168,
      "loss": 4.9726,
      "step": 8625
    },
    {
      "epoch": 1.1813201862503424,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008206110469716988,
      "loss": 4.8864,
      "step": 8626
    },
    {
      "epoch": 1.1814571350314982,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008205878098500386,
      "loss": 4.9195,
      "step": 8627
    },
    {
      "epoch": 1.1815940838126542,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008205645696572289,
      "loss": 4.9191,
      "step": 8628
    },
    {
      "epoch": 1.18173103259381,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008205413263934624,
      "loss": 4.9076,
      "step": 8629
    },
    {
      "epoch": 1.1818679813749657,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008205180800589316,
      "loss": 4.8484,
      "step": 8630
    },
    {
      "epoch": 1.1820049301561215,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008204948306538292,
      "loss": 4.8972,
      "step": 8631
    },
    {
      "epoch": 1.1821418789372775,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008204715781783478,
      "loss": 4.9554,
      "step": 8632
    },
    {
      "epoch": 1.1822788277184333,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008204483226326802,
      "loss": 4.95,
      "step": 8633
    },
    {
      "epoch": 1.182415776499589,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008204250640170191,
      "loss": 4.9357,
      "step": 8634
    },
    {
      "epoch": 1.182552725280745,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008204018023315571,
      "loss": 4.879,
      "step": 8635
    },
    {
      "epoch": 1.1826896740619008,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008203785375764873,
      "loss": 4.9046,
      "step": 8636
    },
    {
      "epoch": 1.1828266228430566,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008203552697520021,
      "loss": 4.9238,
      "step": 8637
    },
    {
      "epoch": 1.1829635716242126,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008203319988582946,
      "loss": 4.8504,
      "step": 8638
    },
    {
      "epoch": 1.1831005204053684,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008203087248955577,
      "loss": 4.9414,
      "step": 8639
    },
    {
      "epoch": 1.1832374691865242,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000820285447863984,
      "loss": 4.9281,
      "step": 8640
    },
    {
      "epoch": 1.1833744179676802,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008202621677637665,
      "loss": 4.8956,
      "step": 8641
    },
    {
      "epoch": 1.183511366748836,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008202388845950984,
      "loss": 4.868,
      "step": 8642
    },
    {
      "epoch": 1.1836483155299917,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008202155983581723,
      "loss": 4.9746,
      "step": 8643
    },
    {
      "epoch": 1.1837852643111477,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008201923090531813,
      "loss": 4.9134,
      "step": 8644
    },
    {
      "epoch": 1.1839222130923035,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008201690166803185,
      "loss": 4.8041,
      "step": 8645
    },
    {
      "epoch": 1.1840591618734593,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008201457212397769,
      "loss": 4.841,
      "step": 8646
    },
    {
      "epoch": 1.1841961106546153,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0008201224227317495,
      "loss": 4.895,
      "step": 8647
    },
    {
      "epoch": 1.184333059435771,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008200991211564292,
      "loss": 4.9353,
      "step": 8648
    },
    {
      "epoch": 1.1844700082169268,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008200758165140095,
      "loss": 4.9536,
      "step": 8649
    },
    {
      "epoch": 1.1846069569980826,
      "grad_norm": 0.375,
      "learning_rate": 0.0008200525088046832,
      "loss": 4.931,
      "step": 8650
    },
    {
      "epoch": 1.1847439057792386,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008200291980286435,
      "loss": 4.8309,
      "step": 8651
    },
    {
      "epoch": 1.1848808545603944,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008200058841860839,
      "loss": 4.8993,
      "step": 8652
    },
    {
      "epoch": 1.1850178033415504,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008199825672771972,
      "loss": 4.8732,
      "step": 8653
    },
    {
      "epoch": 1.1851547521227062,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008199592473021769,
      "loss": 4.8817,
      "step": 8654
    },
    {
      "epoch": 1.185291700903862,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008199359242612162,
      "loss": 4.8456,
      "step": 8655
    },
    {
      "epoch": 1.1854286496850177,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008199125981545082,
      "loss": 4.8144,
      "step": 8656
    },
    {
      "epoch": 1.1855655984661737,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008198892689822464,
      "loss": 4.9178,
      "step": 8657
    },
    {
      "epoch": 1.1857025472473295,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008198659367446242,
      "loss": 4.9837,
      "step": 8658
    },
    {
      "epoch": 1.1858394960284853,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008198426014418347,
      "loss": 4.8648,
      "step": 8659
    },
    {
      "epoch": 1.1859764448096413,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008198192630740714,
      "loss": 4.8446,
      "step": 8660
    },
    {
      "epoch": 1.186113393590797,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008197959216415279,
      "loss": 4.9367,
      "step": 8661
    },
    {
      "epoch": 1.1862503423719528,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008197725771443975,
      "loss": 4.8301,
      "step": 8662
    },
    {
      "epoch": 1.1863872911531088,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008197492295828735,
      "loss": 4.938,
      "step": 8663
    },
    {
      "epoch": 1.1865242399342646,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008197258789571496,
      "loss": 4.8879,
      "step": 8664
    },
    {
      "epoch": 1.1866611887154204,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008197025252674191,
      "loss": 4.9125,
      "step": 8665
    },
    {
      "epoch": 1.1867981374965764,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008196791685138758,
      "loss": 4.9241,
      "step": 8666
    },
    {
      "epoch": 1.1869350862777321,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008196558086967132,
      "loss": 4.8453,
      "step": 8667
    },
    {
      "epoch": 1.187072035058888,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008196324458161246,
      "loss": 4.848,
      "step": 8668
    },
    {
      "epoch": 1.187208983840044,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008196090798723042,
      "loss": 4.9068,
      "step": 8669
    },
    {
      "epoch": 1.1873459326211997,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000819585710865445,
      "loss": 4.9168,
      "step": 8670
    },
    {
      "epoch": 1.1874828814023555,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000819562338795741,
      "loss": 4.9252,
      "step": 8671
    },
    {
      "epoch": 1.1876198301835115,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008195389636633859,
      "loss": 4.9313,
      "step": 8672
    },
    {
      "epoch": 1.1877567789646672,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008195155854685734,
      "loss": 4.8899,
      "step": 8673
    },
    {
      "epoch": 1.187893727745823,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008194922042114971,
      "loss": 4.8969,
      "step": 8674
    },
    {
      "epoch": 1.1880306765269788,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008194688198923509,
      "loss": 4.9384,
      "step": 8675
    },
    {
      "epoch": 1.1881676253081348,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008194454325113285,
      "loss": 4.8662,
      "step": 8676
    },
    {
      "epoch": 1.1883045740892906,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008194220420686239,
      "loss": 4.859,
      "step": 8677
    },
    {
      "epoch": 1.1884415228704466,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008193986485644307,
      "loss": 4.8971,
      "step": 8678
    },
    {
      "epoch": 1.1885784716516024,
      "grad_norm": 0.375,
      "learning_rate": 0.0008193752519989429,
      "loss": 5.023,
      "step": 8679
    },
    {
      "epoch": 1.1887154204327581,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008193518523723543,
      "loss": 4.9034,
      "step": 8680
    },
    {
      "epoch": 1.188852369213914,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008193284496848589,
      "loss": 4.8512,
      "step": 8681
    },
    {
      "epoch": 1.18898931799507,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008193050439366506,
      "loss": 4.9218,
      "step": 8682
    },
    {
      "epoch": 1.1891262667762257,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008192816351279235,
      "loss": 4.9659,
      "step": 8683
    },
    {
      "epoch": 1.1892632155573815,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008192582232588715,
      "loss": 4.9503,
      "step": 8684
    },
    {
      "epoch": 1.1894001643385375,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008192348083296886,
      "loss": 4.9246,
      "step": 8685
    },
    {
      "epoch": 1.1895371131196932,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008192113903405687,
      "loss": 4.8593,
      "step": 8686
    },
    {
      "epoch": 1.189674061900849,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008191879692917062,
      "loss": 4.7678,
      "step": 8687
    },
    {
      "epoch": 1.189811010682005,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000819164545183295,
      "loss": 4.871,
      "step": 8688
    },
    {
      "epoch": 1.1899479594631608,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008191411180155293,
      "loss": 4.9681,
      "step": 8689
    },
    {
      "epoch": 1.1900849082443166,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008191176877886032,
      "loss": 4.92,
      "step": 8690
    },
    {
      "epoch": 1.1902218570254726,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008190942545027107,
      "loss": 4.8752,
      "step": 8691
    },
    {
      "epoch": 1.1903588058066283,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008190708181580463,
      "loss": 4.9103,
      "step": 8692
    },
    {
      "epoch": 1.1904957545877841,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000819047378754804,
      "loss": 4.8912,
      "step": 8693
    },
    {
      "epoch": 1.19063270336894,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008190239362931783,
      "loss": 4.8921,
      "step": 8694
    },
    {
      "epoch": 1.190769652150096,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008190004907733631,
      "loss": 4.8941,
      "step": 8695
    },
    {
      "epoch": 1.1909066009312517,
      "grad_norm": 0.390625,
      "learning_rate": 0.000818977042195553,
      "loss": 4.9432,
      "step": 8696
    },
    {
      "epoch": 1.1910435497124077,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008189535905599422,
      "loss": 4.8757,
      "step": 8697
    },
    {
      "epoch": 1.1911804984935634,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008189301358667251,
      "loss": 4.9518,
      "step": 8698
    },
    {
      "epoch": 1.1913174472747192,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000818906678116096,
      "loss": 4.9984,
      "step": 8699
    },
    {
      "epoch": 1.191454396055875,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008188832173082496,
      "loss": 4.8756,
      "step": 8700
    },
    {
      "epoch": 1.191591344837031,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008188597534433797,
      "loss": 4.9414,
      "step": 8701
    },
    {
      "epoch": 1.1917282936181868,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008188362865216814,
      "loss": 4.9052,
      "step": 8702
    },
    {
      "epoch": 1.1918652423993426,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008188128165433488,
      "loss": 4.9243,
      "step": 8703
    },
    {
      "epoch": 1.1920021911804985,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008187893435085765,
      "loss": 4.9209,
      "step": 8704
    },
    {
      "epoch": 1.1921391399616543,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008187658674175592,
      "loss": 4.7799,
      "step": 8705
    },
    {
      "epoch": 1.19227608874281,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000818742388270491,
      "loss": 4.8506,
      "step": 8706
    },
    {
      "epoch": 1.192413037523966,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008187189060675669,
      "loss": 4.9778,
      "step": 8707
    },
    {
      "epoch": 1.1925499863051219,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008186954208089813,
      "loss": 4.933,
      "step": 8708
    },
    {
      "epoch": 1.1926869350862777,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000818671932494929,
      "loss": 4.9466,
      "step": 8709
    },
    {
      "epoch": 1.1928238838674337,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008186484411256045,
      "loss": 4.9152,
      "step": 8710
    },
    {
      "epoch": 1.1929608326485894,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008186249467012026,
      "loss": 4.8927,
      "step": 8711
    },
    {
      "epoch": 1.1930977814297452,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008186014492219178,
      "loss": 4.9188,
      "step": 8712
    },
    {
      "epoch": 1.1932347302109012,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008185779486879451,
      "loss": 4.8527,
      "step": 8713
    },
    {
      "epoch": 1.193371678992057,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000818554445099479,
      "loss": 4.8785,
      "step": 8714
    },
    {
      "epoch": 1.1935086277732128,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008185309384567144,
      "loss": 4.8899,
      "step": 8715
    },
    {
      "epoch": 1.1936455765543688,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008185074287598461,
      "loss": 4.9421,
      "step": 8716
    },
    {
      "epoch": 1.1937825253355245,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008184839160090689,
      "loss": 4.8864,
      "step": 8717
    },
    {
      "epoch": 1.1939194741166803,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008184604002045777,
      "loss": 4.9115,
      "step": 8718
    },
    {
      "epoch": 1.194056422897836,
      "grad_norm": 0.375,
      "learning_rate": 0.0008184368813465674,
      "loss": 4.8852,
      "step": 8719
    },
    {
      "epoch": 1.194193371678992,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008184133594352328,
      "loss": 4.9255,
      "step": 8720
    },
    {
      "epoch": 1.1943303204601479,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000818389834470769,
      "loss": 4.8335,
      "step": 8721
    },
    {
      "epoch": 1.1944672692413039,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008183663064533709,
      "loss": 4.9242,
      "step": 8722
    },
    {
      "epoch": 1.1946042180224596,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008183427753832334,
      "loss": 4.8359,
      "step": 8723
    },
    {
      "epoch": 1.1947411668036154,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008183192412605514,
      "loss": 4.9311,
      "step": 8724
    },
    {
      "epoch": 1.1948781155847712,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008182957040855202,
      "loss": 4.9062,
      "step": 8725
    },
    {
      "epoch": 1.1950150643659272,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008182721638583349,
      "loss": 4.9199,
      "step": 8726
    },
    {
      "epoch": 1.195152013147083,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008182486205791901,
      "loss": 4.8241,
      "step": 8727
    },
    {
      "epoch": 1.1952889619282387,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008182250742482814,
      "loss": 4.8945,
      "step": 8728
    },
    {
      "epoch": 1.1954259107093947,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008182015248658038,
      "loss": 4.9274,
      "step": 8729
    },
    {
      "epoch": 1.1955628594905505,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008181779724319524,
      "loss": 4.8801,
      "step": 8730
    },
    {
      "epoch": 1.1956998082717063,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008181544169469224,
      "loss": 4.9144,
      "step": 8731
    },
    {
      "epoch": 1.1958367570528623,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000818130858410909,
      "loss": 4.9349,
      "step": 8732
    },
    {
      "epoch": 1.195973705834018,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008181072968241075,
      "loss": 4.8778,
      "step": 8733
    },
    {
      "epoch": 1.1961106546151739,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000818083732186713,
      "loss": 4.9078,
      "step": 8734
    },
    {
      "epoch": 1.1962476033963299,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000818060164498921,
      "loss": 4.8988,
      "step": 8735
    },
    {
      "epoch": 1.1963845521774856,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008180365937609266,
      "loss": 4.9079,
      "step": 8736
    },
    {
      "epoch": 1.1965215009586414,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008180130199729255,
      "loss": 4.9005,
      "step": 8737
    },
    {
      "epoch": 1.1966584497397974,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0008179894431351126,
      "loss": 4.9794,
      "step": 8738
    },
    {
      "epoch": 1.1967953985209532,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008179658632476835,
      "loss": 4.8981,
      "step": 8739
    },
    {
      "epoch": 1.196932347302109,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008179422803108337,
      "loss": 4.9871,
      "step": 8740
    },
    {
      "epoch": 1.197069296083265,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008179186943247585,
      "loss": 4.8943,
      "step": 8741
    },
    {
      "epoch": 1.1972062448644207,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008178951052896534,
      "loss": 4.947,
      "step": 8742
    },
    {
      "epoch": 1.1973431936455765,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000817871513205714,
      "loss": 4.9021,
      "step": 8743
    },
    {
      "epoch": 1.1974801424267323,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008178479180731357,
      "loss": 4.7893,
      "step": 8744
    },
    {
      "epoch": 1.1976170912078883,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008178243198921139,
      "loss": 4.9744,
      "step": 8745
    },
    {
      "epoch": 1.197754039989044,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008178007186628443,
      "loss": 4.9865,
      "step": 8746
    },
    {
      "epoch": 1.1978909887702,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008177771143855228,
      "loss": 4.8753,
      "step": 8747
    },
    {
      "epoch": 1.1980279375513558,
      "grad_norm": 0.53125,
      "learning_rate": 0.0008177535070603444,
      "loss": 4.8646,
      "step": 8748
    },
    {
      "epoch": 1.1981648863325116,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008177298966875052,
      "loss": 4.9266,
      "step": 8749
    },
    {
      "epoch": 1.1983018351136674,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0008177062832672007,
      "loss": 4.9417,
      "step": 8750
    },
    {
      "epoch": 1.1984387838948234,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008176826667996266,
      "loss": 4.8575,
      "step": 8751
    },
    {
      "epoch": 1.1985757326759792,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008176590472849787,
      "loss": 4.8893,
      "step": 8752
    },
    {
      "epoch": 1.198712681457135,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008176354247234524,
      "loss": 4.95,
      "step": 8753
    },
    {
      "epoch": 1.198849630238291,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000817611799115244,
      "loss": 4.8652,
      "step": 8754
    },
    {
      "epoch": 1.1989865790194467,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000817588170460549,
      "loss": 4.8431,
      "step": 8755
    },
    {
      "epoch": 1.1991235278006025,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008175645387595631,
      "loss": 4.8452,
      "step": 8756
    },
    {
      "epoch": 1.1992604765817585,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008175409040124824,
      "loss": 4.8943,
      "step": 8757
    },
    {
      "epoch": 1.1993974253629143,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008175172662195026,
      "loss": 4.868,
      "step": 8758
    },
    {
      "epoch": 1.19953437414407,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008174936253808198,
      "loss": 4.9694,
      "step": 8759
    },
    {
      "epoch": 1.199671322925226,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008174699814966296,
      "loss": 4.9707,
      "step": 8760
    },
    {
      "epoch": 1.1998082717063818,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000817446334567128,
      "loss": 4.8703,
      "step": 8761
    },
    {
      "epoch": 1.1999452204875376,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008174226845925112,
      "loss": 4.9041,
      "step": 8762
    },
    {
      "epoch": 1.2000821692686936,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000817399031572975,
      "loss": 4.9053,
      "step": 8763
    },
    {
      "epoch": 1.2002191180498494,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008173753755087156,
      "loss": 4.95,
      "step": 8764
    },
    {
      "epoch": 1.2003560668310052,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008173517163999289,
      "loss": 4.9272,
      "step": 8765
    },
    {
      "epoch": 1.2004930156121612,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008173280542468109,
      "loss": 4.8566,
      "step": 8766
    },
    {
      "epoch": 1.200629964393317,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008173043890495578,
      "loss": 4.8978,
      "step": 8767
    },
    {
      "epoch": 1.2007669131744727,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008172807208083658,
      "loss": 4.8708,
      "step": 8768
    },
    {
      "epoch": 1.2009038619556285,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008172570495234309,
      "loss": 4.9818,
      "step": 8769
    },
    {
      "epoch": 1.2010408107367845,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008172333751949492,
      "loss": 4.8503,
      "step": 8770
    },
    {
      "epoch": 1.2011777595179403,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008172096978231171,
      "loss": 4.9426,
      "step": 8771
    },
    {
      "epoch": 1.2013147082990963,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008171860174081307,
      "loss": 4.9016,
      "step": 8772
    },
    {
      "epoch": 1.201451657080252,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008171623339501864,
      "loss": 4.8851,
      "step": 8773
    },
    {
      "epoch": 1.2015886058614078,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008171386474494802,
      "loss": 4.9336,
      "step": 8774
    },
    {
      "epoch": 1.2017255546425636,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008171149579062086,
      "loss": 4.7848,
      "step": 8775
    },
    {
      "epoch": 1.2018625034237196,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008170912653205679,
      "loss": 4.9127,
      "step": 8776
    },
    {
      "epoch": 1.2019994522048754,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008170675696927544,
      "loss": 4.9077,
      "step": 8777
    },
    {
      "epoch": 1.2021364009860311,
      "grad_norm": 0.375,
      "learning_rate": 0.0008170438710229644,
      "loss": 4.9543,
      "step": 8778
    },
    {
      "epoch": 1.2022733497671871,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008170201693113946,
      "loss": 4.8978,
      "step": 8779
    },
    {
      "epoch": 1.202410298548343,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000816996464558241,
      "loss": 4.9199,
      "step": 8780
    },
    {
      "epoch": 1.2025472473294987,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008169727567637003,
      "loss": 4.9367,
      "step": 8781
    },
    {
      "epoch": 1.2026841961106547,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008169490459279689,
      "loss": 4.857,
      "step": 8782
    },
    {
      "epoch": 1.2028211448918105,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008169253320512433,
      "loss": 4.8807,
      "step": 8783
    },
    {
      "epoch": 1.2029580936729662,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008169016151337201,
      "loss": 4.8519,
      "step": 8784
    },
    {
      "epoch": 1.2030950424541222,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008168778951755957,
      "loss": 4.9542,
      "step": 8785
    },
    {
      "epoch": 1.203231991235278,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008168541721770669,
      "loss": 4.8872,
      "step": 8786
    },
    {
      "epoch": 1.2033689400164338,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008168304461383301,
      "loss": 4.8945,
      "step": 8787
    },
    {
      "epoch": 1.2035058887975898,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000816806717059582,
      "loss": 4.8709,
      "step": 8788
    },
    {
      "epoch": 1.2036428375787456,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008167829849410191,
      "loss": 4.8609,
      "step": 8789
    },
    {
      "epoch": 1.2037797863599013,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008167592497828382,
      "loss": 4.8436,
      "step": 8790
    },
    {
      "epoch": 1.2039167351410573,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008167355115852363,
      "loss": 4.7993,
      "step": 8791
    },
    {
      "epoch": 1.2040536839222131,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008167117703484095,
      "loss": 4.9304,
      "step": 8792
    },
    {
      "epoch": 1.204190632703369,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000816688026072555,
      "loss": 4.8792,
      "step": 8793
    },
    {
      "epoch": 1.2043275814845247,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008166642787578695,
      "loss": 4.8698,
      "step": 8794
    },
    {
      "epoch": 1.2044645302656807,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008166405284045497,
      "loss": 4.9036,
      "step": 8795
    },
    {
      "epoch": 1.2046014790468365,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008166167750127924,
      "loss": 4.9064,
      "step": 8796
    },
    {
      "epoch": 1.2047384278279925,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008165930185827946,
      "loss": 4.8678,
      "step": 8797
    },
    {
      "epoch": 1.2048753766091482,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008165692591147533,
      "loss": 4.9092,
      "step": 8798
    },
    {
      "epoch": 1.205012325390304,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008165454966088648,
      "loss": 4.8429,
      "step": 8799
    },
    {
      "epoch": 1.2051492741714598,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008165217310653267,
      "loss": 4.8908,
      "step": 8800
    },
    {
      "epoch": 1.2052862229526158,
      "grad_norm": 0.5,
      "learning_rate": 0.0008164979624843356,
      "loss": 4.9783,
      "step": 8801
    },
    {
      "epoch": 1.2054231717337716,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008164741908660885,
      "loss": 4.9039,
      "step": 8802
    },
    {
      "epoch": 1.2055601205149273,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008164504162107827,
      "loss": 4.7804,
      "step": 8803
    },
    {
      "epoch": 1.2056970692960833,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008164266385186146,
      "loss": 4.8763,
      "step": 8804
    },
    {
      "epoch": 1.2058340180772391,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008164028577897818,
      "loss": 4.869,
      "step": 8805
    },
    {
      "epoch": 1.2059709668583949,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008163790740244813,
      "loss": 4.8795,
      "step": 8806
    },
    {
      "epoch": 1.2061079156395509,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008163552872229101,
      "loss": 4.823,
      "step": 8807
    },
    {
      "epoch": 1.2062448644207067,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008163314973852652,
      "loss": 4.9603,
      "step": 8808
    },
    {
      "epoch": 1.2063818132018624,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000816307704511744,
      "loss": 4.8472,
      "step": 8809
    },
    {
      "epoch": 1.2065187619830184,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008162839086025436,
      "loss": 4.8843,
      "step": 8810
    },
    {
      "epoch": 1.2066557107641742,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008162601096578611,
      "loss": 4.8368,
      "step": 8811
    },
    {
      "epoch": 1.20679265954533,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008162363076778938,
      "loss": 4.8838,
      "step": 8812
    },
    {
      "epoch": 1.2069296083264858,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000816212502662839,
      "loss": 4.8916,
      "step": 8813
    },
    {
      "epoch": 1.2070665571076418,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008161886946128939,
      "loss": 4.8992,
      "step": 8814
    },
    {
      "epoch": 1.2072035058887975,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008161648835282557,
      "loss": 4.8267,
      "step": 8815
    },
    {
      "epoch": 1.2073404546699535,
      "grad_norm": 0.390625,
      "learning_rate": 0.000816141069409122,
      "loss": 4.9415,
      "step": 8816
    },
    {
      "epoch": 1.2074774034511093,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008161172522556901,
      "loss": 4.913,
      "step": 8817
    },
    {
      "epoch": 1.207614352232265,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008160934320681571,
      "loss": 4.8881,
      "step": 8818
    },
    {
      "epoch": 1.2077513010134209,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008160696088467208,
      "loss": 4.9585,
      "step": 8819
    },
    {
      "epoch": 1.2078882497945769,
      "grad_norm": 0.375,
      "learning_rate": 0.0008160457825915781,
      "loss": 4.9266,
      "step": 8820
    },
    {
      "epoch": 1.2080251985757327,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008160219533029269,
      "loss": 4.8694,
      "step": 8821
    },
    {
      "epoch": 1.2081621473568884,
      "grad_norm": 0.375,
      "learning_rate": 0.0008159981209809647,
      "loss": 4.8515,
      "step": 8822
    },
    {
      "epoch": 1.2082990961380444,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0008159742856258888,
      "loss": 4.958,
      "step": 8823
    },
    {
      "epoch": 1.2084360449192002,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008159504472378969,
      "loss": 4.9442,
      "step": 8824
    },
    {
      "epoch": 1.208572993700356,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008159266058171862,
      "loss": 4.8409,
      "step": 8825
    },
    {
      "epoch": 1.208709942481512,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008159027613639547,
      "loss": 4.8866,
      "step": 8826
    },
    {
      "epoch": 1.2088468912626678,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008158789138783997,
      "loss": 4.8959,
      "step": 8827
    },
    {
      "epoch": 1.2089838400438235,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000815855063360719,
      "loss": 4.8993,
      "step": 8828
    },
    {
      "epoch": 1.2091207888249795,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008158312098111103,
      "loss": 4.9441,
      "step": 8829
    },
    {
      "epoch": 1.2092577376061353,
      "grad_norm": 0.34375,
      "learning_rate": 0.000815807353229771,
      "loss": 4.8773,
      "step": 8830
    },
    {
      "epoch": 1.209394686387291,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008157834936168991,
      "loss": 4.939,
      "step": 8831
    },
    {
      "epoch": 1.209531635168447,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008157596309726923,
      "loss": 4.9547,
      "step": 8832
    },
    {
      "epoch": 1.2096685839496029,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008157357652973482,
      "loss": 4.8966,
      "step": 8833
    },
    {
      "epoch": 1.2098055327307586,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008157118965910647,
      "loss": 4.959,
      "step": 8834
    },
    {
      "epoch": 1.2099424815119146,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008156880248540395,
      "loss": 4.791,
      "step": 8835
    },
    {
      "epoch": 1.2100794302930704,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008156641500864706,
      "loss": 4.8864,
      "step": 8836
    },
    {
      "epoch": 1.2102163790742262,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008156402722885556,
      "loss": 4.9668,
      "step": 8837
    },
    {
      "epoch": 1.210353327855382,
      "grad_norm": 0.375,
      "learning_rate": 0.0008156163914604927,
      "loss": 4.7959,
      "step": 8838
    },
    {
      "epoch": 1.210490276636538,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008155925076024797,
      "loss": 4.9373,
      "step": 8839
    },
    {
      "epoch": 1.2106272254176937,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008155686207147142,
      "loss": 4.9024,
      "step": 8840
    },
    {
      "epoch": 1.2107641741988497,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008155447307973947,
      "loss": 4.8401,
      "step": 8841
    },
    {
      "epoch": 1.2109011229800055,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008155208378507187,
      "loss": 4.9097,
      "step": 8842
    },
    {
      "epoch": 1.2110380717611613,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008154969418748846,
      "loss": 4.9345,
      "step": 8843
    },
    {
      "epoch": 1.211175020542317,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00081547304287009,
      "loss": 4.8683,
      "step": 8844
    },
    {
      "epoch": 1.211311969323473,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008154491408365335,
      "loss": 4.9118,
      "step": 8845
    },
    {
      "epoch": 1.2114489181046288,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008154252357744127,
      "loss": 4.7582,
      "step": 8846
    },
    {
      "epoch": 1.2115858668857846,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000815401327683926,
      "loss": 4.9977,
      "step": 8847
    },
    {
      "epoch": 1.2117228156669406,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008153774165652715,
      "loss": 4.889,
      "step": 8848
    },
    {
      "epoch": 1.2118597644480964,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000815353502418647,
      "loss": 4.8555,
      "step": 8849
    },
    {
      "epoch": 1.2119967132292522,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008153295852442512,
      "loss": 4.8613,
      "step": 8850
    },
    {
      "epoch": 1.2121336620104082,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000815305665042282,
      "loss": 4.8814,
      "step": 8851
    },
    {
      "epoch": 1.212270610791564,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008152817418129377,
      "loss": 4.901,
      "step": 8852
    },
    {
      "epoch": 1.2124075595727197,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008152578155564166,
      "loss": 4.8843,
      "step": 8853
    },
    {
      "epoch": 1.2125445083538757,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008152338862729169,
      "loss": 4.9101,
      "step": 8854
    },
    {
      "epoch": 1.2126814571350315,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000815209953962637,
      "loss": 4.8626,
      "step": 8855
    },
    {
      "epoch": 1.2128184059161873,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008151860186257753,
      "loss": 4.9782,
      "step": 8856
    },
    {
      "epoch": 1.2129553546973433,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008151620802625297,
      "loss": 4.9042,
      "step": 8857
    },
    {
      "epoch": 1.213092303478499,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008151381388730992,
      "loss": 4.8926,
      "step": 8858
    },
    {
      "epoch": 1.2132292522596548,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008151141944576819,
      "loss": 4.8665,
      "step": 8859
    },
    {
      "epoch": 1.2133662010408108,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008150902470164762,
      "loss": 4.9259,
      "step": 8860
    },
    {
      "epoch": 1.2135031498219666,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008150662965496807,
      "loss": 4.9068,
      "step": 8861
    },
    {
      "epoch": 1.2136400986031224,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0008150423430574937,
      "loss": 4.9348,
      "step": 8862
    },
    {
      "epoch": 1.2137770473842782,
      "grad_norm": 0.375,
      "learning_rate": 0.0008150183865401139,
      "loss": 4.8645,
      "step": 8863
    },
    {
      "epoch": 1.2139139961654342,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008149944269977398,
      "loss": 4.8235,
      "step": 8864
    },
    {
      "epoch": 1.21405094494659,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008149704644305699,
      "loss": 4.8908,
      "step": 8865
    },
    {
      "epoch": 1.214187893727746,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008149464988388027,
      "loss": 4.9027,
      "step": 8866
    },
    {
      "epoch": 1.2143248425089017,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000814922530222637,
      "loss": 4.8378,
      "step": 8867
    },
    {
      "epoch": 1.2144617912900575,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008148985585822712,
      "loss": 4.9585,
      "step": 8868
    },
    {
      "epoch": 1.2145987400712133,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008148745839179043,
      "loss": 4.8736,
      "step": 8869
    },
    {
      "epoch": 1.2147356888523693,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008148506062297346,
      "loss": 4.8514,
      "step": 8870
    },
    {
      "epoch": 1.214872637633525,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000814826625517961,
      "loss": 4.8296,
      "step": 8871
    },
    {
      "epoch": 1.2150095864146808,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008148026417827823,
      "loss": 4.7948,
      "step": 8872
    },
    {
      "epoch": 1.2151465351958368,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008147786550243972,
      "loss": 4.8991,
      "step": 8873
    },
    {
      "epoch": 1.2152834839769926,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008147546652430043,
      "loss": 4.9433,
      "step": 8874
    },
    {
      "epoch": 1.2154204327581484,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008147306724388027,
      "loss": 4.7892,
      "step": 8875
    },
    {
      "epoch": 1.2155573815393044,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008147066766119912,
      "loss": 4.8273,
      "step": 8876
    },
    {
      "epoch": 1.2156943303204601,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008146826777627684,
      "loss": 4.8213,
      "step": 8877
    },
    {
      "epoch": 1.215831279101616,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008146586758913335,
      "loss": 4.8865,
      "step": 8878
    },
    {
      "epoch": 1.215968227882772,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008146346709978851,
      "loss": 4.7802,
      "step": 8879
    },
    {
      "epoch": 1.2161051766639277,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008146106630826223,
      "loss": 4.8893,
      "step": 8880
    },
    {
      "epoch": 1.2162421254450835,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008145866521457442,
      "loss": 4.8589,
      "step": 8881
    },
    {
      "epoch": 1.2163790742262395,
      "grad_norm": 0.375,
      "learning_rate": 0.0008145626381874495,
      "loss": 4.8732,
      "step": 8882
    },
    {
      "epoch": 1.2165160230073953,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008145386212079375,
      "loss": 4.9311,
      "step": 8883
    },
    {
      "epoch": 1.216652971788551,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008145146012074069,
      "loss": 4.8948,
      "step": 8884
    },
    {
      "epoch": 1.216789920569707,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008144905781860569,
      "loss": 4.8274,
      "step": 8885
    },
    {
      "epoch": 1.2169268693508628,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008144665521440869,
      "loss": 4.9337,
      "step": 8886
    },
    {
      "epoch": 1.2170638181320186,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008144425230816955,
      "loss": 4.8135,
      "step": 8887
    },
    {
      "epoch": 1.2172007669131744,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008144184909990822,
      "loss": 4.8492,
      "step": 8888
    },
    {
      "epoch": 1.2173377156943304,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008143944558964459,
      "loss": 4.862,
      "step": 8889
    },
    {
      "epoch": 1.2174746644754861,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000814370417773986,
      "loss": 4.8815,
      "step": 8890
    },
    {
      "epoch": 1.2176116132566421,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008143463766319016,
      "loss": 4.8822,
      "step": 8891
    },
    {
      "epoch": 1.217748562037798,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008143223324703921,
      "loss": 4.8346,
      "step": 8892
    },
    {
      "epoch": 1.2178855108189537,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008142982852896563,
      "loss": 4.9173,
      "step": 8893
    },
    {
      "epoch": 1.2180224596001095,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008142742350898941,
      "loss": 4.8594,
      "step": 8894
    },
    {
      "epoch": 1.2181594083812655,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008142501818713043,
      "loss": 4.8536,
      "step": 8895
    },
    {
      "epoch": 1.2182963571624212,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008142261256340867,
      "loss": 4.8713,
      "step": 8896
    },
    {
      "epoch": 1.218433305943577,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008142020663784402,
      "loss": 4.9567,
      "step": 8897
    },
    {
      "epoch": 1.218570254724733,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008141780041045645,
      "loss": 4.9409,
      "step": 8898
    },
    {
      "epoch": 1.2187072035058888,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008141539388126589,
      "loss": 4.8872,
      "step": 8899
    },
    {
      "epoch": 1.2188441522870446,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008141298705029227,
      "loss": 4.892,
      "step": 8900
    },
    {
      "epoch": 1.2189811010682006,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008141057991755557,
      "loss": 4.8509,
      "step": 8901
    },
    {
      "epoch": 1.2191180498493563,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008140817248307571,
      "loss": 4.958,
      "step": 8902
    },
    {
      "epoch": 1.2192549986305121,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008140576474687263,
      "loss": 4.9184,
      "step": 8903
    },
    {
      "epoch": 1.2193919474116681,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008140335670896631,
      "loss": 4.9226,
      "step": 8904
    },
    {
      "epoch": 1.219528896192824,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008140094836937672,
      "loss": 4.8416,
      "step": 8905
    },
    {
      "epoch": 1.2196658449739797,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008139853972812378,
      "loss": 4.8961,
      "step": 8906
    },
    {
      "epoch": 1.2198027937551357,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008139613078522747,
      "loss": 4.8702,
      "step": 8907
    },
    {
      "epoch": 1.2199397425362914,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008139372154070775,
      "loss": 4.8286,
      "step": 8908
    },
    {
      "epoch": 1.2200766913174472,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008139131199458459,
      "loss": 4.9319,
      "step": 8909
    },
    {
      "epoch": 1.2202136400986032,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008138890214687796,
      "loss": 4.8804,
      "step": 8910
    },
    {
      "epoch": 1.220350588879759,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008138649199760783,
      "loss": 4.8454,
      "step": 8911
    },
    {
      "epoch": 1.2204875376609148,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008138408154679416,
      "loss": 4.9529,
      "step": 8912
    },
    {
      "epoch": 1.2206244864420706,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008138167079445694,
      "loss": 4.8696,
      "step": 8913
    },
    {
      "epoch": 1.2207614352232266,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008137925974061614,
      "loss": 4.9074,
      "step": 8914
    },
    {
      "epoch": 1.2208983840043823,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008137684838529176,
      "loss": 4.9333,
      "step": 8915
    },
    {
      "epoch": 1.2210353327855383,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008137443672850374,
      "loss": 4.8379,
      "step": 8916
    },
    {
      "epoch": 1.221172281566694,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008137202477027213,
      "loss": 4.9432,
      "step": 8917
    },
    {
      "epoch": 1.2213092303478499,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008136961251061686,
      "loss": 4.8438,
      "step": 8918
    },
    {
      "epoch": 1.2214461791290057,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008136719994955796,
      "loss": 4.8981,
      "step": 8919
    },
    {
      "epoch": 1.2215831279101617,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000813647870871154,
      "loss": 4.8128,
      "step": 8920
    },
    {
      "epoch": 1.2217200766913174,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008136237392330918,
      "loss": 4.9114,
      "step": 8921
    },
    {
      "epoch": 1.2218570254724732,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000813599604581593,
      "loss": 4.9316,
      "step": 8922
    },
    {
      "epoch": 1.2219939742536292,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008135754669168579,
      "loss": 4.9686,
      "step": 8923
    },
    {
      "epoch": 1.222130923034785,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000813551326239086,
      "loss": 4.8918,
      "step": 8924
    },
    {
      "epoch": 1.2222678718159408,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008135271825484777,
      "loss": 4.842,
      "step": 8925
    },
    {
      "epoch": 1.2224048205970968,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008135030358452331,
      "loss": 4.948,
      "step": 8926
    },
    {
      "epoch": 1.2225417693782525,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008134788861295522,
      "loss": 4.9561,
      "step": 8927
    },
    {
      "epoch": 1.2226787181594083,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000813454733401635,
      "loss": 4.8023,
      "step": 8928
    },
    {
      "epoch": 1.2228156669405643,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000813430577661682,
      "loss": 4.9576,
      "step": 8929
    },
    {
      "epoch": 1.22295261572172,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008134064189098932,
      "loss": 4.899,
      "step": 8930
    },
    {
      "epoch": 1.2230895645028759,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008133822571464687,
      "loss": 4.9137,
      "step": 8931
    },
    {
      "epoch": 1.2232265132840319,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008133580923716089,
      "loss": 4.9492,
      "step": 8932
    },
    {
      "epoch": 1.2233634620651876,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000813333924585514,
      "loss": 4.8999,
      "step": 8933
    },
    {
      "epoch": 1.2235004108463434,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008133097537883843,
      "loss": 4.9287,
      "step": 8934
    },
    {
      "epoch": 1.2236373596274994,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008132855799804201,
      "loss": 4.8533,
      "step": 8935
    },
    {
      "epoch": 1.2237743084086552,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008132614031618217,
      "loss": 4.8711,
      "step": 8936
    },
    {
      "epoch": 1.223911257189811,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008132372233327895,
      "loss": 4.8064,
      "step": 8937
    },
    {
      "epoch": 1.2240482059709668,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008132130404935239,
      "loss": 4.8303,
      "step": 8938
    },
    {
      "epoch": 1.2241851547521228,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008131888546442252,
      "loss": 4.9275,
      "step": 8939
    },
    {
      "epoch": 1.2243221035332785,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008131646657850941,
      "loss": 4.9058,
      "step": 8940
    },
    {
      "epoch": 1.2244590523144345,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008131404739163306,
      "loss": 4.9161,
      "step": 8941
    },
    {
      "epoch": 1.2245960010955903,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008131162790381357,
      "loss": 4.9142,
      "step": 8942
    },
    {
      "epoch": 1.224732949876746,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008130920811507096,
      "loss": 4.8234,
      "step": 8943
    },
    {
      "epoch": 1.2248698986579019,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008130678802542528,
      "loss": 4.9252,
      "step": 8944
    },
    {
      "epoch": 1.2250068474390579,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008130436763489661,
      "loss": 4.8035,
      "step": 8945
    },
    {
      "epoch": 1.2251437962202136,
      "grad_norm": 0.375,
      "learning_rate": 0.00081301946943505,
      "loss": 4.9544,
      "step": 8946
    },
    {
      "epoch": 1.2252807450013694,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000812995259512705,
      "loss": 4.8556,
      "step": 8947
    },
    {
      "epoch": 1.2254176937825254,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008129710465821317,
      "loss": 4.8992,
      "step": 8948
    },
    {
      "epoch": 1.2255546425636812,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000812946830643531,
      "loss": 5.0205,
      "step": 8949
    },
    {
      "epoch": 1.225691591344837,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008129226116971032,
      "loss": 4.876,
      "step": 8950
    },
    {
      "epoch": 1.225828540125993,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008128983897430493,
      "loss": 4.8757,
      "step": 8951
    },
    {
      "epoch": 1.2259654889071487,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008128741647815701,
      "loss": 4.8688,
      "step": 8952
    },
    {
      "epoch": 1.2261024376883045,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008128499368128663,
      "loss": 4.9036,
      "step": 8953
    },
    {
      "epoch": 1.2262393864694605,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008128257058371384,
      "loss": 4.865,
      "step": 8954
    },
    {
      "epoch": 1.2263763352506163,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008128014718545875,
      "loss": 5.0066,
      "step": 8955
    },
    {
      "epoch": 1.226513284031772,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008127772348654143,
      "loss": 4.852,
      "step": 8956
    },
    {
      "epoch": 1.2266502328129278,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008127529948698198,
      "loss": 4.8973,
      "step": 8957
    },
    {
      "epoch": 1.2267871815940838,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008127287518680047,
      "loss": 4.9381,
      "step": 8958
    },
    {
      "epoch": 1.2269241303752396,
      "grad_norm": 0.375,
      "learning_rate": 0.00081270450586017,
      "loss": 4.932,
      "step": 8959
    },
    {
      "epoch": 1.2270610791563956,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008126802568465167,
      "loss": 4.8608,
      "step": 8960
    },
    {
      "epoch": 1.2271980279375514,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008126560048272457,
      "loss": 4.9276,
      "step": 8961
    },
    {
      "epoch": 1.2273349767187072,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008126317498025579,
      "loss": 4.9344,
      "step": 8962
    },
    {
      "epoch": 1.227471925499863,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008126074917726544,
      "loss": 4.8545,
      "step": 8963
    },
    {
      "epoch": 1.227608874281019,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008125832307377363,
      "loss": 4.8756,
      "step": 8964
    },
    {
      "epoch": 1.2277458230621747,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008125589666980045,
      "loss": 4.8741,
      "step": 8965
    },
    {
      "epoch": 1.2278827718433305,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008125346996536602,
      "loss": 4.9257,
      "step": 8966
    },
    {
      "epoch": 1.2280197206244865,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008125104296049043,
      "loss": 4.9453,
      "step": 8967
    },
    {
      "epoch": 1.2281566694056423,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008124861565519383,
      "loss": 4.8715,
      "step": 8968
    },
    {
      "epoch": 1.228293618186798,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000812461880494963,
      "loss": 4.9175,
      "step": 8969
    },
    {
      "epoch": 1.228430566967954,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008124376014341797,
      "loss": 4.9569,
      "step": 8970
    },
    {
      "epoch": 1.2285675157491098,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008124133193697897,
      "loss": 4.8949,
      "step": 8971
    },
    {
      "epoch": 1.2287044645302656,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008123890343019941,
      "loss": 4.8896,
      "step": 8972
    },
    {
      "epoch": 1.2288414133114216,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008123647462309943,
      "loss": 4.8927,
      "step": 8973
    },
    {
      "epoch": 1.2289783620925774,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008123404551569916,
      "loss": 4.8518,
      "step": 8974
    },
    {
      "epoch": 1.2291153108737332,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008123161610801869,
      "loss": 4.9129,
      "step": 8975
    },
    {
      "epoch": 1.2292522596548892,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008122918640007821,
      "loss": 4.9071,
      "step": 8976
    },
    {
      "epoch": 1.229389208436045,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008122675639189782,
      "loss": 4.9745,
      "step": 8977
    },
    {
      "epoch": 1.2295261572172007,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008122432608349766,
      "loss": 4.8974,
      "step": 8978
    },
    {
      "epoch": 1.2296631059983567,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000812218954748979,
      "loss": 4.9387,
      "step": 8979
    },
    {
      "epoch": 1.2298000547795125,
      "grad_norm": 0.375,
      "learning_rate": 0.0008121946456611864,
      "loss": 4.9095,
      "step": 8980
    },
    {
      "epoch": 1.2299370035606683,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008121703335718006,
      "loss": 4.8975,
      "step": 8981
    },
    {
      "epoch": 1.230073952341824,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008121460184810228,
      "loss": 4.8969,
      "step": 8982
    },
    {
      "epoch": 1.23021090112298,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008121217003890549,
      "loss": 4.9305,
      "step": 8983
    },
    {
      "epoch": 1.2303478499041358,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000812097379296098,
      "loss": 4.9623,
      "step": 8984
    },
    {
      "epoch": 1.2304847986852918,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008120730552023538,
      "loss": 4.8545,
      "step": 8985
    },
    {
      "epoch": 1.2306217474664476,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000812048728108024,
      "loss": 4.8939,
      "step": 8986
    },
    {
      "epoch": 1.2307586962476034,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008120243980133101,
      "loss": 4.9002,
      "step": 8987
    },
    {
      "epoch": 1.2308956450287591,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008120000649184136,
      "loss": 4.8877,
      "step": 8988
    },
    {
      "epoch": 1.2310325938099151,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008119757288235366,
      "loss": 4.865,
      "step": 8989
    },
    {
      "epoch": 1.231169542591071,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008119513897288802,
      "loss": 4.8727,
      "step": 8990
    },
    {
      "epoch": 1.2313064913722267,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008119270476346466,
      "loss": 4.9176,
      "step": 8991
    },
    {
      "epoch": 1.2314434401533827,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008119027025410373,
      "loss": 4.8626,
      "step": 8992
    },
    {
      "epoch": 1.2315803889345385,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000811878354448254,
      "loss": 4.9799,
      "step": 8993
    },
    {
      "epoch": 1.2317173377156942,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0008118540033564984,
      "loss": 4.9495,
      "step": 8994
    },
    {
      "epoch": 1.2318542864968502,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008118296492659727,
      "loss": 4.9041,
      "step": 8995
    },
    {
      "epoch": 1.231991235278006,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008118052921768783,
      "loss": 4.8979,
      "step": 8996
    },
    {
      "epoch": 1.2321281840591618,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008117809320894173,
      "loss": 4.9165,
      "step": 8997
    },
    {
      "epoch": 1.2322651328403178,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008117565690037915,
      "loss": 4.8863,
      "step": 8998
    },
    {
      "epoch": 1.2324020816214736,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008117322029202027,
      "loss": 4.8142,
      "step": 8999
    },
    {
      "epoch": 1.2325390304026294,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000811707833838853,
      "loss": 4.8418,
      "step": 9000
    },
    {
      "epoch": 1.2326759791837854,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008116834617599444,
      "loss": 4.8978,
      "step": 9001
    },
    {
      "epoch": 1.2328129279649411,
      "grad_norm": 0.375,
      "learning_rate": 0.0008116590866836787,
      "loss": 4.9179,
      "step": 9002
    },
    {
      "epoch": 1.232949876746097,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008116347086102581,
      "loss": 4.8096,
      "step": 9003
    },
    {
      "epoch": 1.233086825527253,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008116103275398843,
      "loss": 4.9375,
      "step": 9004
    },
    {
      "epoch": 1.2332237743084087,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008115859434727596,
      "loss": 4.9393,
      "step": 9005
    },
    {
      "epoch": 1.2333607230895645,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000811561556409086,
      "loss": 4.862,
      "step": 9006
    },
    {
      "epoch": 1.2334976718707202,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008115371663490657,
      "loss": 4.8963,
      "step": 9007
    },
    {
      "epoch": 1.2336346206518762,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008115127732929008,
      "loss": 4.8846,
      "step": 9008
    },
    {
      "epoch": 1.233771569433032,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008114883772407933,
      "loss": 4.9318,
      "step": 9009
    },
    {
      "epoch": 1.233908518214188,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008114639781929455,
      "loss": 4.945,
      "step": 9010
    },
    {
      "epoch": 1.2340454669953438,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008114395761495595,
      "loss": 4.8246,
      "step": 9011
    },
    {
      "epoch": 1.2341824157764996,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008114151711108377,
      "loss": 4.8299,
      "step": 9012
    },
    {
      "epoch": 1.2343193645576553,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008113907630769823,
      "loss": 4.9295,
      "step": 9013
    },
    {
      "epoch": 1.2344563133388113,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0008113663520481955,
      "loss": 4.8918,
      "step": 9014
    },
    {
      "epoch": 1.2345932621199671,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008113419380246797,
      "loss": 4.932,
      "step": 9015
    },
    {
      "epoch": 1.234730210901123,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000811317521006637,
      "loss": 4.8917,
      "step": 9016
    },
    {
      "epoch": 1.234867159682279,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00081129310099427,
      "loss": 4.9241,
      "step": 9017
    },
    {
      "epoch": 1.2350041084634347,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008112686779877809,
      "loss": 4.8833,
      "step": 9018
    },
    {
      "epoch": 1.2351410572445904,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008112442519873722,
      "loss": 4.8226,
      "step": 9019
    },
    {
      "epoch": 1.2352780060257464,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008112198229932464,
      "loss": 4.9702,
      "step": 9020
    },
    {
      "epoch": 1.2354149548069022,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008111953910056058,
      "loss": 4.9104,
      "step": 9021
    },
    {
      "epoch": 1.235551903588058,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008111709560246529,
      "loss": 4.9572,
      "step": 9022
    },
    {
      "epoch": 1.235688852369214,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0008111465180505903,
      "loss": 4.9457,
      "step": 9023
    },
    {
      "epoch": 1.2358258011503698,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008111220770836204,
      "loss": 4.9574,
      "step": 9024
    },
    {
      "epoch": 1.2359627499315256,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008110976331239457,
      "loss": 4.8824,
      "step": 9025
    },
    {
      "epoch": 1.2360996987126815,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008110731861717689,
      "loss": 4.9297,
      "step": 9026
    },
    {
      "epoch": 1.2362366474938373,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008110487362272926,
      "loss": 4.8415,
      "step": 9027
    },
    {
      "epoch": 1.236373596274993,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008110242832907195,
      "loss": 4.9014,
      "step": 9028
    },
    {
      "epoch": 1.236510545056149,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008109998273622521,
      "loss": 4.9432,
      "step": 9029
    },
    {
      "epoch": 1.2366474938373049,
      "grad_norm": 0.443359375,
      "learning_rate": 0.000810975368442093,
      "loss": 4.9374,
      "step": 9030
    },
    {
      "epoch": 1.2367844426184607,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008109509065304451,
      "loss": 4.9189,
      "step": 9031
    },
    {
      "epoch": 1.2369213913996164,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000810926441627511,
      "loss": 4.8527,
      "step": 9032
    },
    {
      "epoch": 1.2370583401807724,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008109019737334934,
      "loss": 4.8476,
      "step": 9033
    },
    {
      "epoch": 1.2371952889619282,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008108775028485953,
      "loss": 4.968,
      "step": 9034
    },
    {
      "epoch": 1.2373322377430842,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008108530289730192,
      "loss": 4.8186,
      "step": 9035
    },
    {
      "epoch": 1.23746918652424,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008108285521069682,
      "loss": 4.8327,
      "step": 9036
    },
    {
      "epoch": 1.2376061353053958,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008108040722506449,
      "loss": 4.9235,
      "step": 9037
    },
    {
      "epoch": 1.2377430840865515,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008107795894042523,
      "loss": 4.842,
      "step": 9038
    },
    {
      "epoch": 1.2378800328677075,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008107551035679932,
      "loss": 4.8999,
      "step": 9039
    },
    {
      "epoch": 1.2380169816488633,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008107306147420707,
      "loss": 4.8582,
      "step": 9040
    },
    {
      "epoch": 1.238153930430019,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008107061229266876,
      "loss": 4.9478,
      "step": 9041
    },
    {
      "epoch": 1.238290879211175,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008106816281220469,
      "loss": 4.8622,
      "step": 9042
    },
    {
      "epoch": 1.2384278279923309,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008106571303283517,
      "loss": 4.8922,
      "step": 9043
    },
    {
      "epoch": 1.2385647767734866,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008106326295458049,
      "loss": 4.941,
      "step": 9044
    },
    {
      "epoch": 1.2387017255546426,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008106081257746095,
      "loss": 4.8252,
      "step": 9045
    },
    {
      "epoch": 1.2388386743357984,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008105836190149687,
      "loss": 4.8175,
      "step": 9046
    },
    {
      "epoch": 1.2389756231169542,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008105591092670854,
      "loss": 4.9336,
      "step": 9047
    },
    {
      "epoch": 1.2391125718981102,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008105345965311629,
      "loss": 4.8626,
      "step": 9048
    },
    {
      "epoch": 1.239249520679266,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008105100808074042,
      "loss": 4.8173,
      "step": 9049
    },
    {
      "epoch": 1.2393864694604217,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008104855620960129,
      "loss": 4.889,
      "step": 9050
    },
    {
      "epoch": 1.2395234182415777,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008104610403971915,
      "loss": 4.894,
      "step": 9051
    },
    {
      "epoch": 1.2396603670227335,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008104365157111436,
      "loss": 4.8789,
      "step": 9052
    },
    {
      "epoch": 1.2397973158038893,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008104119880380725,
      "loss": 4.8899,
      "step": 9053
    },
    {
      "epoch": 1.2399342645850453,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008103874573781813,
      "loss": 4.9408,
      "step": 9054
    },
    {
      "epoch": 1.240071213366201,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008103629237316734,
      "loss": 4.9089,
      "step": 9055
    },
    {
      "epoch": 1.2402081621473569,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008103383870987522,
      "loss": 4.945,
      "step": 9056
    },
    {
      "epoch": 1.2403451109285126,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008103138474796209,
      "loss": 4.8544,
      "step": 9057
    },
    {
      "epoch": 1.2404820597096686,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008102893048744826,
      "loss": 4.9015,
      "step": 9058
    },
    {
      "epoch": 1.2406190084908244,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008102647592835413,
      "loss": 4.8625,
      "step": 9059
    },
    {
      "epoch": 1.2407559572719804,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000810240210707,
      "loss": 4.8867,
      "step": 9060
    },
    {
      "epoch": 1.2408929060531362,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008102156591450622,
      "loss": 4.9231,
      "step": 9061
    },
    {
      "epoch": 1.241029854834292,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008101911045979313,
      "loss": 4.9636,
      "step": 9062
    },
    {
      "epoch": 1.2411668036154477,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000810166547065811,
      "loss": 4.9006,
      "step": 9063
    },
    {
      "epoch": 1.2413037523966037,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0008101419865489048,
      "loss": 4.9049,
      "step": 9064
    },
    {
      "epoch": 1.2414407011777595,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000810117423047416,
      "loss": 4.8461,
      "step": 9065
    },
    {
      "epoch": 1.2415776499589153,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008100928565615482,
      "loss": 4.8882,
      "step": 9066
    },
    {
      "epoch": 1.2417145987400713,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008100682870915054,
      "loss": 4.9022,
      "step": 9067
    },
    {
      "epoch": 1.241851547521227,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008100437146374908,
      "loss": 4.9407,
      "step": 9068
    },
    {
      "epoch": 1.2419884963023828,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000810019139199708,
      "loss": 4.9054,
      "step": 9069
    },
    {
      "epoch": 1.2421254450835388,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008099945607783608,
      "loss": 4.8656,
      "step": 9070
    },
    {
      "epoch": 1.2422623938646946,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000809969979373653,
      "loss": 4.8689,
      "step": 9071
    },
    {
      "epoch": 1.2423993426458504,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008099453949857882,
      "loss": 4.9138,
      "step": 9072
    },
    {
      "epoch": 1.2425362914270064,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008099208076149701,
      "loss": 4.8912,
      "step": 9073
    },
    {
      "epoch": 1.2426732402081622,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008098962172614026,
      "loss": 4.9486,
      "step": 9074
    },
    {
      "epoch": 1.242810188989318,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008098716239252892,
      "loss": 4.8718,
      "step": 9075
    },
    {
      "epoch": 1.2429471377704737,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008098470276068341,
      "loss": 4.9098,
      "step": 9076
    },
    {
      "epoch": 1.2430840865516297,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008098224283062408,
      "loss": 4.9351,
      "step": 9077
    },
    {
      "epoch": 1.2432210353327855,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008097978260237134,
      "loss": 4.9055,
      "step": 9078
    },
    {
      "epoch": 1.2433579841139415,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008097732207594557,
      "loss": 4.8963,
      "step": 9079
    },
    {
      "epoch": 1.2434949328950973,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008097486125136714,
      "loss": 4.8857,
      "step": 9080
    },
    {
      "epoch": 1.243631881676253,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008097240012865648,
      "loss": 4.8585,
      "step": 9081
    },
    {
      "epoch": 1.2437688304574088,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008096993870783397,
      "loss": 4.8484,
      "step": 9082
    },
    {
      "epoch": 1.2439057792385648,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008096747698891999,
      "loss": 4.8921,
      "step": 9083
    },
    {
      "epoch": 1.2440427280197206,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008096501497193497,
      "loss": 4.9074,
      "step": 9084
    },
    {
      "epoch": 1.2441796768008764,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008096255265689931,
      "loss": 4.8329,
      "step": 9085
    },
    {
      "epoch": 1.2443166255820324,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008096009004383338,
      "loss": 4.9536,
      "step": 9086
    },
    {
      "epoch": 1.2444535743631882,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008095762713275763,
      "loss": 4.9253,
      "step": 9087
    },
    {
      "epoch": 1.244590523144344,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008095516392369245,
      "loss": 4.8819,
      "step": 9088
    },
    {
      "epoch": 1.2447274719255,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008095270041665827,
      "loss": 4.8205,
      "step": 9089
    },
    {
      "epoch": 1.2448644207066557,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000809502366116755,
      "loss": 4.9048,
      "step": 9090
    },
    {
      "epoch": 1.2450013694878115,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008094777250876454,
      "loss": 4.8768,
      "step": 9091
    },
    {
      "epoch": 1.2451383182689675,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008094530810794583,
      "loss": 4.9164,
      "step": 9092
    },
    {
      "epoch": 1.2452752670501233,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008094284340923978,
      "loss": 4.7751,
      "step": 9093
    },
    {
      "epoch": 1.245412215831279,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008094037841266682,
      "loss": 4.9338,
      "step": 9094
    },
    {
      "epoch": 1.245549164612435,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0008093791311824738,
      "loss": 4.8848,
      "step": 9095
    },
    {
      "epoch": 1.2456861133935908,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000809354475260019,
      "loss": 4.848,
      "step": 9096
    },
    {
      "epoch": 1.2458230621747466,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008093298163595079,
      "loss": 4.9627,
      "step": 9097
    },
    {
      "epoch": 1.2459600109559026,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008093051544811451,
      "loss": 4.9005,
      "step": 9098
    },
    {
      "epoch": 1.2460969597370584,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008092804896251348,
      "loss": 4.9263,
      "step": 9099
    },
    {
      "epoch": 1.2462339085182141,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0008092558217916815,
      "loss": 4.9584,
      "step": 9100
    },
    {
      "epoch": 1.24637085729937,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008092311509809895,
      "loss": 4.8286,
      "step": 9101
    },
    {
      "epoch": 1.246507806080526,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008092064771932636,
      "loss": 4.8577,
      "step": 9102
    },
    {
      "epoch": 1.2466447548616817,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008091818004287079,
      "loss": 4.8904,
      "step": 9103
    },
    {
      "epoch": 1.2467817036428377,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0008091571206875269,
      "loss": 4.9567,
      "step": 9104
    },
    {
      "epoch": 1.2469186524239935,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008091324379699254,
      "loss": 4.9419,
      "step": 9105
    },
    {
      "epoch": 1.2470556012051492,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008091077522761078,
      "loss": 4.7611,
      "step": 9106
    },
    {
      "epoch": 1.247192549986305,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008090830636062786,
      "loss": 4.8974,
      "step": 9107
    },
    {
      "epoch": 1.247329498767461,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008090583719606426,
      "loss": 4.9674,
      "step": 9108
    },
    {
      "epoch": 1.2474664475486168,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008090336773394042,
      "loss": 4.9155,
      "step": 9109
    },
    {
      "epoch": 1.2476033963297726,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008090089797427683,
      "loss": 4.9258,
      "step": 9110
    },
    {
      "epoch": 1.2477403451109286,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008089842791709394,
      "loss": 4.8213,
      "step": 9111
    },
    {
      "epoch": 1.2478772938920843,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008089595756241221,
      "loss": 4.8502,
      "step": 9112
    },
    {
      "epoch": 1.2480142426732401,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008089348691025215,
      "loss": 4.9329,
      "step": 9113
    },
    {
      "epoch": 1.2481511914543961,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008089101596063418,
      "loss": 4.8154,
      "step": 9114
    },
    {
      "epoch": 1.248288140235552,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008088854471357884,
      "loss": 4.8873,
      "step": 9115
    },
    {
      "epoch": 1.2484250890167077,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008088607316910656,
      "loss": 4.9348,
      "step": 9116
    },
    {
      "epoch": 1.2485620377978637,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008088360132723785,
      "loss": 4.8662,
      "step": 9117
    },
    {
      "epoch": 1.2486989865790195,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008088112918799317,
      "loss": 4.9143,
      "step": 9118
    },
    {
      "epoch": 1.2488359353601752,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008087865675139304,
      "loss": 4.8627,
      "step": 9119
    },
    {
      "epoch": 1.2489728841413312,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008087618401745792,
      "loss": 4.92,
      "step": 9120
    },
    {
      "epoch": 1.249109832922487,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008087371098620831,
      "loss": 4.9351,
      "step": 9121
    },
    {
      "epoch": 1.2492467817036428,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008087123765766472,
      "loss": 4.9004,
      "step": 9122
    },
    {
      "epoch": 1.2493837304847988,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008086876403184763,
      "loss": 4.9496,
      "step": 9123
    },
    {
      "epoch": 1.2495206792659546,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008086629010877755,
      "loss": 4.9124,
      "step": 9124
    },
    {
      "epoch": 1.2496576280471103,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008086381588847498,
      "loss": 4.7523,
      "step": 9125
    },
    {
      "epoch": 1.2497945768282661,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008086134137096041,
      "loss": 4.9035,
      "step": 9126
    },
    {
      "epoch": 1.2499315256094221,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008085886655625437,
      "loss": 4.9717,
      "step": 9127
    },
    {
      "epoch": 1.2500684743905779,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008085639144437736,
      "loss": 4.8833,
      "step": 9128
    },
    {
      "epoch": 1.2502054231717339,
      "grad_norm": 0.375,
      "learning_rate": 0.0008085391603534988,
      "loss": 4.9116,
      "step": 9129
    },
    {
      "epoch": 1.2503423719528897,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008085144032919246,
      "loss": 4.8019,
      "step": 9130
    },
    {
      "epoch": 1.2504793207340454,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008084896432592561,
      "loss": 4.8623,
      "step": 9131
    },
    {
      "epoch": 1.2506162695152012,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008084648802556986,
      "loss": 4.891,
      "step": 9132
    },
    {
      "epoch": 1.2507532182963572,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000808440114281457,
      "loss": 4.895,
      "step": 9133
    },
    {
      "epoch": 1.250890167077513,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008084153453367369,
      "loss": 4.8826,
      "step": 9134
    },
    {
      "epoch": 1.251027115858669,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008083905734217434,
      "loss": 4.8379,
      "step": 9135
    },
    {
      "epoch": 1.2511640646398248,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008083657985366819,
      "loss": 4.8786,
      "step": 9136
    },
    {
      "epoch": 1.2513010134209805,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008083410206817575,
      "loss": 4.8825,
      "step": 9137
    },
    {
      "epoch": 1.2514379622021363,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008083162398571758,
      "loss": 4.9237,
      "step": 9138
    },
    {
      "epoch": 1.2515749109832923,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008082914560631419,
      "loss": 4.8689,
      "step": 9139
    },
    {
      "epoch": 1.251711859764448,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008082666692998615,
      "loss": 4.8606,
      "step": 9140
    },
    {
      "epoch": 1.2518488085456039,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008082418795675397,
      "loss": 4.9378,
      "step": 9141
    },
    {
      "epoch": 1.2519857573267599,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0008082170868663821,
      "loss": 4.9123,
      "step": 9142
    },
    {
      "epoch": 1.2521227061079157,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008081922911965942,
      "loss": 4.9001,
      "step": 9143
    },
    {
      "epoch": 1.2522596548890714,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008081674925583814,
      "loss": 4.8856,
      "step": 9144
    },
    {
      "epoch": 1.2523966036702272,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008081426909519492,
      "loss": 4.993,
      "step": 9145
    },
    {
      "epoch": 1.2525335524513832,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008081178863775031,
      "loss": 4.8453,
      "step": 9146
    },
    {
      "epoch": 1.252670501232539,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000808093078835249,
      "loss": 4.9033,
      "step": 9147
    },
    {
      "epoch": 1.252807450013695,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008080682683253921,
      "loss": 4.8811,
      "step": 9148
    },
    {
      "epoch": 1.2529443987948508,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000808043454848138,
      "loss": 4.9304,
      "step": 9149
    },
    {
      "epoch": 1.2530813475760065,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008080186384036926,
      "loss": 4.9752,
      "step": 9150
    },
    {
      "epoch": 1.2532182963571623,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008079938189922615,
      "loss": 4.939,
      "step": 9151
    },
    {
      "epoch": 1.2533552451383183,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008079689966140502,
      "loss": 4.8919,
      "step": 9152
    },
    {
      "epoch": 1.253492193919474,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008079441712692646,
      "loss": 4.8751,
      "step": 9153
    },
    {
      "epoch": 1.25362914270063,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008079193429581104,
      "loss": 4.9149,
      "step": 9154
    },
    {
      "epoch": 1.2537660914817859,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008078945116807932,
      "loss": 4.9167,
      "step": 9155
    },
    {
      "epoch": 1.2539030402629416,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008078696774375189,
      "loss": 4.875,
      "step": 9156
    },
    {
      "epoch": 1.2540399890440974,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008078448402284934,
      "loss": 4.8897,
      "step": 9157
    },
    {
      "epoch": 1.2541769378252534,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008078200000539224,
      "loss": 4.9327,
      "step": 9158
    },
    {
      "epoch": 1.2543138866064092,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008077951569140118,
      "loss": 4.887,
      "step": 9159
    },
    {
      "epoch": 1.2544508353875652,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008077703108089674,
      "loss": 4.7621,
      "step": 9160
    },
    {
      "epoch": 1.254587784168721,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008077454617389953,
      "loss": 4.9312,
      "step": 9161
    },
    {
      "epoch": 1.2547247329498767,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008077206097043012,
      "loss": 4.929,
      "step": 9162
    },
    {
      "epoch": 1.2548616817310325,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008076957547050912,
      "loss": 4.8435,
      "step": 9163
    },
    {
      "epoch": 1.2549986305121885,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008076708967415713,
      "loss": 4.8017,
      "step": 9164
    },
    {
      "epoch": 1.2551355792933443,
      "grad_norm": 0.375,
      "learning_rate": 0.0008076460358139475,
      "loss": 4.8608,
      "step": 9165
    },
    {
      "epoch": 1.2552725280745,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008076211719224256,
      "loss": 4.8614,
      "step": 9166
    },
    {
      "epoch": 1.255409476855656,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000807596305067212,
      "loss": 4.7981,
      "step": 9167
    },
    {
      "epoch": 1.2555464256368118,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008075714352485125,
      "loss": 4.8505,
      "step": 9168
    },
    {
      "epoch": 1.2556833744179676,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008075465624665333,
      "loss": 4.904,
      "step": 9169
    },
    {
      "epoch": 1.2558203231991234,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008075216867214807,
      "loss": 4.9444,
      "step": 9170
    },
    {
      "epoch": 1.2559572719802794,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008074968080135606,
      "loss": 4.919,
      "step": 9171
    },
    {
      "epoch": 1.2560942207614352,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008074719263429792,
      "loss": 4.8681,
      "step": 9172
    },
    {
      "epoch": 1.2562311695425912,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008074470417099429,
      "loss": 4.8756,
      "step": 9173
    },
    {
      "epoch": 1.256368118323747,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008074221541146576,
      "loss": 4.8121,
      "step": 9174
    },
    {
      "epoch": 1.2565050671049027,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008073972635573298,
      "loss": 4.8712,
      "step": 9175
    },
    {
      "epoch": 1.2566420158860585,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008073723700381658,
      "loss": 4.8236,
      "step": 9176
    },
    {
      "epoch": 1.2567789646672145,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008073474735573718,
      "loss": 4.8656,
      "step": 9177
    },
    {
      "epoch": 1.2569159134483703,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008073225741151541,
      "loss": 4.8177,
      "step": 9178
    },
    {
      "epoch": 1.2570528622295263,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000807297671711719,
      "loss": 4.8695,
      "step": 9179
    },
    {
      "epoch": 1.257189811010682,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008072727663472731,
      "loss": 4.8462,
      "step": 9180
    },
    {
      "epoch": 1.2573267597918378,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008072478580220227,
      "loss": 4.9995,
      "step": 9181
    },
    {
      "epoch": 1.2574637085729936,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000807222946736174,
      "loss": 4.9584,
      "step": 9182
    },
    {
      "epoch": 1.2576006573541496,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0008071980324899337,
      "loss": 4.9492,
      "step": 9183
    },
    {
      "epoch": 1.2577376061353054,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008071731152835081,
      "loss": 4.9325,
      "step": 9184
    },
    {
      "epoch": 1.2578745549164612,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008071481951171038,
      "loss": 4.9336,
      "step": 9185
    },
    {
      "epoch": 1.2580115036976172,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008071232719909274,
      "loss": 4.9755,
      "step": 9186
    },
    {
      "epoch": 1.258148452478773,
      "grad_norm": 0.359375,
      "learning_rate": 0.0008070983459051854,
      "loss": 4.9392,
      "step": 9187
    },
    {
      "epoch": 1.2582854012599287,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0008070734168600842,
      "loss": 4.891,
      "step": 9188
    },
    {
      "epoch": 1.2584223500410847,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008070484848558305,
      "loss": 4.9097,
      "step": 9189
    },
    {
      "epoch": 1.2585592988222405,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000807023549892631,
      "loss": 4.9299,
      "step": 9190
    },
    {
      "epoch": 1.2586962476033963,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008069986119706923,
      "loss": 4.864,
      "step": 9191
    },
    {
      "epoch": 1.2588331963845523,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008069736710902208,
      "loss": 4.903,
      "step": 9192
    },
    {
      "epoch": 1.258970145165708,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008069487272514236,
      "loss": 4.8041,
      "step": 9193
    },
    {
      "epoch": 1.2591070939468638,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008069237804545073,
      "loss": 4.9815,
      "step": 9194
    },
    {
      "epoch": 1.2592440427280196,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008068988306996785,
      "loss": 4.8264,
      "step": 9195
    },
    {
      "epoch": 1.2593809915091756,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0008068738779871441,
      "loss": 4.8072,
      "step": 9196
    },
    {
      "epoch": 1.2595179402903314,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008068489223171107,
      "loss": 4.8778,
      "step": 9197
    },
    {
      "epoch": 1.2596548890714874,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008068239636897854,
      "loss": 4.8579,
      "step": 9198
    },
    {
      "epoch": 1.2597918378526431,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008067990021053748,
      "loss": 4.8485,
      "step": 9199
    },
    {
      "epoch": 1.259928786633799,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000806774037564086,
      "loss": 4.7907,
      "step": 9200
    },
    {
      "epoch": 1.2600657354149547,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008067490700661255,
      "loss": 4.8987,
      "step": 9201
    },
    {
      "epoch": 1.2602026841961107,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008067240996117005,
      "loss": 4.9263,
      "step": 9202
    },
    {
      "epoch": 1.2603396329772665,
      "grad_norm": 0.494140625,
      "learning_rate": 0.000806699126201018,
      "loss": 4.8099,
      "step": 9203
    },
    {
      "epoch": 1.2604765817584225,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0008066741498342849,
      "loss": 4.8524,
      "step": 9204
    },
    {
      "epoch": 1.2606135305395783,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008066491705117078,
      "loss": 4.9535,
      "step": 9205
    },
    {
      "epoch": 1.260750479320734,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008066241882334944,
      "loss": 4.9545,
      "step": 9206
    },
    {
      "epoch": 1.2608874281018898,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008065992029998513,
      "loss": 4.8575,
      "step": 9207
    },
    {
      "epoch": 1.2610243768830458,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008065742148109854,
      "loss": 4.8626,
      "step": 9208
    },
    {
      "epoch": 1.2611613256642016,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008065492236671043,
      "loss": 4.9279,
      "step": 9209
    },
    {
      "epoch": 1.2612982744453574,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0008065242295684146,
      "loss": 4.8862,
      "step": 9210
    },
    {
      "epoch": 1.2614352232265134,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008064992325151238,
      "loss": 4.7612,
      "step": 9211
    },
    {
      "epoch": 1.2615721720076691,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0008064742325074389,
      "loss": 4.8482,
      "step": 9212
    },
    {
      "epoch": 1.261709120788825,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000806449229545567,
      "loss": 4.9222,
      "step": 9213
    },
    {
      "epoch": 1.261846069569981,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008064242236297154,
      "loss": 4.9271,
      "step": 9214
    },
    {
      "epoch": 1.2619830183511367,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008063992147600914,
      "loss": 4.805,
      "step": 9215
    },
    {
      "epoch": 1.2621199671322925,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008063742029369022,
      "loss": 4.8835,
      "step": 9216
    },
    {
      "epoch": 1.2622569159134485,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008063491881603551,
      "loss": 4.8513,
      "step": 9217
    },
    {
      "epoch": 1.2623938646946042,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008063241704306572,
      "loss": 4.8672,
      "step": 9218
    },
    {
      "epoch": 1.26253081347576,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008062991497480161,
      "loss": 4.9012,
      "step": 9219
    },
    {
      "epoch": 1.2626677622569158,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000806274126112639,
      "loss": 4.9102,
      "step": 9220
    },
    {
      "epoch": 1.2628047110380718,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008062490995247334,
      "loss": 4.8616,
      "step": 9221
    },
    {
      "epoch": 1.2629416598192276,
      "grad_norm": 0.375,
      "learning_rate": 0.0008062240699845065,
      "loss": 4.888,
      "step": 9222
    },
    {
      "epoch": 1.2630786086003836,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008061990374921659,
      "loss": 4.9203,
      "step": 9223
    },
    {
      "epoch": 1.2632155573815393,
      "grad_norm": 0.375,
      "learning_rate": 0.000806174002047919,
      "loss": 4.8792,
      "step": 9224
    },
    {
      "epoch": 1.2633525061626951,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0008061489636519733,
      "loss": 4.9165,
      "step": 9225
    },
    {
      "epoch": 1.263489454943851,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0008061239223045363,
      "loss": 4.9598,
      "step": 9226
    },
    {
      "epoch": 1.263626403725007,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008060988780058155,
      "loss": 4.7861,
      "step": 9227
    },
    {
      "epoch": 1.2637633525061627,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008060738307560184,
      "loss": 4.992,
      "step": 9228
    },
    {
      "epoch": 1.2639003012873187,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0008060487805553525,
      "loss": 4.8596,
      "step": 9229
    },
    {
      "epoch": 1.2640372500684744,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008060237274040257,
      "loss": 4.8751,
      "step": 9230
    },
    {
      "epoch": 1.2641741988496302,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0008059986713022454,
      "loss": 4.8903,
      "step": 9231
    },
    {
      "epoch": 1.264311147630786,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008059736122502192,
      "loss": 4.8411,
      "step": 9232
    },
    {
      "epoch": 1.264448096411942,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000805948550248155,
      "loss": 4.9118,
      "step": 9233
    },
    {
      "epoch": 1.2645850451930978,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0008059234852962602,
      "loss": 4.798,
      "step": 9234
    },
    {
      "epoch": 1.2647219939742536,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008058984173947427,
      "loss": 4.8825,
      "step": 9235
    },
    {
      "epoch": 1.2648589427554096,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008058733465438102,
      "loss": 4.8936,
      "step": 9236
    },
    {
      "epoch": 1.2649958915365653,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008058482727436705,
      "loss": 4.9171,
      "step": 9237
    },
    {
      "epoch": 1.265132840317721,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008058231959945314,
      "loss": 4.8711,
      "step": 9238
    },
    {
      "epoch": 1.2652697890988769,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008057981162966006,
      "loss": 4.9382,
      "step": 9239
    },
    {
      "epoch": 1.2654067378800329,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000805773033650086,
      "loss": 4.8842,
      "step": 9240
    },
    {
      "epoch": 1.2655436866611887,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008057479480551957,
      "loss": 4.8304,
      "step": 9241
    },
    {
      "epoch": 1.2656806354423447,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008057228595121372,
      "loss": 4.7955,
      "step": 9242
    },
    {
      "epoch": 1.2658175842235004,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008056977680211187,
      "loss": 4.9201,
      "step": 9243
    },
    {
      "epoch": 1.2659545330046562,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000805672673582348,
      "loss": 4.9312,
      "step": 9244
    },
    {
      "epoch": 1.266091481785812,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000805647576196033,
      "loss": 4.9365,
      "step": 9245
    },
    {
      "epoch": 1.266228430566968,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000805622475862382,
      "loss": 4.8761,
      "step": 9246
    },
    {
      "epoch": 1.2663653793481238,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008055973725816027,
      "loss": 4.8839,
      "step": 9247
    },
    {
      "epoch": 1.2665023281292798,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008055722663539032,
      "loss": 4.8661,
      "step": 9248
    },
    {
      "epoch": 1.2666392769104355,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008055471571794916,
      "loss": 4.867,
      "step": 9249
    },
    {
      "epoch": 1.2667762256915913,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000805522045058576,
      "loss": 4.857,
      "step": 9250
    },
    {
      "epoch": 1.266913174472747,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008054969299913646,
      "loss": 4.8227,
      "step": 9251
    },
    {
      "epoch": 1.267050123253903,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008054718119780653,
      "loss": 4.8995,
      "step": 9252
    },
    {
      "epoch": 1.2671870720350589,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008054466910188865,
      "loss": 4.8413,
      "step": 9253
    },
    {
      "epoch": 1.2673240208162149,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008054215671140362,
      "loss": 4.8643,
      "step": 9254
    },
    {
      "epoch": 1.2674609695973706,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008053964402637226,
      "loss": 4.8989,
      "step": 9255
    },
    {
      "epoch": 1.2675979183785264,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008053713104681542,
      "loss": 4.907,
      "step": 9256
    },
    {
      "epoch": 1.2677348671596822,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008053461777275388,
      "loss": 4.8943,
      "step": 9257
    },
    {
      "epoch": 1.2678718159408382,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008053210420420852,
      "loss": 4.9755,
      "step": 9258
    },
    {
      "epoch": 1.268008764721994,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008052959034120015,
      "loss": 4.9374,
      "step": 9259
    },
    {
      "epoch": 1.2681457135031498,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008052707618374957,
      "loss": 4.9552,
      "step": 9260
    },
    {
      "epoch": 1.2682826622843058,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008052456173187766,
      "loss": 4.8651,
      "step": 9261
    },
    {
      "epoch": 1.2684196110654615,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008052204698560525,
      "loss": 4.7996,
      "step": 9262
    },
    {
      "epoch": 1.2685565598466173,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008051953194495316,
      "loss": 4.894,
      "step": 9263
    },
    {
      "epoch": 1.268693508627773,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008051701660994225,
      "loss": 4.9137,
      "step": 9264
    },
    {
      "epoch": 1.268830457408929,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008051450098059337,
      "loss": 4.9072,
      "step": 9265
    },
    {
      "epoch": 1.2689674061900849,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008051198505692734,
      "loss": 4.9329,
      "step": 9266
    },
    {
      "epoch": 1.2691043549712409,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008050946883896504,
      "loss": 4.9145,
      "step": 9267
    },
    {
      "epoch": 1.2692413037523966,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008050695232672731,
      "loss": 4.8401,
      "step": 9268
    },
    {
      "epoch": 1.2693782525335524,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0008050443552023499,
      "loss": 4.8785,
      "step": 9269
    },
    {
      "epoch": 1.2695152013147082,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008050191841950897,
      "loss": 4.8799,
      "step": 9270
    },
    {
      "epoch": 1.2696521500958642,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008049940102457007,
      "loss": 4.8071,
      "step": 9271
    },
    {
      "epoch": 1.26978909887702,
      "grad_norm": 0.375,
      "learning_rate": 0.000804968833354392,
      "loss": 4.9292,
      "step": 9272
    },
    {
      "epoch": 1.269926047658176,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008049436535213719,
      "loss": 4.8054,
      "step": 9273
    },
    {
      "epoch": 1.2700629964393317,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008049184707468492,
      "loss": 4.9052,
      "step": 9274
    },
    {
      "epoch": 1.2701999452204875,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0008048932850310325,
      "loss": 4.8467,
      "step": 9275
    },
    {
      "epoch": 1.2703368940016433,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0008048680963741305,
      "loss": 4.8196,
      "step": 9276
    },
    {
      "epoch": 1.2704738427827993,
      "grad_norm": 0.375,
      "learning_rate": 0.0008048429047763522,
      "loss": 4.7692,
      "step": 9277
    },
    {
      "epoch": 1.270610791563955,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008048177102379061,
      "loss": 4.9585,
      "step": 9278
    },
    {
      "epoch": 1.270747740345111,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008047925127590011,
      "loss": 4.8265,
      "step": 9279
    },
    {
      "epoch": 1.2708846891262668,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000804767312339846,
      "loss": 4.8528,
      "step": 9280
    },
    {
      "epoch": 1.2710216379074226,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008047421089806497,
      "loss": 4.8622,
      "step": 9281
    },
    {
      "epoch": 1.2711585866885784,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008047169026816209,
      "loss": 4.8717,
      "step": 9282
    },
    {
      "epoch": 1.2712955354697344,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008046916934429687,
      "loss": 4.8724,
      "step": 9283
    },
    {
      "epoch": 1.2714324842508902,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000804666481264902,
      "loss": 4.8506,
      "step": 9284
    },
    {
      "epoch": 1.271569433032046,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0008046412661476295,
      "loss": 4.8265,
      "step": 9285
    },
    {
      "epoch": 1.271706381813202,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008046160480913603,
      "loss": 4.8384,
      "step": 9286
    },
    {
      "epoch": 1.2718433305943577,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008045908270963034,
      "loss": 4.8641,
      "step": 9287
    },
    {
      "epoch": 1.2719802793755135,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008045656031626681,
      "loss": 4.8693,
      "step": 9288
    },
    {
      "epoch": 1.2721172281566693,
      "grad_norm": 0.453125,
      "learning_rate": 0.000804540376290663,
      "loss": 4.825,
      "step": 9289
    },
    {
      "epoch": 1.2722541769378253,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0008045151464804971,
      "loss": 4.8777,
      "step": 9290
    },
    {
      "epoch": 1.272391125718981,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008044899137323799,
      "loss": 4.8656,
      "step": 9291
    },
    {
      "epoch": 1.272528074500137,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008044646780465204,
      "loss": 4.8959,
      "step": 9292
    },
    {
      "epoch": 1.2726650232812928,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008044394394231276,
      "loss": 4.8529,
      "step": 9293
    },
    {
      "epoch": 1.2728019720624486,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008044141978624105,
      "loss": 4.8746,
      "step": 9294
    },
    {
      "epoch": 1.2729389208436044,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008043889533645786,
      "loss": 4.7964,
      "step": 9295
    },
    {
      "epoch": 1.2730758696247604,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008043637059298412,
      "loss": 4.9019,
      "step": 9296
    },
    {
      "epoch": 1.2732128184059162,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0008043384555584071,
      "loss": 4.8793,
      "step": 9297
    },
    {
      "epoch": 1.2733497671870722,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008043132022504857,
      "loss": 4.9204,
      "step": 9298
    },
    {
      "epoch": 1.273486715968228,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008042879460062866,
      "loss": 4.9049,
      "step": 9299
    },
    {
      "epoch": 1.2736236647493837,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008042626868260188,
      "loss": 4.8928,
      "step": 9300
    },
    {
      "epoch": 1.2737606135305395,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008042374247098918,
      "loss": 4.9433,
      "step": 9301
    },
    {
      "epoch": 1.2738975623116955,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0008042121596581146,
      "loss": 4.8204,
      "step": 9302
    },
    {
      "epoch": 1.2740345110928513,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008041868916708971,
      "loss": 4.9253,
      "step": 9303
    },
    {
      "epoch": 1.274171459874007,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0008041616207484483,
      "loss": 4.941,
      "step": 9304
    },
    {
      "epoch": 1.274308408655163,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008041363468909776,
      "loss": 4.8891,
      "step": 9305
    },
    {
      "epoch": 1.2744453574363188,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008041110700986948,
      "loss": 4.8526,
      "step": 9306
    },
    {
      "epoch": 1.2745823062174746,
      "grad_norm": 0.375,
      "learning_rate": 0.0008040857903718091,
      "loss": 4.8599,
      "step": 9307
    },
    {
      "epoch": 1.2747192549986306,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008040605077105302,
      "loss": 4.8491,
      "step": 9308
    },
    {
      "epoch": 1.2748562037797864,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008040352221150675,
      "loss": 4.9375,
      "step": 9309
    },
    {
      "epoch": 1.2749931525609421,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008040099335856304,
      "loss": 4.8495,
      "step": 9310
    },
    {
      "epoch": 1.2751301013420981,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008039846421224287,
      "loss": 4.9843,
      "step": 9311
    },
    {
      "epoch": 1.275267050123254,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000803959347725672,
      "loss": 4.9402,
      "step": 9312
    },
    {
      "epoch": 1.2754039989044097,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008039340503955696,
      "loss": 4.7867,
      "step": 9313
    },
    {
      "epoch": 1.2755409476855655,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008039087501323316,
      "loss": 4.8695,
      "step": 9314
    },
    {
      "epoch": 1.2756778964667215,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008038834469361675,
      "loss": 4.8658,
      "step": 9315
    },
    {
      "epoch": 1.2758148452478772,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008038581408072869,
      "loss": 4.8584,
      "step": 9316
    },
    {
      "epoch": 1.2759517940290332,
      "grad_norm": 0.375,
      "learning_rate": 0.0008038328317458995,
      "loss": 4.8917,
      "step": 9317
    },
    {
      "epoch": 1.276088742810189,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008038075197522151,
      "loss": 4.8855,
      "step": 9318
    },
    {
      "epoch": 1.2762256915913448,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008037822048264435,
      "loss": 4.9563,
      "step": 9319
    },
    {
      "epoch": 1.2763626403725006,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0008037568869687946,
      "loss": 4.8041,
      "step": 9320
    },
    {
      "epoch": 1.2764995891536566,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008037315661794779,
      "loss": 4.8636,
      "step": 9321
    },
    {
      "epoch": 1.2766365379348124,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008037062424587035,
      "loss": 4.916,
      "step": 9322
    },
    {
      "epoch": 1.2767734867159684,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008036809158066811,
      "loss": 4.894,
      "step": 9323
    },
    {
      "epoch": 1.2769104354971241,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008036555862236208,
      "loss": 4.8305,
      "step": 9324
    },
    {
      "epoch": 1.27704738427828,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008036302537097322,
      "loss": 4.8909,
      "step": 9325
    },
    {
      "epoch": 1.2771843330594357,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008036049182652257,
      "loss": 4.8888,
      "step": 9326
    },
    {
      "epoch": 1.2773212818405917,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008035795798903108,
      "loss": 4.882,
      "step": 9327
    },
    {
      "epoch": 1.2774582306217475,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008035542385851977,
      "loss": 4.8169,
      "step": 9328
    },
    {
      "epoch": 1.2775951794029032,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0008035288943500963,
      "loss": 4.8787,
      "step": 9329
    },
    {
      "epoch": 1.2777321281840592,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008035035471852168,
      "loss": 4.7987,
      "step": 9330
    },
    {
      "epoch": 1.277869076965215,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008034781970907692,
      "loss": 4.7758,
      "step": 9331
    },
    {
      "epoch": 1.2780060257463708,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008034528440669635,
      "loss": 4.9028,
      "step": 9332
    },
    {
      "epoch": 1.2781429745275268,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0008034274881140098,
      "loss": 4.8708,
      "step": 9333
    },
    {
      "epoch": 1.2782799233086826,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008034021292321184,
      "loss": 4.8454,
      "step": 9334
    },
    {
      "epoch": 1.2784168720898383,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008033767674214992,
      "loss": 4.8647,
      "step": 9335
    },
    {
      "epoch": 1.2785538208709943,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008033514026823627,
      "loss": 4.8583,
      "step": 9336
    },
    {
      "epoch": 1.2786907696521501,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000803326035014919,
      "loss": 4.9332,
      "step": 9337
    },
    {
      "epoch": 1.278827718433306,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000803300664419378,
      "loss": 4.8664,
      "step": 9338
    },
    {
      "epoch": 1.2789646672144617,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008032752908959503,
      "loss": 4.9245,
      "step": 9339
    },
    {
      "epoch": 1.2791016159956177,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000803249914444846,
      "loss": 4.9246,
      "step": 9340
    },
    {
      "epoch": 1.2792385647767734,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008032245350662755,
      "loss": 4.8663,
      "step": 9341
    },
    {
      "epoch": 1.2793755135579294,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008031991527604492,
      "loss": 4.9093,
      "step": 9342
    },
    {
      "epoch": 1.2795124623390852,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008031737675275774,
      "loss": 4.8901,
      "step": 9343
    },
    {
      "epoch": 1.279649411120241,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008031483793678705,
      "loss": 4.9399,
      "step": 9344
    },
    {
      "epoch": 1.2797863599013968,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008031229882815386,
      "loss": 4.8916,
      "step": 9345
    },
    {
      "epoch": 1.2799233086825528,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0008030975942687925,
      "loss": 4.8857,
      "step": 9346
    },
    {
      "epoch": 1.2800602574637086,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0008030721973298425,
      "loss": 4.8931,
      "step": 9347
    },
    {
      "epoch": 1.2801972062448645,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000803046797464899,
      "loss": 4.8688,
      "step": 9348
    },
    {
      "epoch": 1.2803341550260203,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008030213946741726,
      "loss": 4.9072,
      "step": 9349
    },
    {
      "epoch": 1.280471103807176,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008029959889578737,
      "loss": 4.974,
      "step": 9350
    },
    {
      "epoch": 1.2806080525883319,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008029705803162131,
      "loss": 4.8504,
      "step": 9351
    },
    {
      "epoch": 1.2807450013694879,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000802945168749401,
      "loss": 4.8377,
      "step": 9352
    },
    {
      "epoch": 1.2808819501506437,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008029197542576483,
      "loss": 4.902,
      "step": 9353
    },
    {
      "epoch": 1.2810188989317994,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008028943368411655,
      "loss": 4.91,
      "step": 9354
    },
    {
      "epoch": 1.2811558477129554,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008028689165001632,
      "loss": 4.8095,
      "step": 9355
    },
    {
      "epoch": 1.2812927964941112,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0008028434932348522,
      "loss": 4.8307,
      "step": 9356
    },
    {
      "epoch": 1.281429745275267,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008028180670454429,
      "loss": 4.9132,
      "step": 9357
    },
    {
      "epoch": 1.2815666940564228,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008027926379321463,
      "loss": 4.9817,
      "step": 9358
    },
    {
      "epoch": 1.2817036428375788,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008027672058951731,
      "loss": 4.9638,
      "step": 9359
    },
    {
      "epoch": 1.2818405916187345,
      "grad_norm": 0.421875,
      "learning_rate": 0.000802741770934734,
      "loss": 4.8484,
      "step": 9360
    },
    {
      "epoch": 1.2819775403998905,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008027163330510397,
      "loss": 4.8652,
      "step": 9361
    },
    {
      "epoch": 1.2821144891810463,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0008026908922443012,
      "loss": 4.9336,
      "step": 9362
    },
    {
      "epoch": 1.282251437962202,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008026654485147292,
      "loss": 4.8672,
      "step": 9363
    },
    {
      "epoch": 1.2823883867433579,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008026400018625346,
      "loss": 4.8219,
      "step": 9364
    },
    {
      "epoch": 1.2825253355245139,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008026145522879283,
      "loss": 4.9203,
      "step": 9365
    },
    {
      "epoch": 1.2826622843056696,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000802589099791121,
      "loss": 4.8233,
      "step": 9366
    },
    {
      "epoch": 1.2827992330868256,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000802563644372324,
      "loss": 4.8839,
      "step": 9367
    },
    {
      "epoch": 1.2829361818679814,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0008025381860317479,
      "loss": 4.8558,
      "step": 9368
    },
    {
      "epoch": 1.2830731306491372,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000802512724769604,
      "loss": 4.8835,
      "step": 9369
    },
    {
      "epoch": 1.283210079430293,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008024872605861031,
      "loss": 4.9143,
      "step": 9370
    },
    {
      "epoch": 1.283347028211449,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008024617934814564,
      "loss": 4.8669,
      "step": 9371
    },
    {
      "epoch": 1.2834839769926047,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008024363234558747,
      "loss": 4.8613,
      "step": 9372
    },
    {
      "epoch": 1.2836209257737607,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008024108505095692,
      "loss": 4.9402,
      "step": 9373
    },
    {
      "epoch": 1.2837578745549165,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008023853746427511,
      "loss": 4.844,
      "step": 9374
    },
    {
      "epoch": 1.2838948233360723,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008023598958556313,
      "loss": 4.84,
      "step": 9375
    },
    {
      "epoch": 1.284031772117228,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008023344141484212,
      "loss": 4.8697,
      "step": 9376
    },
    {
      "epoch": 1.284168720898384,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008023089295213318,
      "loss": 4.7768,
      "step": 9377
    },
    {
      "epoch": 1.2843056696795399,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008022834419745743,
      "loss": 4.9422,
      "step": 9378
    },
    {
      "epoch": 1.2844426184606956,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0008022579515083601,
      "loss": 4.811,
      "step": 9379
    },
    {
      "epoch": 1.2845795672418516,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008022324581229001,
      "loss": 4.8987,
      "step": 9380
    },
    {
      "epoch": 1.2847165160230074,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0008022069618184061,
      "loss": 4.8881,
      "step": 9381
    },
    {
      "epoch": 1.2848534648041632,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008021814625950889,
      "loss": 4.8584,
      "step": 9382
    },
    {
      "epoch": 1.284990413585319,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00080215596045316,
      "loss": 4.8853,
      "step": 9383
    },
    {
      "epoch": 1.285127362366475,
      "grad_norm": 0.453125,
      "learning_rate": 0.0008021304553928307,
      "loss": 4.8519,
      "step": 9384
    },
    {
      "epoch": 1.2852643111476307,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008021049474143126,
      "loss": 4.9237,
      "step": 9385
    },
    {
      "epoch": 1.2854012599287867,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008020794365178166,
      "loss": 4.8663,
      "step": 9386
    },
    {
      "epoch": 1.2855382087099425,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008020539227035546,
      "loss": 4.8822,
      "step": 9387
    },
    {
      "epoch": 1.2856751574910983,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008020284059717379,
      "loss": 4.9531,
      "step": 9388
    },
    {
      "epoch": 1.285812106272254,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008020028863225778,
      "loss": 4.859,
      "step": 9389
    },
    {
      "epoch": 1.28594905505341,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000801977363756286,
      "loss": 4.8663,
      "step": 9390
    },
    {
      "epoch": 1.2860860038345658,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008019518382730738,
      "loss": 4.8701,
      "step": 9391
    },
    {
      "epoch": 1.2862229526157218,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0008019263098731529,
      "loss": 4.8369,
      "step": 9392
    },
    {
      "epoch": 1.2863599013968776,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008019007785567348,
      "loss": 4.8668,
      "step": 9393
    },
    {
      "epoch": 1.2864968501780334,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008018752443240311,
      "loss": 4.8282,
      "step": 9394
    },
    {
      "epoch": 1.2866337989591892,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008018497071752535,
      "loss": 4.8652,
      "step": 9395
    },
    {
      "epoch": 1.2867707477403452,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008018241671106135,
      "loss": 4.8587,
      "step": 9396
    },
    {
      "epoch": 1.286907696521501,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0008017986241303227,
      "loss": 4.8726,
      "step": 9397
    },
    {
      "epoch": 1.287044645302657,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008017730782345928,
      "loss": 4.9132,
      "step": 9398
    },
    {
      "epoch": 1.2871815940838127,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008017475294236357,
      "loss": 4.9956,
      "step": 9399
    },
    {
      "epoch": 1.2873185428649685,
      "grad_norm": 0.484375,
      "learning_rate": 0.0008017219776976631,
      "loss": 4.8733,
      "step": 9400
    },
    {
      "epoch": 1.2874554916461243,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0008016964230568866,
      "loss": 4.9801,
      "step": 9401
    },
    {
      "epoch": 1.2875924404272803,
      "grad_norm": 0.52734375,
      "learning_rate": 0.000801670865501518,
      "loss": 4.8212,
      "step": 9402
    },
    {
      "epoch": 1.287729389208436,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0008016453050317691,
      "loss": 4.8913,
      "step": 9403
    },
    {
      "epoch": 1.2878663379895918,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0008016197416478518,
      "loss": 4.8555,
      "step": 9404
    },
    {
      "epoch": 1.2880032867707478,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0008015941753499779,
      "loss": 4.8121,
      "step": 9405
    },
    {
      "epoch": 1.2881402355519036,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0008015686061383593,
      "loss": 4.8913,
      "step": 9406
    },
    {
      "epoch": 1.2882771843330594,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008015430340132079,
      "loss": 4.7533,
      "step": 9407
    },
    {
      "epoch": 1.2884141331142152,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008015174589747356,
      "loss": 4.9444,
      "step": 9408
    },
    {
      "epoch": 1.2885510818953712,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008014918810231544,
      "loss": 4.9247,
      "step": 9409
    },
    {
      "epoch": 1.288688030676527,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000801466300158676,
      "loss": 4.8211,
      "step": 9410
    },
    {
      "epoch": 1.288824979457683,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008014407163815129,
      "loss": 4.9053,
      "step": 9411
    },
    {
      "epoch": 1.2889619282388387,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0008014151296918768,
      "loss": 4.8588,
      "step": 9412
    },
    {
      "epoch": 1.2890988770199945,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0008013895400899798,
      "loss": 4.8601,
      "step": 9413
    },
    {
      "epoch": 1.2892358258011503,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008013639475760338,
      "loss": 4.8697,
      "step": 9414
    },
    {
      "epoch": 1.2893727745823063,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0008013383521502511,
      "loss": 4.8443,
      "step": 9415
    },
    {
      "epoch": 1.289509723363462,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008013127538128437,
      "loss": 4.8298,
      "step": 9416
    },
    {
      "epoch": 1.289646672144618,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0008012871525640239,
      "loss": 4.8642,
      "step": 9417
    },
    {
      "epoch": 1.2897836209257738,
      "grad_norm": 0.5,
      "learning_rate": 0.0008012615484040038,
      "loss": 4.9045,
      "step": 9418
    },
    {
      "epoch": 1.2899205697069296,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0008012359413329954,
      "loss": 4.876,
      "step": 9419
    },
    {
      "epoch": 1.2900575184880854,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008012103313512111,
      "loss": 4.9008,
      "step": 9420
    },
    {
      "epoch": 1.2901944672692414,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008011847184588633,
      "loss": 4.8964,
      "step": 9421
    },
    {
      "epoch": 1.2903314160503971,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008011591026561638,
      "loss": 4.8961,
      "step": 9422
    },
    {
      "epoch": 1.290468364831553,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008011334839433253,
      "loss": 4.8429,
      "step": 9423
    },
    {
      "epoch": 1.290605313612709,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0008011078623205598,
      "loss": 4.877,
      "step": 9424
    },
    {
      "epoch": 1.2907422623938647,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0008010822377880799,
      "loss": 4.8206,
      "step": 9425
    },
    {
      "epoch": 1.2908792111750205,
      "grad_norm": 0.34375,
      "learning_rate": 0.0008010566103460977,
      "loss": 4.8797,
      "step": 9426
    },
    {
      "epoch": 1.2910161599561765,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008010309799948258,
      "loss": 4.8067,
      "step": 9427
    },
    {
      "epoch": 1.2911531087373322,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0008010053467344767,
      "loss": 4.8893,
      "step": 9428
    },
    {
      "epoch": 1.291290057518488,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008009797105652624,
      "loss": 4.8243,
      "step": 9429
    },
    {
      "epoch": 1.291427006299644,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008009540714873957,
      "loss": 4.9174,
      "step": 9430
    },
    {
      "epoch": 1.2915639550807998,
      "grad_norm": 0.375,
      "learning_rate": 0.0008009284295010889,
      "loss": 4.8505,
      "step": 9431
    },
    {
      "epoch": 1.2917009038619556,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0008009027846065548,
      "loss": 4.8738,
      "step": 9432
    },
    {
      "epoch": 1.2918378526431114,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008008771368040055,
      "loss": 4.885,
      "step": 9433
    },
    {
      "epoch": 1.2919748014242673,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008008514860936538,
      "loss": 4.8435,
      "step": 9434
    },
    {
      "epoch": 1.2921117502054231,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0008008258324757122,
      "loss": 4.9261,
      "step": 9435
    },
    {
      "epoch": 1.2922486989865791,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008008001759503934,
      "loss": 4.9223,
      "step": 9436
    },
    {
      "epoch": 1.292385647767735,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0008007745165179099,
      "loss": 4.8525,
      "step": 9437
    },
    {
      "epoch": 1.2925225965488907,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0008007488541784745,
      "loss": 4.8673,
      "step": 9438
    },
    {
      "epoch": 1.2926595453300465,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0008007231889322996,
      "loss": 4.8442,
      "step": 9439
    },
    {
      "epoch": 1.2927964941112025,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0008006975207795982,
      "loss": 4.842,
      "step": 9440
    },
    {
      "epoch": 1.2929334428923582,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008006718497205828,
      "loss": 4.9661,
      "step": 9441
    },
    {
      "epoch": 1.2930703916735142,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008006461757554663,
      "loss": 4.9265,
      "step": 9442
    },
    {
      "epoch": 1.29320734045467,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008006204988844613,
      "loss": 4.8264,
      "step": 9443
    },
    {
      "epoch": 1.2933442892358258,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0008005948191077808,
      "loss": 4.9043,
      "step": 9444
    },
    {
      "epoch": 1.2934812380169816,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008005691364256375,
      "loss": 4.9238,
      "step": 9445
    },
    {
      "epoch": 1.2936181867981376,
      "grad_norm": 0.443359375,
      "learning_rate": 0.000800543450838244,
      "loss": 4.861,
      "step": 9446
    },
    {
      "epoch": 1.2937551355792933,
      "grad_norm": 0.4375,
      "learning_rate": 0.0008005177623458136,
      "loss": 4.9613,
      "step": 9447
    },
    {
      "epoch": 1.2938920843604491,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0008004920709485589,
      "loss": 4.866,
      "step": 9448
    },
    {
      "epoch": 1.2940290331416051,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008004663766466928,
      "loss": 4.935,
      "step": 9449
    },
    {
      "epoch": 1.2941659819227609,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0008004406794404285,
      "loss": 4.8807,
      "step": 9450
    },
    {
      "epoch": 1.2943029307039167,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0008004149793299787,
      "loss": 4.8554,
      "step": 9451
    },
    {
      "epoch": 1.2944398794850727,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0008003892763155566,
      "loss": 4.8502,
      "step": 9452
    },
    {
      "epoch": 1.2945768282662284,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008003635703973748,
      "loss": 4.9018,
      "step": 9453
    },
    {
      "epoch": 1.2947137770473842,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0008003378615756468,
      "loss": 4.8612,
      "step": 9454
    },
    {
      "epoch": 1.2948507258285402,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0008003121498505855,
      "loss": 4.9042,
      "step": 9455
    },
    {
      "epoch": 1.294987674609696,
      "grad_norm": 0.421875,
      "learning_rate": 0.0008002864352224038,
      "loss": 4.8079,
      "step": 9456
    },
    {
      "epoch": 1.2951246233908518,
      "grad_norm": 0.390625,
      "learning_rate": 0.000800260717691315,
      "loss": 4.8711,
      "step": 9457
    },
    {
      "epoch": 1.2952615721720075,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0008002349972575322,
      "loss": 4.8926,
      "step": 9458
    },
    {
      "epoch": 1.2953985209531635,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0008002092739212684,
      "loss": 4.8046,
      "step": 9459
    },
    {
      "epoch": 1.2955354697343193,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000800183547682737,
      "loss": 4.8228,
      "step": 9460
    },
    {
      "epoch": 1.2956724185154753,
      "grad_norm": 0.40625,
      "learning_rate": 0.0008001578185421512,
      "loss": 4.8454,
      "step": 9461
    },
    {
      "epoch": 1.295809367296631,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000800132086499724,
      "loss": 4.9409,
      "step": 9462
    },
    {
      "epoch": 1.2959463160777869,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0008001063515556687,
      "loss": 4.8609,
      "step": 9463
    },
    {
      "epoch": 1.2960832648589427,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0008000806137101988,
      "loss": 4.838,
      "step": 9464
    },
    {
      "epoch": 1.2962202136400987,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0008000548729635272,
      "loss": 4.8291,
      "step": 9465
    },
    {
      "epoch": 1.2963571624212544,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0008000291293158678,
      "loss": 4.9372,
      "step": 9466
    },
    {
      "epoch": 1.2964941112024104,
      "grad_norm": 0.390625,
      "learning_rate": 0.0008000033827674333,
      "loss": 4.9246,
      "step": 9467
    },
    {
      "epoch": 1.2966310599835662,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007999776333184375,
      "loss": 4.8636,
      "step": 9468
    },
    {
      "epoch": 1.296768008764722,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007999518809690936,
      "loss": 4.9699,
      "step": 9469
    },
    {
      "epoch": 1.2969049575458778,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000799926125719615,
      "loss": 4.9406,
      "step": 9470
    },
    {
      "epoch": 1.2970419063270338,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007999003675702153,
      "loss": 4.8388,
      "step": 9471
    },
    {
      "epoch": 1.2971788551081895,
      "grad_norm": 0.359375,
      "learning_rate": 0.000799874606521108,
      "loss": 4.9498,
      "step": 9472
    },
    {
      "epoch": 1.2973158038893453,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007998488425725063,
      "loss": 4.846,
      "step": 9473
    },
    {
      "epoch": 1.2974527526705013,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007998230757246239,
      "loss": 4.9013,
      "step": 9474
    },
    {
      "epoch": 1.297589701451657,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007997973059776744,
      "loss": 4.8489,
      "step": 9475
    },
    {
      "epoch": 1.2977266502328129,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007997715333318712,
      "loss": 4.8985,
      "step": 9476
    },
    {
      "epoch": 1.2978635990139686,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000799745757787428,
      "loss": 4.9423,
      "step": 9477
    },
    {
      "epoch": 1.2980005477951246,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007997199793445583,
      "loss": 4.9184,
      "step": 9478
    },
    {
      "epoch": 1.2981374965762804,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007996941980034758,
      "loss": 4.9066,
      "step": 9479
    },
    {
      "epoch": 1.2982744453574364,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007996684137643942,
      "loss": 4.8598,
      "step": 9480
    },
    {
      "epoch": 1.2984113941385922,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000799642626627527,
      "loss": 4.9064,
      "step": 9481
    },
    {
      "epoch": 1.298548342919748,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007996168365930883,
      "loss": 4.8825,
      "step": 9482
    },
    {
      "epoch": 1.2986852917009037,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007995910436612915,
      "loss": 4.8805,
      "step": 9483
    },
    {
      "epoch": 1.2988222404820597,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007995652478323504,
      "loss": 4.8201,
      "step": 9484
    },
    {
      "epoch": 1.2989591892632155,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007995394491064786,
      "loss": 4.9274,
      "step": 9485
    },
    {
      "epoch": 1.2990961380443715,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007995136474838902,
      "loss": 4.8395,
      "step": 9486
    },
    {
      "epoch": 1.2992330868255273,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000799487842964799,
      "loss": 4.819,
      "step": 9487
    },
    {
      "epoch": 1.299370035606683,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007994620355494187,
      "loss": 4.8473,
      "step": 9488
    },
    {
      "epoch": 1.2995069843878388,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007994362252379633,
      "loss": 4.8841,
      "step": 9489
    },
    {
      "epoch": 1.2996439331689948,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0007994104120306464,
      "loss": 4.8779,
      "step": 9490
    },
    {
      "epoch": 1.2997808819501506,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007993845959276825,
      "loss": 4.8749,
      "step": 9491
    },
    {
      "epoch": 1.2999178307313066,
      "grad_norm": 0.47265625,
      "learning_rate": 0.000799358776929285,
      "loss": 4.8624,
      "step": 9492
    },
    {
      "epoch": 1.3000547795124624,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007993329550356681,
      "loss": 4.8163,
      "step": 9493
    },
    {
      "epoch": 1.3001917282936182,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007993071302470459,
      "loss": 4.8176,
      "step": 9494
    },
    {
      "epoch": 1.300328677074774,
      "grad_norm": 0.48828125,
      "learning_rate": 0.000799281302563632,
      "loss": 4.8906,
      "step": 9495
    },
    {
      "epoch": 1.30046562585593,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000799255471985641,
      "loss": 4.8974,
      "step": 9496
    },
    {
      "epoch": 1.3006025746370857,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007992296385132866,
      "loss": 4.9126,
      "step": 9497
    },
    {
      "epoch": 1.3007395234182415,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007992038021467828,
      "loss": 4.8606,
      "step": 9498
    },
    {
      "epoch": 1.3008764721993975,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000799177962886344,
      "loss": 4.8957,
      "step": 9499
    },
    {
      "epoch": 1.3010134209805533,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0007991521207321843,
      "loss": 4.874,
      "step": 9500
    },
    {
      "epoch": 1.301150369761709,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007991262756845176,
      "loss": 4.8269,
      "step": 9501
    },
    {
      "epoch": 1.3012873185428648,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007991004277435584,
      "loss": 4.9444,
      "step": 9502
    },
    {
      "epoch": 1.3014242673240208,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007990745769095207,
      "loss": 4.8935,
      "step": 9503
    },
    {
      "epoch": 1.3015612161051766,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007990487231826188,
      "loss": 4.8217,
      "step": 9504
    },
    {
      "epoch": 1.3016981648863326,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000799022866563067,
      "loss": 4.9284,
      "step": 9505
    },
    {
      "epoch": 1.3018351136674884,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0007989970070510795,
      "loss": 4.8696,
      "step": 9506
    },
    {
      "epoch": 1.3019720624486442,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007989711446468706,
      "loss": 4.8192,
      "step": 9507
    },
    {
      "epoch": 1.3021090112298,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007989452793506548,
      "loss": 4.8299,
      "step": 9508
    },
    {
      "epoch": 1.302245960010956,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007989194111626463,
      "loss": 4.8393,
      "step": 9509
    },
    {
      "epoch": 1.3023829087921117,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007988935400830594,
      "loss": 4.855,
      "step": 9510
    },
    {
      "epoch": 1.3025198575732677,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007988676661121086,
      "loss": 4.8804,
      "step": 9511
    },
    {
      "epoch": 1.3026568063544235,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007988417892500083,
      "loss": 4.905,
      "step": 9512
    },
    {
      "epoch": 1.3027937551355793,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007988159094969731,
      "loss": 4.8435,
      "step": 9513
    },
    {
      "epoch": 1.302930703916735,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007987900268532172,
      "loss": 4.8318,
      "step": 9514
    },
    {
      "epoch": 1.303067652697891,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007987641413189553,
      "loss": 4.8794,
      "step": 9515
    },
    {
      "epoch": 1.3032046014790468,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007987382528944019,
      "loss": 4.9352,
      "step": 9516
    },
    {
      "epoch": 1.3033415502602028,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007987123615797715,
      "loss": 4.8179,
      "step": 9517
    },
    {
      "epoch": 1.3034784990413586,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007986864673752786,
      "loss": 4.9837,
      "step": 9518
    },
    {
      "epoch": 1.3036154478225144,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007986605702811378,
      "loss": 4.9268,
      "step": 9519
    },
    {
      "epoch": 1.3037523966036701,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000798634670297564,
      "loss": 4.9387,
      "step": 9520
    },
    {
      "epoch": 1.3038893453848261,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007986087674247714,
      "loss": 4.8694,
      "step": 9521
    },
    {
      "epoch": 1.304026294165982,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000798582861662975,
      "loss": 4.8599,
      "step": 9522
    },
    {
      "epoch": 1.3041632429471377,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007985569530123892,
      "loss": 4.8487,
      "step": 9523
    },
    {
      "epoch": 1.3043001917282937,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007985310414732291,
      "loss": 4.9342,
      "step": 9524
    },
    {
      "epoch": 1.3044371405094495,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000798505127045709,
      "loss": 4.8591,
      "step": 9525
    },
    {
      "epoch": 1.3045740892906053,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007984792097300441,
      "loss": 4.8648,
      "step": 9526
    },
    {
      "epoch": 1.304711038071761,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007984532895264488,
      "loss": 4.8664,
      "step": 9527
    },
    {
      "epoch": 1.304847986852917,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007984273664351381,
      "loss": 4.8392,
      "step": 9528
    },
    {
      "epoch": 1.3049849356340728,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007984014404563267,
      "loss": 4.9306,
      "step": 9529
    },
    {
      "epoch": 1.3051218844152288,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007983755115902295,
      "loss": 4.8149,
      "step": 9530
    },
    {
      "epoch": 1.3052588331963846,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007983495798370616,
      "loss": 4.9272,
      "step": 9531
    },
    {
      "epoch": 1.3053957819775404,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007983236451970375,
      "loss": 4.794,
      "step": 9532
    },
    {
      "epoch": 1.3055327307586961,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007982977076703727,
      "loss": 4.8897,
      "step": 9533
    },
    {
      "epoch": 1.3056696795398521,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007982717672572815,
      "loss": 4.8608,
      "step": 9534
    },
    {
      "epoch": 1.305806628321008,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007982458239579793,
      "loss": 4.9516,
      "step": 9535
    },
    {
      "epoch": 1.305943577102164,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000798219877772681,
      "loss": 4.9235,
      "step": 9536
    },
    {
      "epoch": 1.3060805258833197,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007981939287016015,
      "loss": 4.8084,
      "step": 9537
    },
    {
      "epoch": 1.3062174746644755,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000798167976744956,
      "loss": 4.9007,
      "step": 9538
    },
    {
      "epoch": 1.3063544234456312,
      "grad_norm": 0.375,
      "learning_rate": 0.0007981420219029594,
      "loss": 4.8738,
      "step": 9539
    },
    {
      "epoch": 1.3064913722267872,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007981160641758271,
      "loss": 4.8896,
      "step": 9540
    },
    {
      "epoch": 1.306628321007943,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007980901035637739,
      "loss": 4.8879,
      "step": 9541
    },
    {
      "epoch": 1.306765269789099,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000798064140067015,
      "loss": 4.8556,
      "step": 9542
    },
    {
      "epoch": 1.3069022185702548,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007980381736857658,
      "loss": 4.9055,
      "step": 9543
    },
    {
      "epoch": 1.3070391673514106,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007980122044202412,
      "loss": 4.9371,
      "step": 9544
    },
    {
      "epoch": 1.3071761161325663,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007979862322706567,
      "loss": 4.8363,
      "step": 9545
    },
    {
      "epoch": 1.3073130649137223,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007979602572372272,
      "loss": 5.0359,
      "step": 9546
    },
    {
      "epoch": 1.3074500136948781,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007979342793201682,
      "loss": 4.9491,
      "step": 9547
    },
    {
      "epoch": 1.307586962476034,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000797908298519695,
      "loss": 4.8885,
      "step": 9548
    },
    {
      "epoch": 1.30772391125719,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007978823148360227,
      "loss": 4.8533,
      "step": 9549
    },
    {
      "epoch": 1.3078608600383457,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007978563282693668,
      "loss": 4.9288,
      "step": 9550
    },
    {
      "epoch": 1.3079978088195015,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007978303388199426,
      "loss": 4.8132,
      "step": 9551
    },
    {
      "epoch": 1.3081347576006572,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007978043464879654,
      "loss": 4.9514,
      "step": 9552
    },
    {
      "epoch": 1.3082717063818132,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000797778351273651,
      "loss": 4.8362,
      "step": 9553
    },
    {
      "epoch": 1.308408655162969,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007977523531772142,
      "loss": 4.8596,
      "step": 9554
    },
    {
      "epoch": 1.308545603944125,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007977263521988711,
      "loss": 4.9329,
      "step": 9555
    },
    {
      "epoch": 1.3086825527252808,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007977003483388367,
      "loss": 4.7359,
      "step": 9556
    },
    {
      "epoch": 1.3088195015064366,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007976743415973265,
      "loss": 4.9946,
      "step": 9557
    },
    {
      "epoch": 1.3089564502875923,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007976483319745563,
      "loss": 4.8477,
      "step": 9558
    },
    {
      "epoch": 1.3090933990687483,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007976223194707417,
      "loss": 4.9016,
      "step": 9559
    },
    {
      "epoch": 1.309230347849904,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000797596304086098,
      "loss": 4.9279,
      "step": 9560
    },
    {
      "epoch": 1.30936729663106,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007975702858208408,
      "loss": 4.904,
      "step": 9561
    },
    {
      "epoch": 1.3095042454122159,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000797544264675186,
      "loss": 4.8881,
      "step": 9562
    },
    {
      "epoch": 1.3096411941933717,
      "grad_norm": 0.40625,
      "learning_rate": 0.000797518240649349,
      "loss": 4.8716,
      "step": 9563
    },
    {
      "epoch": 1.3097781429745274,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007974922137435453,
      "loss": 4.8165,
      "step": 9564
    },
    {
      "epoch": 1.3099150917556834,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007974661839579911,
      "loss": 4.9706,
      "step": 9565
    },
    {
      "epoch": 1.3100520405368392,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007974401512929018,
      "loss": 4.8467,
      "step": 9566
    },
    {
      "epoch": 1.310188989317995,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007974141157484931,
      "loss": 4.7554,
      "step": 9567
    },
    {
      "epoch": 1.310325938099151,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007973880773249808,
      "loss": 4.844,
      "step": 9568
    },
    {
      "epoch": 1.3104628868803068,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007973620360225809,
      "loss": 4.906,
      "step": 9569
    },
    {
      "epoch": 1.3105998356614625,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007973359918415089,
      "loss": 4.9147,
      "step": 9570
    },
    {
      "epoch": 1.3107367844426185,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007973099447819808,
      "loss": 4.8955,
      "step": 9571
    },
    {
      "epoch": 1.3108737332237743,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007972838948442125,
      "loss": 4.8765,
      "step": 9572
    },
    {
      "epoch": 1.31101068200493,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007972578420284198,
      "loss": 4.9388,
      "step": 9573
    },
    {
      "epoch": 1.311147630786086,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007972317863348185,
      "loss": 4.7884,
      "step": 9574
    },
    {
      "epoch": 1.3112845795672419,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007972057277636246,
      "loss": 4.9005,
      "step": 9575
    },
    {
      "epoch": 1.3114215283483976,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007971796663150542,
      "loss": 4.8684,
      "step": 9576
    },
    {
      "epoch": 1.3115584771295534,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007971536019893233,
      "loss": 4.8565,
      "step": 9577
    },
    {
      "epoch": 1.3116954259107094,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007971275347866476,
      "loss": 4.8169,
      "step": 9578
    },
    {
      "epoch": 1.3118323746918652,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007971014647072433,
      "loss": 4.8001,
      "step": 9579
    },
    {
      "epoch": 1.3119693234730212,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007970753917513265,
      "loss": 4.883,
      "step": 9580
    },
    {
      "epoch": 1.312106272254177,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007970493159191132,
      "loss": 4.899,
      "step": 9581
    },
    {
      "epoch": 1.3122432210353328,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007970232372108196,
      "loss": 4.8961,
      "step": 9582
    },
    {
      "epoch": 1.3123801698164885,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007969971556266616,
      "loss": 4.8449,
      "step": 9583
    },
    {
      "epoch": 1.3125171185976445,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007969710711668557,
      "loss": 4.9071,
      "step": 9584
    },
    {
      "epoch": 1.3126540673788003,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007969449838316177,
      "loss": 4.8369,
      "step": 9585
    },
    {
      "epoch": 1.3127910161599563,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000796918893621164,
      "loss": 4.8037,
      "step": 9586
    },
    {
      "epoch": 1.312927964941112,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007968928005357107,
      "loss": 4.8182,
      "step": 9587
    },
    {
      "epoch": 1.3130649137222679,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007968667045754742,
      "loss": 4.8811,
      "step": 9588
    },
    {
      "epoch": 1.3132018625034236,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007968406057406707,
      "loss": 4.9198,
      "step": 9589
    },
    {
      "epoch": 1.3133388112845796,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007968145040315164,
      "loss": 4.8607,
      "step": 9590
    },
    {
      "epoch": 1.3134757600657354,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007967883994482275,
      "loss": 4.825,
      "step": 9591
    },
    {
      "epoch": 1.3136127088468912,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007967622919910207,
      "loss": 4.8636,
      "step": 9592
    },
    {
      "epoch": 1.3137496576280472,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007967361816601121,
      "loss": 4.887,
      "step": 9593
    },
    {
      "epoch": 1.313886606409203,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000796710068455718,
      "loss": 4.9403,
      "step": 9594
    },
    {
      "epoch": 1.3140235551903587,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007966839523780551,
      "loss": 4.9122,
      "step": 9595
    },
    {
      "epoch": 1.3141605039715145,
      "grad_norm": 0.515625,
      "learning_rate": 0.0007966578334273397,
      "loss": 4.9539,
      "step": 9596
    },
    {
      "epoch": 1.3142974527526705,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007966317116037882,
      "loss": 4.9271,
      "step": 9597
    },
    {
      "epoch": 1.3144344015338263,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000796605586907617,
      "loss": 4.8414,
      "step": 9598
    },
    {
      "epoch": 1.3145713503149823,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007965794593390429,
      "loss": 4.8323,
      "step": 9599
    },
    {
      "epoch": 1.314708299096138,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007965533288982822,
      "loss": 4.9225,
      "step": 9600
    },
    {
      "epoch": 1.3148452478772938,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007965271955855513,
      "loss": 4.792,
      "step": 9601
    },
    {
      "epoch": 1.3149821966584496,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007965010594010671,
      "loss": 4.9097,
      "step": 9602
    },
    {
      "epoch": 1.3151191454396056,
      "grad_norm": 0.478515625,
      "learning_rate": 0.000796474920345046,
      "loss": 4.8737,
      "step": 9603
    },
    {
      "epoch": 1.3152560942207614,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007964487784177047,
      "loss": 4.8536,
      "step": 9604
    },
    {
      "epoch": 1.3153930430019174,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007964226336192598,
      "loss": 4.8803,
      "step": 9605
    },
    {
      "epoch": 1.3155299917830732,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000796396485949928,
      "loss": 4.8618,
      "step": 9606
    },
    {
      "epoch": 1.315666940564229,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000796370335409926,
      "loss": 4.9259,
      "step": 9607
    },
    {
      "epoch": 1.3158038893453847,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007963441819994704,
      "loss": 4.8528,
      "step": 9608
    },
    {
      "epoch": 1.3159408381265407,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007963180257187783,
      "loss": 4.9022,
      "step": 9609
    },
    {
      "epoch": 1.3160777869076965,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000796291866568066,
      "loss": 4.8517,
      "step": 9610
    },
    {
      "epoch": 1.3162147356888525,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007962657045475504,
      "loss": 4.8489,
      "step": 9611
    },
    {
      "epoch": 1.3163516844700083,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007962395396574485,
      "loss": 4.8296,
      "step": 9612
    },
    {
      "epoch": 1.316488633251164,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007962133718979771,
      "loss": 4.9124,
      "step": 9613
    },
    {
      "epoch": 1.3166255820323198,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007961872012693529,
      "loss": 4.928,
      "step": 9614
    },
    {
      "epoch": 1.3167625308134758,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007961610277717929,
      "loss": 4.8644,
      "step": 9615
    },
    {
      "epoch": 1.3168994795946316,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007961348514055139,
      "loss": 4.8496,
      "step": 9616
    },
    {
      "epoch": 1.3170364283757874,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000796108672170733,
      "loss": 4.9315,
      "step": 9617
    },
    {
      "epoch": 1.3171733771569434,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000796082490067667,
      "loss": 4.9132,
      "step": 9618
    },
    {
      "epoch": 1.3173103259380992,
      "grad_norm": 0.375,
      "learning_rate": 0.000796056305096533,
      "loss": 4.8396,
      "step": 9619
    },
    {
      "epoch": 1.317447274719255,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007960301172575478,
      "loss": 4.878,
      "step": 9620
    },
    {
      "epoch": 1.3175842235004107,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007960039265509287,
      "loss": 4.9183,
      "step": 9621
    },
    {
      "epoch": 1.3177211722815667,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007959777329768926,
      "loss": 4.8512,
      "step": 9622
    },
    {
      "epoch": 1.3178581210627225,
      "grad_norm": 0.375,
      "learning_rate": 0.0007959515365356567,
      "loss": 4.893,
      "step": 9623
    },
    {
      "epoch": 1.3179950698438785,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007959253372274378,
      "loss": 4.9198,
      "step": 9624
    },
    {
      "epoch": 1.3181320186250343,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007958991350524532,
      "loss": 4.89,
      "step": 9625
    },
    {
      "epoch": 1.31826896740619,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007958729300109201,
      "loss": 4.8887,
      "step": 9626
    },
    {
      "epoch": 1.3184059161873458,
      "grad_norm": 0.375,
      "learning_rate": 0.0007958467221030556,
      "loss": 4.8625,
      "step": 9627
    },
    {
      "epoch": 1.3185428649685018,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007958205113290769,
      "loss": 4.972,
      "step": 9628
    },
    {
      "epoch": 1.3186798137496576,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007957942976892013,
      "loss": 4.9617,
      "step": 9629
    },
    {
      "epoch": 1.3188167625308136,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007957680811836458,
      "loss": 4.8361,
      "step": 9630
    },
    {
      "epoch": 1.3189537113119694,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000795741861812628,
      "loss": 4.8569,
      "step": 9631
    },
    {
      "epoch": 1.3190906600931251,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007957156395763649,
      "loss": 4.8629,
      "step": 9632
    },
    {
      "epoch": 1.319227608874281,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000795689414475074,
      "loss": 4.8214,
      "step": 9633
    },
    {
      "epoch": 1.319364557655437,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007956631865089724,
      "loss": 4.9105,
      "step": 9634
    },
    {
      "epoch": 1.3195015064365927,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007956369556782777,
      "loss": 4.9547,
      "step": 9635
    },
    {
      "epoch": 1.3196384552177487,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007956107219832071,
      "loss": 4.9184,
      "step": 9636
    },
    {
      "epoch": 1.3197754039989045,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007955844854239781,
      "loss": 4.8752,
      "step": 9637
    },
    {
      "epoch": 1.3199123527800602,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007955582460008082,
      "loss": 4.9261,
      "step": 9638
    },
    {
      "epoch": 1.320049301561216,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007955320037139149,
      "loss": 4.8623,
      "step": 9639
    },
    {
      "epoch": 1.320186250342372,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007955057585635152,
      "loss": 4.8424,
      "step": 9640
    },
    {
      "epoch": 1.3203231991235278,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007954795105498273,
      "loss": 4.9158,
      "step": 9641
    },
    {
      "epoch": 1.3204601479046836,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007954532596730682,
      "loss": 4.9404,
      "step": 9642
    },
    {
      "epoch": 1.3205970966858396,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007954270059334557,
      "loss": 4.8664,
      "step": 9643
    },
    {
      "epoch": 1.3207340454669954,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007954007493312071,
      "loss": 4.922,
      "step": 9644
    },
    {
      "epoch": 1.3208709942481511,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007953744898665403,
      "loss": 4.8682,
      "step": 9645
    },
    {
      "epoch": 1.321007943029307,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000795348227539673,
      "loss": 4.8219,
      "step": 9646
    },
    {
      "epoch": 1.321144891810463,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007953219623508224,
      "loss": 4.9458,
      "step": 9647
    },
    {
      "epoch": 1.3212818405916187,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007952956943002064,
      "loss": 4.83,
      "step": 9648
    },
    {
      "epoch": 1.3214187893727747,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007952694233880429,
      "loss": 4.8655,
      "step": 9649
    },
    {
      "epoch": 1.3215557381539305,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0007952431496145493,
      "loss": 4.8441,
      "step": 9650
    },
    {
      "epoch": 1.3216926869350862,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007952168729799434,
      "loss": 4.8227,
      "step": 9651
    },
    {
      "epoch": 1.321829635716242,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007951905934844431,
      "loss": 4.8648,
      "step": 9652
    },
    {
      "epoch": 1.321966584497398,
      "grad_norm": 0.375,
      "learning_rate": 0.000795164311128266,
      "loss": 4.8579,
      "step": 9653
    },
    {
      "epoch": 1.3221035332785538,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007951380259116301,
      "loss": 4.7788,
      "step": 9654
    },
    {
      "epoch": 1.3222404820597098,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007951117378347532,
      "loss": 4.8724,
      "step": 9655
    },
    {
      "epoch": 1.3223774308408656,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007950854468978529,
      "loss": 4.8438,
      "step": 9656
    },
    {
      "epoch": 1.3225143796220213,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007950591531011474,
      "loss": 4.9095,
      "step": 9657
    },
    {
      "epoch": 1.3226513284031771,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007950328564448544,
      "loss": 4.8871,
      "step": 9658
    },
    {
      "epoch": 1.3227882771843331,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007950065569291919,
      "loss": 4.8565,
      "step": 9659
    },
    {
      "epoch": 1.322925225965489,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000794980254554378,
      "loss": 4.864,
      "step": 9660
    },
    {
      "epoch": 1.323062174746645,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007949539493206304,
      "loss": 4.9253,
      "step": 9661
    },
    {
      "epoch": 1.3231991235278007,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007949276412281672,
      "loss": 4.9394,
      "step": 9662
    },
    {
      "epoch": 1.3233360723089564,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007949013302772066,
      "loss": 4.8324,
      "step": 9663
    },
    {
      "epoch": 1.3234730210901122,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007948750164679663,
      "loss": 4.8772,
      "step": 9664
    },
    {
      "epoch": 1.3236099698712682,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007948486998006648,
      "loss": 4.839,
      "step": 9665
    },
    {
      "epoch": 1.323746918652424,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007948223802755197,
      "loss": 4.856,
      "step": 9666
    },
    {
      "epoch": 1.3238838674335798,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007947960578927495,
      "loss": 4.8698,
      "step": 9667
    },
    {
      "epoch": 1.3240208162147358,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007947697326525721,
      "loss": 4.8539,
      "step": 9668
    },
    {
      "epoch": 1.3241577649958916,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000794743404555206,
      "loss": 4.8795,
      "step": 9669
    },
    {
      "epoch": 1.3242947137770473,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007947170736008689,
      "loss": 4.9131,
      "step": 9670
    },
    {
      "epoch": 1.324431662558203,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007946907397897794,
      "loss": 4.9576,
      "step": 9671
    },
    {
      "epoch": 1.324568611339359,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007946644031221556,
      "loss": 4.7966,
      "step": 9672
    },
    {
      "epoch": 1.3247055601205149,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007946380635982156,
      "loss": 4.8649,
      "step": 9673
    },
    {
      "epoch": 1.3248425089016709,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000794611721218178,
      "loss": 4.9673,
      "step": 9674
    },
    {
      "epoch": 1.3249794576828267,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007945853759822611,
      "loss": 4.8282,
      "step": 9675
    },
    {
      "epoch": 1.3251164064639824,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007945590278906828,
      "loss": 4.828,
      "step": 9676
    },
    {
      "epoch": 1.3252533552451382,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0007945326769436619,
      "loss": 4.8538,
      "step": 9677
    },
    {
      "epoch": 1.3253903040262942,
      "grad_norm": 0.375,
      "learning_rate": 0.0007945063231414166,
      "loss": 4.8961,
      "step": 9678
    },
    {
      "epoch": 1.32552725280745,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0007944799664841652,
      "loss": 4.8627,
      "step": 9679
    },
    {
      "epoch": 1.325664201588606,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007944536069721263,
      "loss": 4.8779,
      "step": 9680
    },
    {
      "epoch": 1.3258011503697618,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007944272446055182,
      "loss": 4.8969,
      "step": 9681
    },
    {
      "epoch": 1.3259380991509175,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007944008793845595,
      "loss": 4.8367,
      "step": 9682
    },
    {
      "epoch": 1.3260750479320733,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007943745113094688,
      "loss": 4.8553,
      "step": 9683
    },
    {
      "epoch": 1.3262119967132293,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0007943481403804644,
      "loss": 4.8394,
      "step": 9684
    },
    {
      "epoch": 1.326348945494385,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007943217665977647,
      "loss": 4.8305,
      "step": 9685
    },
    {
      "epoch": 1.3264858942755409,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007942953899615886,
      "loss": 4.8588,
      "step": 9686
    },
    {
      "epoch": 1.3266228430566969,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007942690104721547,
      "loss": 4.8869,
      "step": 9687
    },
    {
      "epoch": 1.3267597918378526,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007942426281296812,
      "loss": 4.9419,
      "step": 9688
    },
    {
      "epoch": 1.3268967406190084,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007942162429343872,
      "loss": 4.8853,
      "step": 9689
    },
    {
      "epoch": 1.3270336894001644,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007941898548864912,
      "loss": 4.8413,
      "step": 9690
    },
    {
      "epoch": 1.3271706381813202,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007941634639862118,
      "loss": 4.8839,
      "step": 9691
    },
    {
      "epoch": 1.327307586962476,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007941370702337678,
      "loss": 4.8653,
      "step": 9692
    },
    {
      "epoch": 1.327444535743632,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007941106736293778,
      "loss": 4.8697,
      "step": 9693
    },
    {
      "epoch": 1.3275814845247877,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0007940842741732607,
      "loss": 4.8902,
      "step": 9694
    },
    {
      "epoch": 1.3277184333059435,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007940578718656353,
      "loss": 4.9023,
      "step": 9695
    },
    {
      "epoch": 1.3278553820870993,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007940314667067203,
      "loss": 4.7186,
      "step": 9696
    },
    {
      "epoch": 1.3279923308682553,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007940050586967345,
      "loss": 4.9077,
      "step": 9697
    },
    {
      "epoch": 1.328129279649411,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007939786478358968,
      "loss": 4.8485,
      "step": 9698
    },
    {
      "epoch": 1.328266228430567,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007939522341244261,
      "loss": 4.8897,
      "step": 9699
    },
    {
      "epoch": 1.3284031772117229,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007939258175625413,
      "loss": 4.8824,
      "step": 9700
    },
    {
      "epoch": 1.3285401259928786,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007938993981504613,
      "loss": 4.9031,
      "step": 9701
    },
    {
      "epoch": 1.3286770747740344,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007938729758884051,
      "loss": 4.8402,
      "step": 9702
    },
    {
      "epoch": 1.3288140235551904,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007938465507765915,
      "loss": 4.935,
      "step": 9703
    },
    {
      "epoch": 1.3289509723363462,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007938201228152396,
      "loss": 4.9923,
      "step": 9704
    },
    {
      "epoch": 1.3290879211175022,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007937936920045683,
      "loss": 4.8783,
      "step": 9705
    },
    {
      "epoch": 1.329224869898658,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007937672583447971,
      "loss": 4.8567,
      "step": 9706
    },
    {
      "epoch": 1.3293618186798137,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007937408218361444,
      "loss": 4.7873,
      "step": 9707
    },
    {
      "epoch": 1.3294987674609695,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007937143824788297,
      "loss": 4.8656,
      "step": 9708
    },
    {
      "epoch": 1.3296357162421255,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007936879402730721,
      "loss": 4.8227,
      "step": 9709
    },
    {
      "epoch": 1.3297726650232813,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0007936614952190905,
      "loss": 4.9424,
      "step": 9710
    },
    {
      "epoch": 1.329909613804437,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0007936350473171043,
      "loss": 4.817,
      "step": 9711
    },
    {
      "epoch": 1.330046562585593,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007936085965673325,
      "loss": 4.9058,
      "step": 9712
    },
    {
      "epoch": 1.3301835113667488,
      "grad_norm": 0.5,
      "learning_rate": 0.0007935821429699945,
      "loss": 4.9113,
      "step": 9713
    },
    {
      "epoch": 1.3303204601479046,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007935556865253093,
      "loss": 4.8777,
      "step": 9714
    },
    {
      "epoch": 1.3304574089290606,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007935292272334963,
      "loss": 4.8923,
      "step": 9715
    },
    {
      "epoch": 1.3305943577102164,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007935027650947747,
      "loss": 4.901,
      "step": 9716
    },
    {
      "epoch": 1.3307313064913722,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007934763001093639,
      "loss": 4.9154,
      "step": 9717
    },
    {
      "epoch": 1.3308682552725282,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007934498322774832,
      "loss": 4.884,
      "step": 9718
    },
    {
      "epoch": 1.331005204053684,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007934233615993517,
      "loss": 4.9012,
      "step": 9719
    },
    {
      "epoch": 1.3311421528348397,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007933968880751893,
      "loss": 4.8526,
      "step": 9720
    },
    {
      "epoch": 1.3312791016159955,
      "grad_norm": 0.375,
      "learning_rate": 0.0007933704117052148,
      "loss": 4.8375,
      "step": 9721
    },
    {
      "epoch": 1.3314160503971515,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007933439324896478,
      "loss": 4.8376,
      "step": 9722
    },
    {
      "epoch": 1.3315529991783073,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007933174504287081,
      "loss": 4.8934,
      "step": 9723
    },
    {
      "epoch": 1.3316899479594633,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007932909655226146,
      "loss": 4.8912,
      "step": 9724
    },
    {
      "epoch": 1.331826896740619,
      "grad_norm": 0.375,
      "learning_rate": 0.0007932644777715873,
      "loss": 4.8511,
      "step": 9725
    },
    {
      "epoch": 1.3319638455217748,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007932379871758452,
      "loss": 4.8443,
      "step": 9726
    },
    {
      "epoch": 1.3321007943029306,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007932114937356084,
      "loss": 4.8854,
      "step": 9727
    },
    {
      "epoch": 1.3322377430840866,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007931849974510961,
      "loss": 4.8201,
      "step": 9728
    },
    {
      "epoch": 1.3323746918652424,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000793158498322528,
      "loss": 4.9128,
      "step": 9729
    },
    {
      "epoch": 1.3325116406463984,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007931319963501234,
      "loss": 4.9571,
      "step": 9730
    },
    {
      "epoch": 1.3326485894275542,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007931054915341024,
      "loss": 4.9257,
      "step": 9731
    },
    {
      "epoch": 1.33278553820871,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007930789838746843,
      "loss": 4.815,
      "step": 9732
    },
    {
      "epoch": 1.3329224869898657,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000793052473372089,
      "loss": 4.8763,
      "step": 9733
    },
    {
      "epoch": 1.3330594357710217,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007930259600265361,
      "loss": 4.854,
      "step": 9734
    },
    {
      "epoch": 1.3331963845521775,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007929994438382453,
      "loss": 4.9606,
      "step": 9735
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007929729248074364,
      "loss": 4.7963,
      "step": 9736
    },
    {
      "epoch": 1.3334702821144893,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007929464029343292,
      "loss": 4.8948,
      "step": 9737
    },
    {
      "epoch": 1.333607230895645,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007929198782191434,
      "loss": 4.919,
      "step": 9738
    },
    {
      "epoch": 1.3337441796768008,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007928933506620987,
      "loss": 4.8836,
      "step": 9739
    },
    {
      "epoch": 1.3338811284579566,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007928668202634154,
      "loss": 4.9162,
      "step": 9740
    },
    {
      "epoch": 1.3340180772391126,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007928402870233128,
      "loss": 4.7165,
      "step": 9741
    },
    {
      "epoch": 1.3341550260202684,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007928137509420112,
      "loss": 4.815,
      "step": 9742
    },
    {
      "epoch": 1.3342919748014244,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007927872120197302,
      "loss": 4.8499,
      "step": 9743
    },
    {
      "epoch": 1.3344289235825801,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00079276067025669,
      "loss": 4.928,
      "step": 9744
    },
    {
      "epoch": 1.334565872363736,
      "grad_norm": 0.375,
      "learning_rate": 0.0007927341256531104,
      "loss": 4.894,
      "step": 9745
    },
    {
      "epoch": 1.3347028211448917,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007927075782092115,
      "loss": 4.8518,
      "step": 9746
    },
    {
      "epoch": 1.3348397699260477,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007926810279252133,
      "loss": 4.9046,
      "step": 9747
    },
    {
      "epoch": 1.3349767187072035,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007926544748013356,
      "loss": 4.8798,
      "step": 9748
    },
    {
      "epoch": 1.3351136674883595,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007926279188377987,
      "loss": 4.8901,
      "step": 9749
    },
    {
      "epoch": 1.3352506162695152,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007926013600348226,
      "loss": 4.8702,
      "step": 9750
    },
    {
      "epoch": 1.335387565050671,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007925747983926275,
      "loss": 4.8453,
      "step": 9751
    },
    {
      "epoch": 1.3355245138318268,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007925482339114332,
      "loss": 4.8771,
      "step": 9752
    },
    {
      "epoch": 1.3356614626129828,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007925216665914601,
      "loss": 4.8511,
      "step": 9753
    },
    {
      "epoch": 1.3357984113941386,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007924950964329284,
      "loss": 4.8512,
      "step": 9754
    },
    {
      "epoch": 1.3359353601752946,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007924685234360582,
      "loss": 4.9146,
      "step": 9755
    },
    {
      "epoch": 1.3360723089564503,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007924419476010698,
      "loss": 4.8688,
      "step": 9756
    },
    {
      "epoch": 1.3362092577376061,
      "grad_norm": 0.375,
      "learning_rate": 0.0007924153689281833,
      "loss": 4.8533,
      "step": 9757
    },
    {
      "epoch": 1.336346206518762,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000792388787417619,
      "loss": 4.8509,
      "step": 9758
    },
    {
      "epoch": 1.336483155299918,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007923622030695974,
      "loss": 4.9135,
      "step": 9759
    },
    {
      "epoch": 1.3366201040810737,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007923356158843386,
      "loss": 5.011,
      "step": 9760
    },
    {
      "epoch": 1.3367570528622295,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007923090258620628,
      "loss": 4.8912,
      "step": 9761
    },
    {
      "epoch": 1.3368940016433855,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007922824330029907,
      "loss": 4.899,
      "step": 9762
    },
    {
      "epoch": 1.3370309504245412,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007922558373073424,
      "loss": 4.8914,
      "step": 9763
    },
    {
      "epoch": 1.337167899205697,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007922292387753385,
      "loss": 4.8343,
      "step": 9764
    },
    {
      "epoch": 1.3373048479868528,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007922026374071994,
      "loss": 4.8696,
      "step": 9765
    },
    {
      "epoch": 1.3374417967680088,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007921760332031454,
      "loss": 4.8723,
      "step": 9766
    },
    {
      "epoch": 1.3375787455491646,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007921494261633972,
      "loss": 4.8421,
      "step": 9767
    },
    {
      "epoch": 1.3377156943303206,
      "grad_norm": 0.375,
      "learning_rate": 0.0007921228162881751,
      "loss": 4.7405,
      "step": 9768
    },
    {
      "epoch": 1.3378526431114763,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007920962035776997,
      "loss": 4.8088,
      "step": 9769
    },
    {
      "epoch": 1.3379895918926321,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007920695880321915,
      "loss": 4.8291,
      "step": 9770
    },
    {
      "epoch": 1.338126540673788,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007920429696518711,
      "loss": 4.9582,
      "step": 9771
    },
    {
      "epoch": 1.3382634894549439,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007920163484369593,
      "loss": 4.8898,
      "step": 9772
    },
    {
      "epoch": 1.3384004382360997,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007919897243876762,
      "loss": 4.9095,
      "step": 9773
    },
    {
      "epoch": 1.3385373870172557,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007919630975042431,
      "loss": 4.8007,
      "step": 9774
    },
    {
      "epoch": 1.3386743357984114,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007919364677868801,
      "loss": 4.8307,
      "step": 9775
    },
    {
      "epoch": 1.3388112845795672,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007919098352358082,
      "loss": 4.8827,
      "step": 9776
    },
    {
      "epoch": 1.338948233360723,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000791883199851248,
      "loss": 4.8438,
      "step": 9777
    },
    {
      "epoch": 1.339085182141879,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007918565616334203,
      "loss": 4.8499,
      "step": 9778
    },
    {
      "epoch": 1.3392221309230348,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007918299205825457,
      "loss": 4.9453,
      "step": 9779
    },
    {
      "epoch": 1.3393590797041908,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007918032766988453,
      "loss": 4.8124,
      "step": 9780
    },
    {
      "epoch": 1.3394960284853465,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007917766299825396,
      "loss": 4.8944,
      "step": 9781
    },
    {
      "epoch": 1.3396329772665023,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007917499804338494,
      "loss": 4.77,
      "step": 9782
    },
    {
      "epoch": 1.339769926047658,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007917233280529958,
      "loss": 4.8349,
      "step": 9783
    },
    {
      "epoch": 1.339906874828814,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007916966728401994,
      "loss": 4.8134,
      "step": 9784
    },
    {
      "epoch": 1.3400438236099699,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007916700147956815,
      "loss": 4.9215,
      "step": 9785
    },
    {
      "epoch": 1.3401807723911257,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007916433539196626,
      "loss": 4.9125,
      "step": 9786
    },
    {
      "epoch": 1.3403177211722817,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007916166902123639,
      "loss": 4.9093,
      "step": 9787
    },
    {
      "epoch": 1.3404546699534374,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007915900236740062,
      "loss": 4.9294,
      "step": 9788
    },
    {
      "epoch": 1.3405916187345932,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007915633543048108,
      "loss": 4.7659,
      "step": 9789
    },
    {
      "epoch": 1.340728567515749,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007915366821049982,
      "loss": 4.8717,
      "step": 9790
    },
    {
      "epoch": 1.340865516296905,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007915100070747898,
      "loss": 4.9671,
      "step": 9791
    },
    {
      "epoch": 1.3410024650780608,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007914833292144067,
      "loss": 4.9069,
      "step": 9792
    },
    {
      "epoch": 1.3411394138592168,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007914566485240698,
      "loss": 4.7365,
      "step": 9793
    },
    {
      "epoch": 1.3412763626403725,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007914299650040004,
      "loss": 4.9063,
      "step": 9794
    },
    {
      "epoch": 1.3414133114215283,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007914032786544193,
      "loss": 4.936,
      "step": 9795
    },
    {
      "epoch": 1.341550260202684,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007913765894755481,
      "loss": 4.8711,
      "step": 9796
    },
    {
      "epoch": 1.34168720898384,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007913498974676076,
      "loss": 4.8721,
      "step": 9797
    },
    {
      "epoch": 1.3418241577649959,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007913232026308193,
      "loss": 4.8502,
      "step": 9798
    },
    {
      "epoch": 1.3419611065461519,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007912965049654042,
      "loss": 4.8609,
      "step": 9799
    },
    {
      "epoch": 1.3420980553273076,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007912698044715837,
      "loss": 4.8112,
      "step": 9800
    },
    {
      "epoch": 1.3422350041084634,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007912431011495789,
      "loss": 4.915,
      "step": 9801
    },
    {
      "epoch": 1.3423719528896192,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007912163949996113,
      "loss": 4.8825,
      "step": 9802
    },
    {
      "epoch": 1.3425089016707752,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000791189686021902,
      "loss": 4.9357,
      "step": 9803
    },
    {
      "epoch": 1.342645850451931,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007911629742166727,
      "loss": 4.9228,
      "step": 9804
    },
    {
      "epoch": 1.3427827992330867,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007911362595841444,
      "loss": 4.7928,
      "step": 9805
    },
    {
      "epoch": 1.3429197480142427,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007911095421245387,
      "loss": 4.9362,
      "step": 9806
    },
    {
      "epoch": 1.3430566967953985,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007910828218380769,
      "loss": 4.8546,
      "step": 9807
    },
    {
      "epoch": 1.3431936455765543,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007910560987249803,
      "loss": 4.9658,
      "step": 9808
    },
    {
      "epoch": 1.3433305943577103,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007910293727854708,
      "loss": 4.9599,
      "step": 9809
    },
    {
      "epoch": 1.343467543138866,
      "grad_norm": 0.375,
      "learning_rate": 0.0007910026440197694,
      "loss": 4.8653,
      "step": 9810
    },
    {
      "epoch": 1.3436044919200218,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000790975912428098,
      "loss": 4.9496,
      "step": 9811
    },
    {
      "epoch": 1.3437414407011778,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000790949178010678,
      "loss": 4.8821,
      "step": 9812
    },
    {
      "epoch": 1.3438783894823336,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007909224407677308,
      "loss": 4.84,
      "step": 9813
    },
    {
      "epoch": 1.3440153382634894,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007908957006994782,
      "loss": 4.9112,
      "step": 9814
    },
    {
      "epoch": 1.3441522870446452,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007908689578061417,
      "loss": 4.8775,
      "step": 9815
    },
    {
      "epoch": 1.3442892358258012,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007908422120879427,
      "loss": 4.8429,
      "step": 9816
    },
    {
      "epoch": 1.344426184606957,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007908154635451032,
      "loss": 4.8727,
      "step": 9817
    },
    {
      "epoch": 1.344563133388113,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007907887121778449,
      "loss": 4.7837,
      "step": 9818
    },
    {
      "epoch": 1.3447000821692687,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007907619579863892,
      "loss": 4.9059,
      "step": 9819
    },
    {
      "epoch": 1.3448370309504245,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007907352009709578,
      "loss": 4.8248,
      "step": 9820
    },
    {
      "epoch": 1.3449739797315803,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007907084411317729,
      "loss": 4.85,
      "step": 9821
    },
    {
      "epoch": 1.3451109285127363,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007906816784690558,
      "loss": 4.7323,
      "step": 9822
    },
    {
      "epoch": 1.345247877293892,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007906549129830285,
      "loss": 4.9058,
      "step": 9823
    },
    {
      "epoch": 1.345384826075048,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007906281446739127,
      "loss": 4.8639,
      "step": 9824
    },
    {
      "epoch": 1.3455217748562038,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007906013735419303,
      "loss": 4.8315,
      "step": 9825
    },
    {
      "epoch": 1.3456587236373596,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007905745995873031,
      "loss": 4.8278,
      "step": 9826
    },
    {
      "epoch": 1.3457956724185154,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007905478228102532,
      "loss": 4.7514,
      "step": 9827
    },
    {
      "epoch": 1.3459326211996714,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007905210432110022,
      "loss": 4.8499,
      "step": 9828
    },
    {
      "epoch": 1.3460695699808272,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0007904942607897722,
      "loss": 4.8735,
      "step": 9829
    },
    {
      "epoch": 1.346206518761983,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007904674755467853,
      "loss": 4.8992,
      "step": 9830
    },
    {
      "epoch": 1.346343467543139,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007904406874822632,
      "loss": 4.8895,
      "step": 9831
    },
    {
      "epoch": 1.3464804163242947,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000790413896596428,
      "loss": 4.9429,
      "step": 9832
    },
    {
      "epoch": 1.3466173651054505,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007903871028895016,
      "loss": 4.831,
      "step": 9833
    },
    {
      "epoch": 1.3467543138866065,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0007903603063617063,
      "loss": 4.8877,
      "step": 9834
    },
    {
      "epoch": 1.3468912626677623,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007903335070132639,
      "loss": 4.9222,
      "step": 9835
    },
    {
      "epoch": 1.347028211448918,
      "grad_norm": 0.546875,
      "learning_rate": 0.0007903067048443968,
      "loss": 4.948,
      "step": 9836
    },
    {
      "epoch": 1.347165160230074,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0007902798998553268,
      "loss": 4.8578,
      "step": 9837
    },
    {
      "epoch": 1.3473021090112298,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007902530920462763,
      "loss": 4.8663,
      "step": 9838
    },
    {
      "epoch": 1.3474390577923856,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007902262814174673,
      "loss": 4.8271,
      "step": 9839
    },
    {
      "epoch": 1.3475760065735414,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007901994679691221,
      "loss": 4.9163,
      "step": 9840
    },
    {
      "epoch": 1.3477129553546974,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0007901726517014627,
      "loss": 4.8398,
      "step": 9841
    },
    {
      "epoch": 1.3478499041358531,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007901458326147116,
      "loss": 4.9004,
      "step": 9842
    },
    {
      "epoch": 1.3479868529170091,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000790119010709091,
      "loss": 4.8204,
      "step": 9843
    },
    {
      "epoch": 1.348123801698165,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0007900921859848229,
      "loss": 4.8843,
      "step": 9844
    },
    {
      "epoch": 1.3482607504793207,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00079006535844213,
      "loss": 4.9197,
      "step": 9845
    },
    {
      "epoch": 1.3483976992604765,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0007900385280812344,
      "loss": 4.8666,
      "step": 9846
    },
    {
      "epoch": 1.3485346480416325,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007900116949023583,
      "loss": 4.8179,
      "step": 9847
    },
    {
      "epoch": 1.3486715968227883,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007899848589057244,
      "loss": 4.8952,
      "step": 9848
    },
    {
      "epoch": 1.3488085456039443,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007899580200915552,
      "loss": 4.8871,
      "step": 9849
    },
    {
      "epoch": 1.3489454943851,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007899311784600725,
      "loss": 4.8772,
      "step": 9850
    },
    {
      "epoch": 1.3490824431662558,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0007899043340114992,
      "loss": 4.7679,
      "step": 9851
    },
    {
      "epoch": 1.3492193919474116,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007898774867460579,
      "loss": 5.0051,
      "step": 9852
    },
    {
      "epoch": 1.3493563407285676,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007898506366639707,
      "loss": 4.777,
      "step": 9853
    },
    {
      "epoch": 1.3494932895097234,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007898237837654603,
      "loss": 4.8661,
      "step": 9854
    },
    {
      "epoch": 1.3496302382908791,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007897969280507493,
      "loss": 4.8374,
      "step": 9855
    },
    {
      "epoch": 1.3497671870720351,
      "grad_norm": 0.375,
      "learning_rate": 0.0007897700695200601,
      "loss": 4.8924,
      "step": 9856
    },
    {
      "epoch": 1.349904135853191,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007897432081736155,
      "loss": 4.8395,
      "step": 9857
    },
    {
      "epoch": 1.3500410846343467,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007897163440116378,
      "loss": 4.8799,
      "step": 9858
    },
    {
      "epoch": 1.3501780334155025,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007896894770343501,
      "loss": 4.8338,
      "step": 9859
    },
    {
      "epoch": 1.3503149821966585,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007896626072419745,
      "loss": 4.8386,
      "step": 9860
    },
    {
      "epoch": 1.3504519309778142,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000789635734634734,
      "loss": 4.93,
      "step": 9861
    },
    {
      "epoch": 1.3505888797589702,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007896088592128513,
      "loss": 4.8783,
      "step": 9862
    },
    {
      "epoch": 1.350725828540126,
      "grad_norm": 0.375,
      "learning_rate": 0.000789581980976549,
      "loss": 4.8533,
      "step": 9863
    },
    {
      "epoch": 1.3508627773212818,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007895550999260499,
      "loss": 4.7975,
      "step": 9864
    },
    {
      "epoch": 1.3509997261024376,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007895282160615767,
      "loss": 4.8137,
      "step": 9865
    },
    {
      "epoch": 1.3511366748835936,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007895013293833524,
      "loss": 4.8982,
      "step": 9866
    },
    {
      "epoch": 1.3512736236647493,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007894744398915997,
      "loss": 4.8744,
      "step": 9867
    },
    {
      "epoch": 1.3514105724459053,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007894475475865413,
      "loss": 4.9189,
      "step": 9868
    },
    {
      "epoch": 1.3515475212270611,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007894206524684003,
      "loss": 4.9191,
      "step": 9869
    },
    {
      "epoch": 1.351684470008217,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007893937545373994,
      "loss": 4.866,
      "step": 9870
    },
    {
      "epoch": 1.3518214187893727,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007893668537937616,
      "loss": 4.868,
      "step": 9871
    },
    {
      "epoch": 1.3519583675705287,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007893399502377097,
      "loss": 4.787,
      "step": 9872
    },
    {
      "epoch": 1.3520953163516845,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007893130438694669,
      "loss": 4.9051,
      "step": 9873
    },
    {
      "epoch": 1.3522322651328404,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007892861346892561,
      "loss": 4.8293,
      "step": 9874
    },
    {
      "epoch": 1.3523692139139962,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007892592226973003,
      "loss": 4.8413,
      "step": 9875
    },
    {
      "epoch": 1.352506162695152,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007892323078938224,
      "loss": 4.7855,
      "step": 9876
    },
    {
      "epoch": 1.3526431114763078,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007892053902790455,
      "loss": 4.8685,
      "step": 9877
    },
    {
      "epoch": 1.3527800602574638,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007891784698531928,
      "loss": 4.8141,
      "step": 9878
    },
    {
      "epoch": 1.3529170090386196,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007891515466164871,
      "loss": 4.8062,
      "step": 9879
    },
    {
      "epoch": 1.3530539578197753,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007891246205691519,
      "loss": 4.8785,
      "step": 9880
    },
    {
      "epoch": 1.3531909066009313,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00078909769171141,
      "loss": 4.8559,
      "step": 9881
    },
    {
      "epoch": 1.353327855382087,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007890707600434849,
      "loss": 4.7307,
      "step": 9882
    },
    {
      "epoch": 1.3534648041632429,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007890438255655994,
      "loss": 4.9653,
      "step": 9883
    },
    {
      "epoch": 1.3536017529443987,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007890168882779771,
      "loss": 4.8786,
      "step": 9884
    },
    {
      "epoch": 1.3537387017255547,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000788989948180841,
      "loss": 4.8796,
      "step": 9885
    },
    {
      "epoch": 1.3538756505067104,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007889630052744144,
      "loss": 4.9062,
      "step": 9886
    },
    {
      "epoch": 1.3540125992878664,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007889360595589208,
      "loss": 4.7622,
      "step": 9887
    },
    {
      "epoch": 1.3541495480690222,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007889091110345832,
      "loss": 4.9018,
      "step": 9888
    },
    {
      "epoch": 1.354286496850178,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007888821597016249,
      "loss": 4.806,
      "step": 9889
    },
    {
      "epoch": 1.3544234456313338,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007888552055602694,
      "loss": 4.8976,
      "step": 9890
    },
    {
      "epoch": 1.3545603944124898,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007888282486107403,
      "loss": 4.8269,
      "step": 9891
    },
    {
      "epoch": 1.3546973431936455,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007888012888532606,
      "loss": 4.8608,
      "step": 9892
    },
    {
      "epoch": 1.3548342919748015,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007887743262880539,
      "loss": 4.9292,
      "step": 9893
    },
    {
      "epoch": 1.3549712407559573,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007887473609153435,
      "loss": 4.8751,
      "step": 9894
    },
    {
      "epoch": 1.355108189537113,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007887203927353533,
      "loss": 4.8593,
      "step": 9895
    },
    {
      "epoch": 1.3552451383182689,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007886934217483062,
      "loss": 4.8385,
      "step": 9896
    },
    {
      "epoch": 1.3553820870994249,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007886664479544263,
      "loss": 4.8361,
      "step": 9897
    },
    {
      "epoch": 1.3555190358805806,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007886394713539367,
      "loss": 4.8643,
      "step": 9898
    },
    {
      "epoch": 1.3556559846617366,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007886124919470611,
      "loss": 4.8808,
      "step": 9899
    },
    {
      "epoch": 1.3557929334428924,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007885855097340231,
      "loss": 4.9006,
      "step": 9900
    },
    {
      "epoch": 1.3559298822240482,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007885585247150462,
      "loss": 4.8629,
      "step": 9901
    },
    {
      "epoch": 1.356066831005204,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007885315368903543,
      "loss": 4.9118,
      "step": 9902
    },
    {
      "epoch": 1.35620377978636,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007885045462601708,
      "loss": 4.8645,
      "step": 9903
    },
    {
      "epoch": 1.3563407285675158,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007884775528247194,
      "loss": 4.8608,
      "step": 9904
    },
    {
      "epoch": 1.3564776773486715,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007884505565842239,
      "loss": 4.9421,
      "step": 9905
    },
    {
      "epoch": 1.3566146261298275,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007884235575389081,
      "loss": 4.7541,
      "step": 9906
    },
    {
      "epoch": 1.3567515749109833,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007883965556889956,
      "loss": 4.8671,
      "step": 9907
    },
    {
      "epoch": 1.356888523692139,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007883695510347102,
      "loss": 4.8491,
      "step": 9908
    },
    {
      "epoch": 1.3570254724732949,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007883425435762757,
      "loss": 4.9453,
      "step": 9909
    },
    {
      "epoch": 1.3571624212544509,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007883155333139158,
      "loss": 4.8829,
      "step": 9910
    },
    {
      "epoch": 1.3572993700356066,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007882885202478545,
      "loss": 4.8961,
      "step": 9911
    },
    {
      "epoch": 1.3574363188167626,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007882615043783157,
      "loss": 4.8219,
      "step": 9912
    },
    {
      "epoch": 1.3575732675979184,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007882344857055232,
      "loss": 4.8491,
      "step": 9913
    },
    {
      "epoch": 1.3577102163790742,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007882074642297009,
      "loss": 4.7984,
      "step": 9914
    },
    {
      "epoch": 1.35784716516023,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007881804399510728,
      "loss": 4.8636,
      "step": 9915
    },
    {
      "epoch": 1.357984113941386,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007881534128698627,
      "loss": 4.86,
      "step": 9916
    },
    {
      "epoch": 1.3581210627225417,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007881263829862948,
      "loss": 4.8739,
      "step": 9917
    },
    {
      "epoch": 1.3582580115036977,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000788099350300593,
      "loss": 4.9447,
      "step": 9918
    },
    {
      "epoch": 1.3583949602848535,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007880723148129813,
      "loss": 4.8091,
      "step": 9919
    },
    {
      "epoch": 1.3585319090660093,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007880452765236837,
      "loss": 4.8111,
      "step": 9920
    },
    {
      "epoch": 1.358668857847165,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007880182354329244,
      "loss": 4.8377,
      "step": 9921
    },
    {
      "epoch": 1.358805806628321,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007879911915409274,
      "loss": 4.8242,
      "step": 9922
    },
    {
      "epoch": 1.3589427554094768,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0007879641448479169,
      "loss": 4.8502,
      "step": 9923
    },
    {
      "epoch": 1.3590797041906326,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007879370953541169,
      "loss": 4.8971,
      "step": 9924
    },
    {
      "epoch": 1.3592166529717886,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007879100430597518,
      "loss": 4.8577,
      "step": 9925
    },
    {
      "epoch": 1.3593536017529444,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007878829879650456,
      "loss": 4.9169,
      "step": 9926
    },
    {
      "epoch": 1.3594905505341002,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007878559300702225,
      "loss": 4.9026,
      "step": 9927
    },
    {
      "epoch": 1.3596274993152562,
      "grad_norm": 0.375,
      "learning_rate": 0.0007878288693755069,
      "loss": 4.9079,
      "step": 9928
    },
    {
      "epoch": 1.359764448096412,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007878018058811229,
      "loss": 4.8461,
      "step": 9929
    },
    {
      "epoch": 1.3599013968775677,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007877747395872948,
      "loss": 4.8407,
      "step": 9930
    },
    {
      "epoch": 1.3600383456587237,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007877476704942471,
      "loss": 4.8223,
      "step": 9931
    },
    {
      "epoch": 1.3601752944398795,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007877205986022038,
      "loss": 4.9012,
      "step": 9932
    },
    {
      "epoch": 1.3603122432210353,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007876935239113895,
      "loss": 4.7887,
      "step": 9933
    },
    {
      "epoch": 1.360449192002191,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007876664464220286,
      "loss": 4.9154,
      "step": 9934
    },
    {
      "epoch": 1.360586140783347,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007876393661343452,
      "loss": 4.8855,
      "step": 9935
    },
    {
      "epoch": 1.3607230895645028,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007876122830485641,
      "loss": 4.8361,
      "step": 9936
    },
    {
      "epoch": 1.3608600383456588,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007875851971649094,
      "loss": 4.8466,
      "step": 9937
    },
    {
      "epoch": 1.3609969871268146,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007875581084836058,
      "loss": 4.89,
      "step": 9938
    },
    {
      "epoch": 1.3611339359079704,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007875310170048778,
      "loss": 4.7682,
      "step": 9939
    },
    {
      "epoch": 1.3612708846891262,
      "grad_norm": 0.484375,
      "learning_rate": 0.0007875039227289497,
      "loss": 4.9216,
      "step": 9940
    },
    {
      "epoch": 1.3614078334702822,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007874768256560463,
      "loss": 4.8686,
      "step": 9941
    },
    {
      "epoch": 1.361544782251438,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007874497257863919,
      "loss": 4.8344,
      "step": 9942
    },
    {
      "epoch": 1.361681731032594,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007874226231202113,
      "loss": 4.9114,
      "step": 9943
    },
    {
      "epoch": 1.3618186798137497,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007873955176577289,
      "loss": 4.8671,
      "step": 9944
    },
    {
      "epoch": 1.3619556285949055,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007873684093991697,
      "loss": 4.8306,
      "step": 9945
    },
    {
      "epoch": 1.3620925773760613,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007873412983447579,
      "loss": 4.8938,
      "step": 9946
    },
    {
      "epoch": 1.3622295261572173,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007873141844947184,
      "loss": 4.8747,
      "step": 9947
    },
    {
      "epoch": 1.362366474938373,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007872870678492759,
      "loss": 4.8607,
      "step": 9948
    },
    {
      "epoch": 1.3625034237195288,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000787259948408655,
      "loss": 4.876,
      "step": 9949
    },
    {
      "epoch": 1.3626403725006848,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007872328261730808,
      "loss": 4.92,
      "step": 9950
    },
    {
      "epoch": 1.3627773212818406,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007872057011427774,
      "loss": 4.8416,
      "step": 9951
    },
    {
      "epoch": 1.3629142700629964,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007871785733179704,
      "loss": 4.7835,
      "step": 9952
    },
    {
      "epoch": 1.3630512188441524,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000787151442698884,
      "loss": 4.86,
      "step": 9953
    },
    {
      "epoch": 1.3631881676253081,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007871243092857433,
      "loss": 4.8521,
      "step": 9954
    },
    {
      "epoch": 1.363325116406464,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007870971730787733,
      "loss": 4.9455,
      "step": 9955
    },
    {
      "epoch": 1.36346206518762,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007870700340781984,
      "loss": 4.8245,
      "step": 9956
    },
    {
      "epoch": 1.3635990139687757,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000787042892284244,
      "loss": 4.9193,
      "step": 9957
    },
    {
      "epoch": 1.3637359627499315,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007870157476971349,
      "loss": 4.852,
      "step": 9958
    },
    {
      "epoch": 1.3638729115310873,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007869886003170958,
      "loss": 4.8515,
      "step": 9959
    },
    {
      "epoch": 1.3640098603122432,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007869614501443519,
      "loss": 4.8808,
      "step": 9960
    },
    {
      "epoch": 1.364146809093399,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007869342971791282,
      "loss": 4.9091,
      "step": 9961
    },
    {
      "epoch": 1.364283757874555,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007869071414216497,
      "loss": 4.8717,
      "step": 9962
    },
    {
      "epoch": 1.3644207066557108,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007868799828721415,
      "loss": 4.7907,
      "step": 9963
    },
    {
      "epoch": 1.3645576554368666,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007868528215308285,
      "loss": 4.8896,
      "step": 9964
    },
    {
      "epoch": 1.3646946042180224,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000786825657397936,
      "loss": 4.8284,
      "step": 9965
    },
    {
      "epoch": 1.3648315529991784,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007867984904736889,
      "loss": 4.8646,
      "step": 9966
    },
    {
      "epoch": 1.3649685017803341,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007867713207583125,
      "loss": 4.9418,
      "step": 9967
    },
    {
      "epoch": 1.3651054505614901,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007867441482520321,
      "loss": 4.8099,
      "step": 9968
    },
    {
      "epoch": 1.365242399342646,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0007867169729550724,
      "loss": 4.938,
      "step": 9969
    },
    {
      "epoch": 1.3653793481238017,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007866897948676591,
      "loss": 4.8632,
      "step": 9970
    },
    {
      "epoch": 1.3655162969049575,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007866626139900172,
      "loss": 4.9084,
      "step": 9971
    },
    {
      "epoch": 1.3656532456861135,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007866354303223721,
      "loss": 4.8695,
      "step": 9972
    },
    {
      "epoch": 1.3657901944672692,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007866082438649488,
      "loss": 4.8587,
      "step": 9973
    },
    {
      "epoch": 1.365927143248425,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007865810546179726,
      "loss": 4.9205,
      "step": 9974
    },
    {
      "epoch": 1.366064092029581,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007865538625816695,
      "loss": 4.8751,
      "step": 9975
    },
    {
      "epoch": 1.3662010408107368,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000786526667756264,
      "loss": 4.9834,
      "step": 9976
    },
    {
      "epoch": 1.3663379895918926,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007864994701419818,
      "loss": 4.8869,
      "step": 9977
    },
    {
      "epoch": 1.3664749383730483,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007864722697390486,
      "loss": 4.8213,
      "step": 9978
    },
    {
      "epoch": 1.3666118871542043,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007864450665476892,
      "loss": 4.8033,
      "step": 9979
    },
    {
      "epoch": 1.3667488359353601,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007864178605681294,
      "loss": 4.8137,
      "step": 9980
    },
    {
      "epoch": 1.3668857847165161,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007863906518005948,
      "loss": 4.8474,
      "step": 9981
    },
    {
      "epoch": 1.367022733497672,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007863634402453106,
      "loss": 4.9275,
      "step": 9982
    },
    {
      "epoch": 1.3671596822788277,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007863362259025024,
      "loss": 4.8504,
      "step": 9983
    },
    {
      "epoch": 1.3672966310599834,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007863090087723958,
      "loss": 4.8108,
      "step": 9984
    },
    {
      "epoch": 1.3674335798411394,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007862817888552163,
      "loss": 4.8637,
      "step": 9985
    },
    {
      "epoch": 1.3675705286222952,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007862545661511895,
      "loss": 4.9031,
      "step": 9986
    },
    {
      "epoch": 1.3677074774034512,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000786227340660541,
      "loss": 4.9258,
      "step": 9987
    },
    {
      "epoch": 1.367844426184607,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007862001123834964,
      "loss": 4.8756,
      "step": 9988
    },
    {
      "epoch": 1.3679813749657628,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007861728813202813,
      "loss": 4.8107,
      "step": 9989
    },
    {
      "epoch": 1.3681183237469186,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007861456474711215,
      "loss": 4.8408,
      "step": 9990
    },
    {
      "epoch": 1.3682552725280746,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007861184108362426,
      "loss": 4.9742,
      "step": 9991
    },
    {
      "epoch": 1.3683922213092303,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007860911714158703,
      "loss": 4.9225,
      "step": 9992
    },
    {
      "epoch": 1.3685291700903863,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007860639292102303,
      "loss": 4.8159,
      "step": 9993
    },
    {
      "epoch": 1.368666118871542,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007860366842195486,
      "loss": 4.9083,
      "step": 9994
    },
    {
      "epoch": 1.3688030676526979,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007860094364440507,
      "loss": 4.8925,
      "step": 9995
    },
    {
      "epoch": 1.3689400164338537,
      "grad_norm": 0.375,
      "learning_rate": 0.0007859821858839627,
      "loss": 4.8854,
      "step": 9996
    },
    {
      "epoch": 1.3690769652150097,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007859549325395101,
      "loss": 4.8494,
      "step": 9997
    },
    {
      "epoch": 1.3692139139961654,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000785927676410919,
      "loss": 4.815,
      "step": 9998
    },
    {
      "epoch": 1.3693508627773212,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007859004174984151,
      "loss": 4.8251,
      "step": 9999
    },
    {
      "epoch": 1.3694878115584772,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007858731558022245,
      "loss": 4.8441,
      "step": 10000
    },
    {
      "epoch": 1.369624760339633,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007858458913225731,
      "loss": 4.8484,
      "step": 10001
    },
    {
      "epoch": 1.3697617091207888,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007858186240596867,
      "loss": 4.8757,
      "step": 10002
    },
    {
      "epoch": 1.3698986579019445,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007857913540137912,
      "loss": 4.8318,
      "step": 10003
    },
    {
      "epoch": 1.3700356066831005,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007857640811851129,
      "loss": 4.8526,
      "step": 10004
    },
    {
      "epoch": 1.3701725554642563,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007857368055738777,
      "loss": 4.8663,
      "step": 10005
    },
    {
      "epoch": 1.3703095042454123,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007857095271803113,
      "loss": 4.8987,
      "step": 10006
    },
    {
      "epoch": 1.370446453026568,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007856822460046403,
      "loss": 4.7499,
      "step": 10007
    },
    {
      "epoch": 1.3705834018077239,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007856549620470904,
      "loss": 4.9219,
      "step": 10008
    },
    {
      "epoch": 1.3707203505888796,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000785627675307888,
      "loss": 4.9109,
      "step": 10009
    },
    {
      "epoch": 1.3708572993700356,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000785600385787259,
      "loss": 4.8391,
      "step": 10010
    },
    {
      "epoch": 1.3709942481511914,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007855730934854296,
      "loss": 4.8072,
      "step": 10011
    },
    {
      "epoch": 1.3711311969323474,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007855457984026261,
      "loss": 4.9293,
      "step": 10012
    },
    {
      "epoch": 1.3712681457135032,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007855185005390746,
      "loss": 4.8958,
      "step": 10013
    },
    {
      "epoch": 1.371405094494659,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007854911998950013,
      "loss": 4.8529,
      "step": 10014
    },
    {
      "epoch": 1.3715420432758147,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007854638964706325,
      "loss": 4.8435,
      "step": 10015
    },
    {
      "epoch": 1.3716789920569707,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007854365902661944,
      "loss": 4.9493,
      "step": 10016
    },
    {
      "epoch": 1.3718159408381265,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007854092812819134,
      "loss": 4.8054,
      "step": 10017
    },
    {
      "epoch": 1.3719528896192825,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007853819695180157,
      "loss": 4.8282,
      "step": 10018
    },
    {
      "epoch": 1.3720898384004383,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007853546549747277,
      "loss": 4.851,
      "step": 10019
    },
    {
      "epoch": 1.372226787181594,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007853273376522759,
      "loss": 4.8304,
      "step": 10020
    },
    {
      "epoch": 1.3723637359627499,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007853000175508865,
      "loss": 4.9208,
      "step": 10021
    },
    {
      "epoch": 1.3725006847439059,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007852726946707859,
      "loss": 4.8657,
      "step": 10022
    },
    {
      "epoch": 1.3726376335250616,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007852453690122006,
      "loss": 4.8877,
      "step": 10023
    },
    {
      "epoch": 1.3727745823062174,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000785218040575357,
      "loss": 4.9045,
      "step": 10024
    },
    {
      "epoch": 1.3729115310873734,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007851907093604816,
      "loss": 4.7661,
      "step": 10025
    },
    {
      "epoch": 1.3730484798685292,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0007851633753678011,
      "loss": 4.8046,
      "step": 10026
    },
    {
      "epoch": 1.373185428649685,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007851360385975417,
      "loss": 4.9355,
      "step": 10027
    },
    {
      "epoch": 1.3733223774308407,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00078510869904993,
      "loss": 4.8449,
      "step": 10028
    },
    {
      "epoch": 1.3734593262119967,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007850813567251929,
      "loss": 4.8762,
      "step": 10029
    },
    {
      "epoch": 1.3735962749931525,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007850540116235567,
      "loss": 4.9618,
      "step": 10030
    },
    {
      "epoch": 1.3737332237743085,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000785026663745248,
      "loss": 4.87,
      "step": 10031
    },
    {
      "epoch": 1.3738701725554643,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007849993130904933,
      "loss": 4.8844,
      "step": 10032
    },
    {
      "epoch": 1.37400712133662,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007849719596595198,
      "loss": 4.9254,
      "step": 10033
    },
    {
      "epoch": 1.3741440701177758,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007849446034525538,
      "loss": 4.9101,
      "step": 10034
    },
    {
      "epoch": 1.3742810188989318,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007849172444698219,
      "loss": 4.8711,
      "step": 10035
    },
    {
      "epoch": 1.3744179676800876,
      "grad_norm": 0.482421875,
      "learning_rate": 0.000784889882711551,
      "loss": 4.8245,
      "step": 10036
    },
    {
      "epoch": 1.3745549164612436,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007848625181779679,
      "loss": 4.779,
      "step": 10037
    },
    {
      "epoch": 1.3746918652423994,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0007848351508692994,
      "loss": 4.8286,
      "step": 10038
    },
    {
      "epoch": 1.3748288140235552,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007848077807857722,
      "loss": 4.8942,
      "step": 10039
    },
    {
      "epoch": 1.374965762804711,
      "grad_norm": 0.421875,
      "learning_rate": 0.000784780407927613,
      "loss": 4.8427,
      "step": 10040
    },
    {
      "epoch": 1.375102711585867,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007847530322950488,
      "loss": 4.9193,
      "step": 10041
    },
    {
      "epoch": 1.3752396603670227,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007847256538883066,
      "loss": 4.7988,
      "step": 10042
    },
    {
      "epoch": 1.3753766091481787,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007846982727076131,
      "loss": 4.8895,
      "step": 10043
    },
    {
      "epoch": 1.3755135579293345,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007846708887531952,
      "loss": 4.9338,
      "step": 10044
    },
    {
      "epoch": 1.3756505067104903,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007846435020252799,
      "loss": 4.9536,
      "step": 10045
    },
    {
      "epoch": 1.375787455491646,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007846161125240941,
      "loss": 4.8422,
      "step": 10046
    },
    {
      "epoch": 1.375924404272802,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007845887202498651,
      "loss": 4.8126,
      "step": 10047
    },
    {
      "epoch": 1.3760613530539578,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007845613252028193,
      "loss": 4.8262,
      "step": 10048
    },
    {
      "epoch": 1.3761983018351136,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007845339273831844,
      "loss": 4.7753,
      "step": 10049
    },
    {
      "epoch": 1.3763352506162696,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007845065267911869,
      "loss": 4.8223,
      "step": 10050
    },
    {
      "epoch": 1.3764721993974254,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007844791234270541,
      "loss": 4.855,
      "step": 10051
    },
    {
      "epoch": 1.3766091481785812,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007844517172910133,
      "loss": 4.8765,
      "step": 10052
    },
    {
      "epoch": 1.376746096959737,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007844243083832913,
      "loss": 4.8057,
      "step": 10053
    },
    {
      "epoch": 1.376883045740893,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007843968967041154,
      "loss": 4.8185,
      "step": 10054
    },
    {
      "epoch": 1.3770199945220487,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007843694822537126,
      "loss": 4.8858,
      "step": 10055
    },
    {
      "epoch": 1.3771569433032047,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007843420650323104,
      "loss": 4.8261,
      "step": 10056
    },
    {
      "epoch": 1.3772938920843605,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007843146450401358,
      "loss": 4.8971,
      "step": 10057
    },
    {
      "epoch": 1.3774308408655163,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000784287222277416,
      "loss": 4.965,
      "step": 10058
    },
    {
      "epoch": 1.377567789646672,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0007842597967443784,
      "loss": 4.7971,
      "step": 10059
    },
    {
      "epoch": 1.377704738427828,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007842323684412502,
      "loss": 4.7285,
      "step": 10060
    },
    {
      "epoch": 1.3778416872089838,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007842049373682587,
      "loss": 4.9358,
      "step": 10061
    },
    {
      "epoch": 1.3779786359901398,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007841775035256312,
      "loss": 4.8707,
      "step": 10062
    },
    {
      "epoch": 1.3781155847712956,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007841500669135952,
      "loss": 4.8944,
      "step": 10063
    },
    {
      "epoch": 1.3782525335524514,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007841226275323779,
      "loss": 4.8294,
      "step": 10064
    },
    {
      "epoch": 1.3783894823336071,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007840951853822068,
      "loss": 4.8753,
      "step": 10065
    },
    {
      "epoch": 1.3785264311147631,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0007840677404633093,
      "loss": 4.8364,
      "step": 10066
    },
    {
      "epoch": 1.378663379895919,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007840402927759128,
      "loss": 4.8653,
      "step": 10067
    },
    {
      "epoch": 1.3788003286770747,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007840128423202449,
      "loss": 4.7886,
      "step": 10068
    },
    {
      "epoch": 1.3789372774582307,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007839853890965327,
      "loss": 4.7977,
      "step": 10069
    },
    {
      "epoch": 1.3790742262393865,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007839579331050043,
      "loss": 4.8837,
      "step": 10070
    },
    {
      "epoch": 1.3792111750205422,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007839304743458869,
      "loss": 4.8708,
      "step": 10071
    },
    {
      "epoch": 1.3793481238016982,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0007839030128194079,
      "loss": 4.8435,
      "step": 10072
    },
    {
      "epoch": 1.379485072582854,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007838755485257952,
      "loss": 4.8623,
      "step": 10073
    },
    {
      "epoch": 1.3796220213640098,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007838480814652762,
      "loss": 4.8171,
      "step": 10074
    },
    {
      "epoch": 1.3797589701451658,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007838206116380787,
      "loss": 4.9684,
      "step": 10075
    },
    {
      "epoch": 1.3798959189263216,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0007837931390444302,
      "loss": 4.8556,
      "step": 10076
    },
    {
      "epoch": 1.3800328677074774,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007837656636845584,
      "loss": 4.8177,
      "step": 10077
    },
    {
      "epoch": 1.3801698164886331,
      "grad_norm": 0.47265625,
      "learning_rate": 0.000783738185558691,
      "loss": 4.8027,
      "step": 10078
    },
    {
      "epoch": 1.3803067652697891,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0007837107046670558,
      "loss": 4.8427,
      "step": 10079
    },
    {
      "epoch": 1.380443714050945,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007836832210098804,
      "loss": 4.8998,
      "step": 10080
    },
    {
      "epoch": 1.380580662832101,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007836557345873927,
      "loss": 4.8362,
      "step": 10081
    },
    {
      "epoch": 1.3807176116132567,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0007836282453998203,
      "loss": 4.8902,
      "step": 10082
    },
    {
      "epoch": 1.3808545603944125,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007836007534473912,
      "loss": 4.812,
      "step": 10083
    },
    {
      "epoch": 1.3809915091755682,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007835732587303332,
      "loss": 4.9129,
      "step": 10084
    },
    {
      "epoch": 1.3811284579567242,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007835457612488742,
      "loss": 4.77,
      "step": 10085
    },
    {
      "epoch": 1.38126540673788,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007835182610032419,
      "loss": 4.8321,
      "step": 10086
    },
    {
      "epoch": 1.381402355519036,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007834907579936642,
      "loss": 4.7091,
      "step": 10087
    },
    {
      "epoch": 1.3815393043001918,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007834632522203693,
      "loss": 4.8537,
      "step": 10088
    },
    {
      "epoch": 1.3816762530813476,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007834357436835848,
      "loss": 4.7686,
      "step": 10089
    },
    {
      "epoch": 1.3818132018625033,
      "grad_norm": 0.375,
      "learning_rate": 0.0007834082323835389,
      "loss": 4.8349,
      "step": 10090
    },
    {
      "epoch": 1.3819501506436593,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007833807183204597,
      "loss": 4.7816,
      "step": 10091
    },
    {
      "epoch": 1.3820870994248151,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007833532014945748,
      "loss": 4.8998,
      "step": 10092
    },
    {
      "epoch": 1.382224048205971,
      "grad_norm": 0.46875,
      "learning_rate": 0.0007833256819061126,
      "loss": 4.8429,
      "step": 10093
    },
    {
      "epoch": 1.3823609969871269,
      "grad_norm": 0.390625,
      "learning_rate": 0.000783298159555301,
      "loss": 4.8998,
      "step": 10094
    },
    {
      "epoch": 1.3824979457682827,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007832706344423682,
      "loss": 4.8509,
      "step": 10095
    },
    {
      "epoch": 1.3826348945494384,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007832431065675422,
      "loss": 4.8642,
      "step": 10096
    },
    {
      "epoch": 1.3827718433305942,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007832155759310511,
      "loss": 4.8294,
      "step": 10097
    },
    {
      "epoch": 1.3829087921117502,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007831880425331233,
      "loss": 4.8656,
      "step": 10098
    },
    {
      "epoch": 1.383045740892906,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007831605063739867,
      "loss": 4.8916,
      "step": 10099
    },
    {
      "epoch": 1.383182689674062,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007831329674538696,
      "loss": 4.9448,
      "step": 10100
    },
    {
      "epoch": 1.3833196384552178,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007831054257730002,
      "loss": 4.905,
      "step": 10101
    },
    {
      "epoch": 1.3834565872363735,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007830778813316066,
      "loss": 4.7975,
      "step": 10102
    },
    {
      "epoch": 1.3835935360175293,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007830503341299175,
      "loss": 4.826,
      "step": 10103
    },
    {
      "epoch": 1.3837304847986853,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007830227841681608,
      "loss": 4.8548,
      "step": 10104
    },
    {
      "epoch": 1.383867433579841,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007829952314465649,
      "loss": 4.8107,
      "step": 10105
    },
    {
      "epoch": 1.384004382360997,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007829676759653582,
      "loss": 4.8858,
      "step": 10106
    },
    {
      "epoch": 1.3841413311421529,
      "grad_norm": 0.421875,
      "learning_rate": 0.000782940117724769,
      "loss": 4.8941,
      "step": 10107
    },
    {
      "epoch": 1.3842782799233087,
      "grad_norm": 0.375,
      "learning_rate": 0.0007829125567250258,
      "loss": 4.9192,
      "step": 10108
    },
    {
      "epoch": 1.3844152287044644,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007828849929663569,
      "loss": 4.9678,
      "step": 10109
    },
    {
      "epoch": 1.3845521774856204,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007828574264489907,
      "loss": 4.747,
      "step": 10110
    },
    {
      "epoch": 1.3846891262667762,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007828298571731557,
      "loss": 4.8613,
      "step": 10111
    },
    {
      "epoch": 1.3848260750479322,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007828022851390803,
      "loss": 4.8923,
      "step": 10112
    },
    {
      "epoch": 1.384963023829088,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007827747103469932,
      "loss": 4.8686,
      "step": 10113
    },
    {
      "epoch": 1.3850999726102438,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007827471327971227,
      "loss": 4.7859,
      "step": 10114
    },
    {
      "epoch": 1.3852369213913995,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007827195524896974,
      "loss": 4.8598,
      "step": 10115
    },
    {
      "epoch": 1.3853738701725555,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007826919694249459,
      "loss": 4.8764,
      "step": 10116
    },
    {
      "epoch": 1.3855108189537113,
      "grad_norm": 0.375,
      "learning_rate": 0.0007826643836030967,
      "loss": 4.8965,
      "step": 10117
    },
    {
      "epoch": 1.385647767734867,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007826367950243787,
      "loss": 4.8388,
      "step": 10118
    },
    {
      "epoch": 1.385784716516023,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007826092036890202,
      "loss": 4.898,
      "step": 10119
    },
    {
      "epoch": 1.3859216652971789,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007825816095972498,
      "loss": 4.8986,
      "step": 10120
    },
    {
      "epoch": 1.3860586140783346,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007825540127492966,
      "loss": 4.915,
      "step": 10121
    },
    {
      "epoch": 1.3861955628594904,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000782526413145389,
      "loss": 4.8935,
      "step": 10122
    },
    {
      "epoch": 1.3863325116406464,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007824988107857558,
      "loss": 4.9707,
      "step": 10123
    },
    {
      "epoch": 1.3864694604218022,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007824712056706256,
      "loss": 4.7719,
      "step": 10124
    },
    {
      "epoch": 1.3866064092029582,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007824435978002274,
      "loss": 4.6996,
      "step": 10125
    },
    {
      "epoch": 1.386743357984114,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007824159871747898,
      "loss": 4.8426,
      "step": 10126
    },
    {
      "epoch": 1.3868803067652697,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007823883737945417,
      "loss": 4.8813,
      "step": 10127
    },
    {
      "epoch": 1.3870172555464255,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007823607576597119,
      "loss": 4.8607,
      "step": 10128
    },
    {
      "epoch": 1.3871542043275815,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007823331387705294,
      "loss": 4.8295,
      "step": 10129
    },
    {
      "epoch": 1.3872911531087373,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000782305517127223,
      "loss": 4.8973,
      "step": 10130
    },
    {
      "epoch": 1.3874281018898933,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0007822778927300217,
      "loss": 4.8455,
      "step": 10131
    },
    {
      "epoch": 1.387565050671049,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007822502655791541,
      "loss": 4.8756,
      "step": 10132
    },
    {
      "epoch": 1.3877019994522048,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007822226356748495,
      "loss": 4.806,
      "step": 10133
    },
    {
      "epoch": 1.3878389482333606,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007821950030173368,
      "loss": 4.8032,
      "step": 10134
    },
    {
      "epoch": 1.3879758970145166,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007821673676068448,
      "loss": 4.8432,
      "step": 10135
    },
    {
      "epoch": 1.3881128457956724,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007821397294436028,
      "loss": 4.8341,
      "step": 10136
    },
    {
      "epoch": 1.3882497945768284,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007821120885278397,
      "loss": 4.8711,
      "step": 10137
    },
    {
      "epoch": 1.3883867433579842,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007820844448597845,
      "loss": 4.7949,
      "step": 10138
    },
    {
      "epoch": 1.38852369213914,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007820567984396664,
      "loss": 4.9395,
      "step": 10139
    },
    {
      "epoch": 1.3886606409202957,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007820291492677146,
      "loss": 4.8556,
      "step": 10140
    },
    {
      "epoch": 1.3887975897014517,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000782001497344158,
      "loss": 4.8935,
      "step": 10141
    },
    {
      "epoch": 1.3889345384826075,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007819738426692259,
      "loss": 4.8869,
      "step": 10142
    },
    {
      "epoch": 1.3890714872637633,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007819461852431476,
      "loss": 4.8899,
      "step": 10143
    },
    {
      "epoch": 1.3892084360449193,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000781918525066152,
      "loss": 4.916,
      "step": 10144
    },
    {
      "epoch": 1.389345384826075,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007818908621384685,
      "loss": 4.8581,
      "step": 10145
    },
    {
      "epoch": 1.3894823336072308,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007818631964603265,
      "loss": 4.8159,
      "step": 10146
    },
    {
      "epoch": 1.3896192823883866,
      "grad_norm": 0.375,
      "learning_rate": 0.0007818355280319549,
      "loss": 4.8386,
      "step": 10147
    },
    {
      "epoch": 1.3897562311695426,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007818078568535833,
      "loss": 4.8876,
      "step": 10148
    },
    {
      "epoch": 1.3898931799506984,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007817801829254409,
      "loss": 4.8666,
      "step": 10149
    },
    {
      "epoch": 1.3900301287318544,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007817525062477572,
      "loss": 4.9004,
      "step": 10150
    },
    {
      "epoch": 1.3901670775130102,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007817248268207613,
      "loss": 4.8173,
      "step": 10151
    },
    {
      "epoch": 1.390304026294166,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007816971446446828,
      "loss": 4.8652,
      "step": 10152
    },
    {
      "epoch": 1.3904409750753217,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007816694597197508,
      "loss": 4.8546,
      "step": 10153
    },
    {
      "epoch": 1.3905779238564777,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007816417720461951,
      "loss": 4.8804,
      "step": 10154
    },
    {
      "epoch": 1.3907148726376335,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000781614081624245,
      "loss": 4.8606,
      "step": 10155
    },
    {
      "epoch": 1.3908518214187895,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007815863884541301,
      "loss": 4.9325,
      "step": 10156
    },
    {
      "epoch": 1.3909887701999453,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007815586925360797,
      "loss": 4.8662,
      "step": 10157
    },
    {
      "epoch": 1.391125718981101,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007815309938703234,
      "loss": 4.8519,
      "step": 10158
    },
    {
      "epoch": 1.3912626677622568,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007815032924570908,
      "loss": 4.8919,
      "step": 10159
    },
    {
      "epoch": 1.3913996165434128,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007814755882966113,
      "loss": 4.9136,
      "step": 10160
    },
    {
      "epoch": 1.3915365653245686,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007814478813891147,
      "loss": 4.7737,
      "step": 10161
    },
    {
      "epoch": 1.3916735141057246,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007814201717348305,
      "loss": 4.8521,
      "step": 10162
    },
    {
      "epoch": 1.3918104628868804,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007813924593339884,
      "loss": 4.9141,
      "step": 10163
    },
    {
      "epoch": 1.3919474116680361,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000781364744186818,
      "loss": 4.8696,
      "step": 10164
    },
    {
      "epoch": 1.392084360449192,
      "grad_norm": 0.421875,
      "learning_rate": 0.000781337026293549,
      "loss": 4.8208,
      "step": 10165
    },
    {
      "epoch": 1.392221309230348,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007813093056544112,
      "loss": 4.8651,
      "step": 10166
    },
    {
      "epoch": 1.3923582580115037,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007812815822696341,
      "loss": 4.8816,
      "step": 10167
    },
    {
      "epoch": 1.3924952067926595,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007812538561394476,
      "loss": 4.8472,
      "step": 10168
    },
    {
      "epoch": 1.3926321555738155,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007812261272640815,
      "loss": 4.8042,
      "step": 10169
    },
    {
      "epoch": 1.3927691043549713,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007811983956437656,
      "loss": 4.8936,
      "step": 10170
    },
    {
      "epoch": 1.392906053136127,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007811706612787297,
      "loss": 4.8731,
      "step": 10171
    },
    {
      "epoch": 1.3930430019172828,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007811429241692035,
      "loss": 4.8541,
      "step": 10172
    },
    {
      "epoch": 1.3931799506984388,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0007811151843154171,
      "loss": 4.8361,
      "step": 10173
    },
    {
      "epoch": 1.3933168994795946,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007810874417176001,
      "loss": 4.8439,
      "step": 10174
    },
    {
      "epoch": 1.3934538482607506,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007810596963759828,
      "loss": 4.8988,
      "step": 10175
    },
    {
      "epoch": 1.3935907970419064,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007810319482907947,
      "loss": 4.8335,
      "step": 10176
    },
    {
      "epoch": 1.3937277458230621,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000781004197462266,
      "loss": 4.8288,
      "step": 10177
    },
    {
      "epoch": 1.393864694604218,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007809764438906268,
      "loss": 4.8564,
      "step": 10178
    },
    {
      "epoch": 1.394001643385374,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007809486875761068,
      "loss": 4.796,
      "step": 10179
    },
    {
      "epoch": 1.3941385921665297,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007809209285189362,
      "loss": 4.8968,
      "step": 10180
    },
    {
      "epoch": 1.3942755409476857,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000780893166719345,
      "loss": 4.8588,
      "step": 10181
    },
    {
      "epoch": 1.3944124897288415,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007808654021775633,
      "loss": 4.9072,
      "step": 10182
    },
    {
      "epoch": 1.3945494385099972,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007808376348938212,
      "loss": 4.7982,
      "step": 10183
    },
    {
      "epoch": 1.394686387291153,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007808098648683486,
      "loss": 4.8712,
      "step": 10184
    },
    {
      "epoch": 1.394823336072309,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000780782092101376,
      "loss": 4.8566,
      "step": 10185
    },
    {
      "epoch": 1.3949602848534648,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007807543165931335,
      "loss": 4.868,
      "step": 10186
    },
    {
      "epoch": 1.3950972336346206,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007807265383438509,
      "loss": 4.8713,
      "step": 10187
    },
    {
      "epoch": 1.3952341824157766,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007806987573537588,
      "loss": 4.8027,
      "step": 10188
    },
    {
      "epoch": 1.3953711311969323,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007806709736230873,
      "loss": 4.8614,
      "step": 10189
    },
    {
      "epoch": 1.3955080799780881,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007806431871520667,
      "loss": 4.8807,
      "step": 10190
    },
    {
      "epoch": 1.3956450287592441,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007806153979409271,
      "loss": 4.8877,
      "step": 10191
    },
    {
      "epoch": 1.3957819775404,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007805876059898989,
      "loss": 4.8213,
      "step": 10192
    },
    {
      "epoch": 1.3959189263215557,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007805598112992125,
      "loss": 4.7981,
      "step": 10193
    },
    {
      "epoch": 1.3960558751027117,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007805320138690982,
      "loss": 4.9067,
      "step": 10194
    },
    {
      "epoch": 1.3961928238838675,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007805042136997864,
      "loss": 4.8636,
      "step": 10195
    },
    {
      "epoch": 1.3963297726650232,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007804764107915073,
      "loss": 4.8749,
      "step": 10196
    },
    {
      "epoch": 1.396466721446179,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007804486051444915,
      "loss": 4.809,
      "step": 10197
    },
    {
      "epoch": 1.396603670227335,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007804207967589692,
      "loss": 4.8626,
      "step": 10198
    },
    {
      "epoch": 1.3967406190084908,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007803929856351712,
      "loss": 4.8902,
      "step": 10199
    },
    {
      "epoch": 1.3968775677896468,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007803651717733278,
      "loss": 4.893,
      "step": 10200
    },
    {
      "epoch": 1.3970145165708026,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007803373551736695,
      "loss": 4.8721,
      "step": 10201
    },
    {
      "epoch": 1.3971514653519583,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007803095358364267,
      "loss": 4.8575,
      "step": 10202
    },
    {
      "epoch": 1.397288414133114,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007802817137618302,
      "loss": 4.8613,
      "step": 10203
    },
    {
      "epoch": 1.39742536291427,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007802538889501104,
      "loss": 4.8252,
      "step": 10204
    },
    {
      "epoch": 1.3975623116954259,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000780226061401498,
      "loss": 4.8797,
      "step": 10205
    },
    {
      "epoch": 1.3976992604765819,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007801982311162233,
      "loss": 4.7972,
      "step": 10206
    },
    {
      "epoch": 1.3978362092577377,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007801703980945174,
      "loss": 4.8959,
      "step": 10207
    },
    {
      "epoch": 1.3979731580388934,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007801425623366106,
      "loss": 4.9083,
      "step": 10208
    },
    {
      "epoch": 1.3981101068200492,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000780114723842734,
      "loss": 4.7748,
      "step": 10209
    },
    {
      "epoch": 1.3982470556012052,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007800868826131176,
      "loss": 4.8236,
      "step": 10210
    },
    {
      "epoch": 1.398384004382361,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007800590386479928,
      "loss": 4.8028,
      "step": 10211
    },
    {
      "epoch": 1.3985209531635168,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00078003119194759,
      "loss": 4.8882,
      "step": 10212
    },
    {
      "epoch": 1.3986579019446728,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007800033425121403,
      "loss": 4.837,
      "step": 10213
    },
    {
      "epoch": 1.3987948507258285,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007799754903418739,
      "loss": 4.7825,
      "step": 10214
    },
    {
      "epoch": 1.3989317995069843,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007799476354370222,
      "loss": 4.8706,
      "step": 10215
    },
    {
      "epoch": 1.3990687482881403,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007799197777978159,
      "loss": 4.8488,
      "step": 10216
    },
    {
      "epoch": 1.399205697069296,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007798919174244857,
      "loss": 4.8956,
      "step": 10217
    },
    {
      "epoch": 1.3993426458504519,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007798640543172625,
      "loss": 4.9367,
      "step": 10218
    },
    {
      "epoch": 1.3994795946316079,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007798361884763773,
      "loss": 4.8798,
      "step": 10219
    },
    {
      "epoch": 1.3996165434127636,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007798083199020611,
      "loss": 4.8572,
      "step": 10220
    },
    {
      "epoch": 1.3997534921939194,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007797804485945446,
      "loss": 4.8992,
      "step": 10221
    },
    {
      "epoch": 1.3998904409750752,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007797525745540592,
      "loss": 4.7757,
      "step": 10222
    },
    {
      "epoch": 1.4000273897562312,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007797246977808354,
      "loss": 4.8009,
      "step": 10223
    },
    {
      "epoch": 1.400164338537387,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007796968182751044,
      "loss": 4.9038,
      "step": 10224
    },
    {
      "epoch": 1.400301287318543,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007796689360370976,
      "loss": 4.814,
      "step": 10225
    },
    {
      "epoch": 1.4004382360996988,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007796410510670456,
      "loss": 4.8552,
      "step": 10226
    },
    {
      "epoch": 1.4005751848808545,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007796131633651796,
      "loss": 4.8372,
      "step": 10227
    },
    {
      "epoch": 1.4007121336620103,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007795852729317309,
      "loss": 4.8774,
      "step": 10228
    },
    {
      "epoch": 1.4008490824431663,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007795573797669305,
      "loss": 4.8369,
      "step": 10229
    },
    {
      "epoch": 1.400986031224322,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007795294838710096,
      "loss": 4.8386,
      "step": 10230
    },
    {
      "epoch": 1.401122980005478,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007795015852441993,
      "loss": 4.8406,
      "step": 10231
    },
    {
      "epoch": 1.4012599287866339,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007794736838867308,
      "loss": 4.895,
      "step": 10232
    },
    {
      "epoch": 1.4013968775677896,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007794457797988352,
      "loss": 4.8132,
      "step": 10233
    },
    {
      "epoch": 1.4015338263489454,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007794178729807442,
      "loss": 4.8518,
      "step": 10234
    },
    {
      "epoch": 1.4016707751301014,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007793899634326888,
      "loss": 4.8181,
      "step": 10235
    },
    {
      "epoch": 1.4018077239112572,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007793620511549001,
      "loss": 4.8345,
      "step": 10236
    },
    {
      "epoch": 1.401944672692413,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007793341361476097,
      "loss": 4.8406,
      "step": 10237
    },
    {
      "epoch": 1.402081621473569,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007793062184110489,
      "loss": 4.8597,
      "step": 10238
    },
    {
      "epoch": 1.4022185702547247,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007792782979454489,
      "loss": 4.7878,
      "step": 10239
    },
    {
      "epoch": 1.4023555190358805,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007792503747510412,
      "loss": 4.9042,
      "step": 10240
    },
    {
      "epoch": 1.4024924678170363,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007792224488280572,
      "loss": 4.9,
      "step": 10241
    },
    {
      "epoch": 1.4026294165981923,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007791945201767282,
      "loss": 4.8962,
      "step": 10242
    },
    {
      "epoch": 1.402766365379348,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000779166588797286,
      "loss": 4.8602,
      "step": 10243
    },
    {
      "epoch": 1.402903314160504,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007791386546899617,
      "loss": 4.9204,
      "step": 10244
    },
    {
      "epoch": 1.4030402629416598,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000779110717854987,
      "loss": 4.8633,
      "step": 10245
    },
    {
      "epoch": 1.4031772117228156,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007790827782925933,
      "loss": 4.7687,
      "step": 10246
    },
    {
      "epoch": 1.4033141605039714,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007790548360030122,
      "loss": 4.808,
      "step": 10247
    },
    {
      "epoch": 1.4034511092851274,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007790268909864752,
      "loss": 4.8877,
      "step": 10248
    },
    {
      "epoch": 1.4035880580662832,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000778998943243214,
      "loss": 4.7983,
      "step": 10249
    },
    {
      "epoch": 1.4037250068474392,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007789709927734603,
      "loss": 4.9324,
      "step": 10250
    },
    {
      "epoch": 1.403861955628595,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007789430395774455,
      "loss": 4.9329,
      "step": 10251
    },
    {
      "epoch": 1.4039989044097507,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007789150836554013,
      "loss": 4.8252,
      "step": 10252
    },
    {
      "epoch": 1.4041358531909065,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007788871250075595,
      "loss": 4.8551,
      "step": 10253
    },
    {
      "epoch": 1.4042728019720625,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007788591636341517,
      "loss": 4.8673,
      "step": 10254
    },
    {
      "epoch": 1.4044097507532183,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007788311995354096,
      "loss": 4.891,
      "step": 10255
    },
    {
      "epoch": 1.4045466995343743,
      "grad_norm": 0.375,
      "learning_rate": 0.000778803232711565,
      "loss": 4.839,
      "step": 10256
    },
    {
      "epoch": 1.40468364831553,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007787752631628497,
      "loss": 4.807,
      "step": 10257
    },
    {
      "epoch": 1.4048205970966858,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007787472908894955,
      "loss": 4.8917,
      "step": 10258
    },
    {
      "epoch": 1.4049575458778416,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007787193158917341,
      "loss": 4.8703,
      "step": 10259
    },
    {
      "epoch": 1.4050944946589976,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007786913381697974,
      "loss": 4.8419,
      "step": 10260
    },
    {
      "epoch": 1.4052314434401534,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007786633577239172,
      "loss": 4.8782,
      "step": 10261
    },
    {
      "epoch": 1.4053683922213092,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007786353745543256,
      "loss": 4.8165,
      "step": 10262
    },
    {
      "epoch": 1.4055053410024652,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007786073886612543,
      "loss": 4.8711,
      "step": 10263
    },
    {
      "epoch": 1.405642289783621,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007785794000449351,
      "loss": 4.9134,
      "step": 10264
    },
    {
      "epoch": 1.4057792385647767,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007785514087056002,
      "loss": 4.846,
      "step": 10265
    },
    {
      "epoch": 1.4059161873459325,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007785234146434815,
      "loss": 4.8317,
      "step": 10266
    },
    {
      "epoch": 1.4060531361270885,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007784954178588111,
      "loss": 4.8923,
      "step": 10267
    },
    {
      "epoch": 1.4061900849082443,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007784674183518208,
      "loss": 4.8668,
      "step": 10268
    },
    {
      "epoch": 1.4063270336894003,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007784394161227426,
      "loss": 4.8307,
      "step": 10269
    },
    {
      "epoch": 1.406463982470556,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007784114111718089,
      "loss": 4.8986,
      "step": 10270
    },
    {
      "epoch": 1.4066009312517118,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007783834034992514,
      "loss": 4.8431,
      "step": 10271
    },
    {
      "epoch": 1.4067378800328676,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007783553931053025,
      "loss": 4.8921,
      "step": 10272
    },
    {
      "epoch": 1.4068748288140236,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007783273799901942,
      "loss": 4.7357,
      "step": 10273
    },
    {
      "epoch": 1.4070117775951794,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007782993641541586,
      "loss": 4.8802,
      "step": 10274
    },
    {
      "epoch": 1.4071487263763354,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007782713455974281,
      "loss": 4.9046,
      "step": 10275
    },
    {
      "epoch": 1.4072856751574911,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007782433243202347,
      "loss": 4.8406,
      "step": 10276
    },
    {
      "epoch": 1.407422623938647,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007782153003228105,
      "loss": 4.8124,
      "step": 10277
    },
    {
      "epoch": 1.4075595727198027,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007781872736053881,
      "loss": 4.7782,
      "step": 10278
    },
    {
      "epoch": 1.4076965215009587,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007781592441681994,
      "loss": 5.0035,
      "step": 10279
    },
    {
      "epoch": 1.4078334702821145,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007781312120114769,
      "loss": 4.8802,
      "step": 10280
    },
    {
      "epoch": 1.4079704190632705,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007781031771354528,
      "loss": 4.8664,
      "step": 10281
    },
    {
      "epoch": 1.4081073678444262,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007780751395403597,
      "loss": 4.9139,
      "step": 10282
    },
    {
      "epoch": 1.408244316625582,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007780470992264295,
      "loss": 4.8582,
      "step": 10283
    },
    {
      "epoch": 1.4083812654067378,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007780190561938949,
      "loss": 4.974,
      "step": 10284
    },
    {
      "epoch": 1.4085182141878938,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007779910104429882,
      "loss": 4.8267,
      "step": 10285
    },
    {
      "epoch": 1.4086551629690496,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007779629619739419,
      "loss": 4.8091,
      "step": 10286
    },
    {
      "epoch": 1.4087921117502054,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007779349107869882,
      "loss": 4.8827,
      "step": 10287
    },
    {
      "epoch": 1.4089290605313614,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007779068568823601,
      "loss": 4.9126,
      "step": 10288
    },
    {
      "epoch": 1.4090660093125171,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007778788002602895,
      "loss": 4.9362,
      "step": 10289
    },
    {
      "epoch": 1.409202958093673,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007778507409210093,
      "loss": 4.8766,
      "step": 10290
    },
    {
      "epoch": 1.4093399068748287,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007778226788647517,
      "loss": 4.8455,
      "step": 10291
    },
    {
      "epoch": 1.4094768556559847,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007777946140917496,
      "loss": 4.925,
      "step": 10292
    },
    {
      "epoch": 1.4096138044371405,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007777665466022352,
      "loss": 4.7784,
      "step": 10293
    },
    {
      "epoch": 1.4097507532182965,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007777384763964414,
      "loss": 4.8497,
      "step": 10294
    },
    {
      "epoch": 1.4098877019994522,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007777104034746009,
      "loss": 4.7669,
      "step": 10295
    },
    {
      "epoch": 1.410024650780608,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007776823278369462,
      "loss": 4.8993,
      "step": 10296
    },
    {
      "epoch": 1.4101615995617638,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00077765424948371,
      "loss": 4.8996,
      "step": 10297
    },
    {
      "epoch": 1.4102985483429198,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007776261684151247,
      "loss": 4.8294,
      "step": 10298
    },
    {
      "epoch": 1.4104354971240756,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007775980846314235,
      "loss": 4.814,
      "step": 10299
    },
    {
      "epoch": 1.4105724459052316,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007775699981328388,
      "loss": 4.7857,
      "step": 10300
    },
    {
      "epoch": 1.4107093946863873,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007775419089196035,
      "loss": 4.9141,
      "step": 10301
    },
    {
      "epoch": 1.4108463434675431,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007775138169919504,
      "loss": 4.9402,
      "step": 10302
    },
    {
      "epoch": 1.410983292248699,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007774857223501122,
      "loss": 4.8409,
      "step": 10303
    },
    {
      "epoch": 1.411120241029855,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007774576249943219,
      "loss": 4.9251,
      "step": 10304
    },
    {
      "epoch": 1.4112571898110107,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000777429524924812,
      "loss": 4.8613,
      "step": 10305
    },
    {
      "epoch": 1.4113941385921664,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007774014221418157,
      "loss": 4.909,
      "step": 10306
    },
    {
      "epoch": 1.4115310873733224,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007773733166455658,
      "loss": 4.7649,
      "step": 10307
    },
    {
      "epoch": 1.4116680361544782,
      "grad_norm": 0.375,
      "learning_rate": 0.0007773452084362953,
      "loss": 4.847,
      "step": 10308
    },
    {
      "epoch": 1.411804984935634,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000777317097514237,
      "loss": 4.852,
      "step": 10309
    },
    {
      "epoch": 1.41194193371679,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000777288983879624,
      "loss": 4.955,
      "step": 10310
    },
    {
      "epoch": 1.4120788824979458,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007772608675326891,
      "loss": 4.7691,
      "step": 10311
    },
    {
      "epoch": 1.4122158312791016,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007772327484736654,
      "loss": 4.8788,
      "step": 10312
    },
    {
      "epoch": 1.4123527800602576,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007772046267027859,
      "loss": 4.8636,
      "step": 10313
    },
    {
      "epoch": 1.4124897288414133,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007771765022202839,
      "loss": 4.8279,
      "step": 10314
    },
    {
      "epoch": 1.412626677622569,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007771483750263921,
      "loss": 4.8414,
      "step": 10315
    },
    {
      "epoch": 1.4127636264037249,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007771202451213438,
      "loss": 4.9097,
      "step": 10316
    },
    {
      "epoch": 1.4129005751848809,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007770921125053719,
      "loss": 4.8893,
      "step": 10317
    },
    {
      "epoch": 1.4130375239660367,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00077706397717871,
      "loss": 4.8696,
      "step": 10318
    },
    {
      "epoch": 1.4131744727471927,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007770358391415909,
      "loss": 4.7352,
      "step": 10319
    },
    {
      "epoch": 1.4133114215283484,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007770076983942478,
      "loss": 4.8369,
      "step": 10320
    },
    {
      "epoch": 1.4134483703095042,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000776979554936914,
      "loss": 4.8399,
      "step": 10321
    },
    {
      "epoch": 1.41358531909066,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007769514087698228,
      "loss": 4.8421,
      "step": 10322
    },
    {
      "epoch": 1.413722267871816,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0007769232598932073,
      "loss": 4.8466,
      "step": 10323
    },
    {
      "epoch": 1.4138592166529718,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007768951083073008,
      "loss": 4.8652,
      "step": 10324
    },
    {
      "epoch": 1.4139961654341278,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0007768669540123368,
      "loss": 4.8857,
      "step": 10325
    },
    {
      "epoch": 1.4141331142152835,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007768387970085483,
      "loss": 4.8153,
      "step": 10326
    },
    {
      "epoch": 1.4142700629964393,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007768106372961689,
      "loss": 4.8224,
      "step": 10327
    },
    {
      "epoch": 1.414407011777595,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007767824748754318,
      "loss": 4.8943,
      "step": 10328
    },
    {
      "epoch": 1.414543960558751,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007767543097465706,
      "loss": 4.8511,
      "step": 10329
    },
    {
      "epoch": 1.4146809093399069,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007767261419098185,
      "loss": 4.8694,
      "step": 10330
    },
    {
      "epoch": 1.4148178581210626,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007766979713654088,
      "loss": 4.9546,
      "step": 10331
    },
    {
      "epoch": 1.4149548069022186,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007766697981135754,
      "loss": 4.8724,
      "step": 10332
    },
    {
      "epoch": 1.4150917556833744,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007766416221545514,
      "loss": 4.8724,
      "step": 10333
    },
    {
      "epoch": 1.4152287044645302,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007766134434885706,
      "loss": 4.9476,
      "step": 10334
    },
    {
      "epoch": 1.4153656532456862,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007765852621158663,
      "loss": 4.8548,
      "step": 10335
    },
    {
      "epoch": 1.415502602026842,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007765570780366722,
      "loss": 4.9247,
      "step": 10336
    },
    {
      "epoch": 1.4156395508079977,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007765288912512217,
      "loss": 4.8833,
      "step": 10337
    },
    {
      "epoch": 1.4157764995891537,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007765007017597484,
      "loss": 4.8958,
      "step": 10338
    },
    {
      "epoch": 1.4159134483703095,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007764725095624861,
      "loss": 4.9403,
      "step": 10339
    },
    {
      "epoch": 1.4160503971514653,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007764443146596682,
      "loss": 4.7861,
      "step": 10340
    },
    {
      "epoch": 1.416187345932621,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007764161170515286,
      "loss": 4.8521,
      "step": 10341
    },
    {
      "epoch": 1.416324294713777,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007763879167383007,
      "loss": 4.7667,
      "step": 10342
    },
    {
      "epoch": 1.4164612434949329,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007763597137202185,
      "loss": 4.8811,
      "step": 10343
    },
    {
      "epoch": 1.4165981922760889,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007763315079975156,
      "loss": 4.7985,
      "step": 10344
    },
    {
      "epoch": 1.4167351410572446,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007763032995704256,
      "loss": 4.813,
      "step": 10345
    },
    {
      "epoch": 1.4168720898384004,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007762750884391825,
      "loss": 4.7654,
      "step": 10346
    },
    {
      "epoch": 1.4170090386195562,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00077624687460402,
      "loss": 4.8459,
      "step": 10347
    },
    {
      "epoch": 1.4171459874007122,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007762186580651719,
      "loss": 4.9103,
      "step": 10348
    },
    {
      "epoch": 1.417282936181868,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000776190438822872,
      "loss": 4.7992,
      "step": 10349
    },
    {
      "epoch": 1.417419884963024,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007761622168773542,
      "loss": 4.8686,
      "step": 10350
    },
    {
      "epoch": 1.4175568337441797,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007761339922288524,
      "loss": 4.8493,
      "step": 10351
    },
    {
      "epoch": 1.4176937825253355,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007761057648776005,
      "loss": 4.9651,
      "step": 10352
    },
    {
      "epoch": 1.4178307313064913,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007760775348238325,
      "loss": 4.7682,
      "step": 10353
    },
    {
      "epoch": 1.4179676800876473,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007760493020677823,
      "loss": 4.7567,
      "step": 10354
    },
    {
      "epoch": 1.418104628868803,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007760210666096837,
      "loss": 4.7844,
      "step": 10355
    },
    {
      "epoch": 1.4182415776499588,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007759928284497709,
      "loss": 4.9255,
      "step": 10356
    },
    {
      "epoch": 1.4183785264311148,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007759645875882778,
      "loss": 4.801,
      "step": 10357
    },
    {
      "epoch": 1.4185154752122706,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007759363440254385,
      "loss": 4.7916,
      "step": 10358
    },
    {
      "epoch": 1.4186524239934264,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007759080977614871,
      "loss": 4.8924,
      "step": 10359
    },
    {
      "epoch": 1.4187893727745822,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007758798487966576,
      "loss": 4.8154,
      "step": 10360
    },
    {
      "epoch": 1.4189263215557382,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000775851597131184,
      "loss": 4.9327,
      "step": 10361
    },
    {
      "epoch": 1.419063270336894,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007758233427653009,
      "loss": 4.8619,
      "step": 10362
    },
    {
      "epoch": 1.41920021911805,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007757950856992419,
      "loss": 4.8397,
      "step": 10363
    },
    {
      "epoch": 1.4193371678992057,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007757668259332414,
      "loss": 4.863,
      "step": 10364
    },
    {
      "epoch": 1.4194741166803615,
      "grad_norm": 0.375,
      "learning_rate": 0.0007757385634675337,
      "loss": 4.9088,
      "step": 10365
    },
    {
      "epoch": 1.4196110654615173,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007757102983023527,
      "loss": 4.8413,
      "step": 10366
    },
    {
      "epoch": 1.4197480142426733,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007756820304379331,
      "loss": 4.878,
      "step": 10367
    },
    {
      "epoch": 1.419884963023829,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007756537598745086,
      "loss": 4.8745,
      "step": 10368
    },
    {
      "epoch": 1.420021911804985,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000775625486612314,
      "loss": 4.9497,
      "step": 10369
    },
    {
      "epoch": 1.4201588605861408,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007755972106515835,
      "loss": 4.8959,
      "step": 10370
    },
    {
      "epoch": 1.4202958093672966,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007755689319925511,
      "loss": 4.8886,
      "step": 10371
    },
    {
      "epoch": 1.4204327581484524,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0007755406506354515,
      "loss": 4.8211,
      "step": 10372
    },
    {
      "epoch": 1.4205697069296084,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007755123665805189,
      "loss": 4.8363,
      "step": 10373
    },
    {
      "epoch": 1.4207066557107642,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007754840798279879,
      "loss": 4.9394,
      "step": 10374
    },
    {
      "epoch": 1.4208436044919202,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007754557903780926,
      "loss": 4.9092,
      "step": 10375
    },
    {
      "epoch": 1.420980553273076,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007754274982310676,
      "loss": 4.8449,
      "step": 10376
    },
    {
      "epoch": 1.4211175020542317,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007753992033871475,
      "loss": 4.8664,
      "step": 10377
    },
    {
      "epoch": 1.4212544508353875,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007753709058465666,
      "loss": 4.8313,
      "step": 10378
    },
    {
      "epoch": 1.4213913996165435,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007753426056095595,
      "loss": 4.7721,
      "step": 10379
    },
    {
      "epoch": 1.4215283483976993,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007753143026763607,
      "loss": 4.8982,
      "step": 10380
    },
    {
      "epoch": 1.421665297178855,
      "grad_norm": 0.375,
      "learning_rate": 0.0007752859970472047,
      "loss": 4.7735,
      "step": 10381
    },
    {
      "epoch": 1.421802245960011,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0007752576887223261,
      "loss": 4.9612,
      "step": 10382
    },
    {
      "epoch": 1.4219391947411668,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007752293777019597,
      "loss": 4.8403,
      "step": 10383
    },
    {
      "epoch": 1.4220761435223226,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007752010639863398,
      "loss": 4.7338,
      "step": 10384
    },
    {
      "epoch": 1.4222130923034784,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007751727475757013,
      "loss": 4.9075,
      "step": 10385
    },
    {
      "epoch": 1.4223500410846344,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007751444284702786,
      "loss": 4.8544,
      "step": 10386
    },
    {
      "epoch": 1.4224869898657901,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007751161066703067,
      "loss": 4.8281,
      "step": 10387
    },
    {
      "epoch": 1.4226239386469461,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007750877821760202,
      "loss": 4.8918,
      "step": 10388
    },
    {
      "epoch": 1.422760887428102,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007750594549876536,
      "loss": 4.9053,
      "step": 10389
    },
    {
      "epoch": 1.4228978362092577,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007750311251054419,
      "loss": 4.8156,
      "step": 10390
    },
    {
      "epoch": 1.4230347849904135,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007750027925296197,
      "loss": 4.8351,
      "step": 10391
    },
    {
      "epoch": 1.4231717337715695,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007749744572604222,
      "loss": 4.8594,
      "step": 10392
    },
    {
      "epoch": 1.4233086825527252,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007749461192980838,
      "loss": 4.9415,
      "step": 10393
    },
    {
      "epoch": 1.4234456313338812,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007749177786428395,
      "loss": 4.8328,
      "step": 10394
    },
    {
      "epoch": 1.423582580115037,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000774889435294924,
      "loss": 4.8589,
      "step": 10395
    },
    {
      "epoch": 1.4237195288961928,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0007748610892545724,
      "loss": 4.7492,
      "step": 10396
    },
    {
      "epoch": 1.4238564776773486,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0007748327405220196,
      "loss": 4.8068,
      "step": 10397
    },
    {
      "epoch": 1.4239934264585046,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007748043890975006,
      "loss": 4.8659,
      "step": 10398
    },
    {
      "epoch": 1.4241303752396604,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00077477603498125,
      "loss": 4.8037,
      "step": 10399
    },
    {
      "epoch": 1.4242673240208163,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007747476781735031,
      "loss": 4.7802,
      "step": 10400
    },
    {
      "epoch": 1.4244042728019721,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007747193186744948,
      "loss": 4.854,
      "step": 10401
    },
    {
      "epoch": 1.424541221583128,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007746909564844602,
      "loss": 4.7781,
      "step": 10402
    },
    {
      "epoch": 1.4246781703642837,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007746625916036342,
      "loss": 4.9095,
      "step": 10403
    },
    {
      "epoch": 1.4248151191454397,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000774634224032252,
      "loss": 4.8693,
      "step": 10404
    },
    {
      "epoch": 1.4249520679265955,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007746058537705484,
      "loss": 4.8724,
      "step": 10405
    },
    {
      "epoch": 1.4250890167077512,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007745774808187591,
      "loss": 4.8468,
      "step": 10406
    },
    {
      "epoch": 1.4252259654889072,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007745491051771186,
      "loss": 4.872,
      "step": 10407
    },
    {
      "epoch": 1.425362914270063,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007745207268458624,
      "loss": 4.9309,
      "step": 10408
    },
    {
      "epoch": 1.4254998630512188,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007744923458252257,
      "loss": 4.8448,
      "step": 10409
    },
    {
      "epoch": 1.4256368118323746,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0007744639621154435,
      "loss": 4.9407,
      "step": 10410
    },
    {
      "epoch": 1.4257737606135306,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007744355757167511,
      "loss": 4.9063,
      "step": 10411
    },
    {
      "epoch": 1.4259107093946863,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007744071866293839,
      "loss": 4.76,
      "step": 10412
    },
    {
      "epoch": 1.4260476581758423,
      "grad_norm": 0.482421875,
      "learning_rate": 0.000774378794853577,
      "loss": 4.8064,
      "step": 10413
    },
    {
      "epoch": 1.4261846069569981,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007743504003895656,
      "loss": 4.8184,
      "step": 10414
    },
    {
      "epoch": 1.426321555738154,
      "grad_norm": 0.484375,
      "learning_rate": 0.0007743220032375854,
      "loss": 4.7972,
      "step": 10415
    },
    {
      "epoch": 1.4264585045193097,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007742936033978713,
      "loss": 4.8609,
      "step": 10416
    },
    {
      "epoch": 1.4265954533004657,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007742652008706589,
      "loss": 4.7718,
      "step": 10417
    },
    {
      "epoch": 1.4267324020816214,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007742367956561835,
      "loss": 4.8754,
      "step": 10418
    },
    {
      "epoch": 1.4268693508627774,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007742083877546805,
      "loss": 4.8992,
      "step": 10419
    },
    {
      "epoch": 1.4270062996439332,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007741799771663853,
      "loss": 4.7303,
      "step": 10420
    },
    {
      "epoch": 1.427143248425089,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007741515638915335,
      "loss": 4.9,
      "step": 10421
    },
    {
      "epoch": 1.4272801972062448,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007741231479303603,
      "loss": 4.8116,
      "step": 10422
    },
    {
      "epoch": 1.4274171459874008,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007740947292831014,
      "loss": 4.922,
      "step": 10423
    },
    {
      "epoch": 1.4275540947685565,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007740663079499922,
      "loss": 4.8522,
      "step": 10424
    },
    {
      "epoch": 1.4276910435497125,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007740378839312683,
      "loss": 4.8328,
      "step": 10425
    },
    {
      "epoch": 1.4278279923308683,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007740094572271654,
      "loss": 4.8229,
      "step": 10426
    },
    {
      "epoch": 1.427964941112024,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007739810278379187,
      "loss": 4.9151,
      "step": 10427
    },
    {
      "epoch": 1.4281018898931799,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007739525957637641,
      "loss": 4.8055,
      "step": 10428
    },
    {
      "epoch": 1.4282388386743359,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007739241610049373,
      "loss": 4.9606,
      "step": 10429
    },
    {
      "epoch": 1.4283757874554917,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007738957235616736,
      "loss": 4.8735,
      "step": 10430
    },
    {
      "epoch": 1.4285127362366474,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007738672834342089,
      "loss": 4.8909,
      "step": 10431
    },
    {
      "epoch": 1.4286496850178034,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000773838840622779,
      "loss": 4.8324,
      "step": 10432
    },
    {
      "epoch": 1.4287866337989592,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007738103951276192,
      "loss": 4.8345,
      "step": 10433
    },
    {
      "epoch": 1.428923582580115,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007737819469489657,
      "loss": 4.8288,
      "step": 10434
    },
    {
      "epoch": 1.4290605313612708,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007737534960870539,
      "loss": 4.9187,
      "step": 10435
    },
    {
      "epoch": 1.4291974801424268,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007737250425421198,
      "loss": 4.7947,
      "step": 10436
    },
    {
      "epoch": 1.4293344289235825,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007736965863143992,
      "loss": 4.8378,
      "step": 10437
    },
    {
      "epoch": 1.4294713777047385,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007736681274041278,
      "loss": 4.9166,
      "step": 10438
    },
    {
      "epoch": 1.4296083264858943,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007736396658115415,
      "loss": 4.8051,
      "step": 10439
    },
    {
      "epoch": 1.42974527526705,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007736112015368761,
      "loss": 4.868,
      "step": 10440
    },
    {
      "epoch": 1.4298822240482059,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007735827345803677,
      "loss": 4.8418,
      "step": 10441
    },
    {
      "epoch": 1.4300191728293619,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000773554264942252,
      "loss": 4.8684,
      "step": 10442
    },
    {
      "epoch": 1.4301561216105176,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000773525792622765,
      "loss": 4.8616,
      "step": 10443
    },
    {
      "epoch": 1.4302930703916736,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007734973176221426,
      "loss": 4.8434,
      "step": 10444
    },
    {
      "epoch": 1.4304300191728294,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000773468839940621,
      "loss": 4.8734,
      "step": 10445
    },
    {
      "epoch": 1.4305669679539852,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0007734403595784361,
      "loss": 4.8528,
      "step": 10446
    },
    {
      "epoch": 1.430703916735141,
      "grad_norm": 0.375,
      "learning_rate": 0.0007734118765358236,
      "loss": 4.9117,
      "step": 10447
    },
    {
      "epoch": 1.430840865516297,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00077338339081302,
      "loss": 4.8683,
      "step": 10448
    },
    {
      "epoch": 1.4309778142974527,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007733549024102611,
      "loss": 4.9142,
      "step": 10449
    },
    {
      "epoch": 1.4311147630786085,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007733264113277831,
      "loss": 4.9094,
      "step": 10450
    },
    {
      "epoch": 1.4312517118597645,
      "grad_norm": 0.7421875,
      "learning_rate": 0.000773297917565822,
      "loss": 4.8252,
      "step": 10451
    },
    {
      "epoch": 1.4313886606409203,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007732694211246142,
      "loss": 4.8813,
      "step": 10452
    },
    {
      "epoch": 1.431525609422076,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007732409220043955,
      "loss": 4.9023,
      "step": 10453
    },
    {
      "epoch": 1.431662558203232,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007732124202054024,
      "loss": 4.7757,
      "step": 10454
    },
    {
      "epoch": 1.4317995069843878,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0007731839157278708,
      "loss": 4.8476,
      "step": 10455
    },
    {
      "epoch": 1.4319364557655436,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007731554085720372,
      "loss": 4.836,
      "step": 10456
    },
    {
      "epoch": 1.4320734045466996,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007731268987381378,
      "loss": 4.8734,
      "step": 10457
    },
    {
      "epoch": 1.4322103533278554,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007730983862264087,
      "loss": 4.856,
      "step": 10458
    },
    {
      "epoch": 1.4323473021090112,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007730698710370862,
      "loss": 4.8756,
      "step": 10459
    },
    {
      "epoch": 1.432484250890167,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007730413531704068,
      "loss": 4.8209,
      "step": 10460
    },
    {
      "epoch": 1.432621199671323,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007730128326266067,
      "loss": 4.9155,
      "step": 10461
    },
    {
      "epoch": 1.4327581484524787,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007729843094059223,
      "loss": 4.8683,
      "step": 10462
    },
    {
      "epoch": 1.4328950972336347,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007729557835085899,
      "loss": 4.8818,
      "step": 10463
    },
    {
      "epoch": 1.4330320460147905,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007729272549348461,
      "loss": 4.7813,
      "step": 10464
    },
    {
      "epoch": 1.4331689947959463,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000772898723684927,
      "loss": 4.891,
      "step": 10465
    },
    {
      "epoch": 1.433305943577102,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007728701897590694,
      "loss": 4.849,
      "step": 10466
    },
    {
      "epoch": 1.433442892358258,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007728416531575095,
      "loss": 4.8459,
      "step": 10467
    },
    {
      "epoch": 1.4335798411394138,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007728131138804839,
      "loss": 4.8547,
      "step": 10468
    },
    {
      "epoch": 1.4337167899205698,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007727845719282293,
      "loss": 4.8339,
      "step": 10469
    },
    {
      "epoch": 1.4338537387017256,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007727560273009818,
      "loss": 4.8872,
      "step": 10470
    },
    {
      "epoch": 1.4339906874828814,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007727274799989782,
      "loss": 4.845,
      "step": 10471
    },
    {
      "epoch": 1.4341276362640372,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007726989300224551,
      "loss": 4.8606,
      "step": 10472
    },
    {
      "epoch": 1.4342645850451932,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007726703773716492,
      "loss": 4.8238,
      "step": 10473
    },
    {
      "epoch": 1.434401533826349,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007726418220467969,
      "loss": 4.825,
      "step": 10474
    },
    {
      "epoch": 1.4345384826075047,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007726132640481349,
      "loss": 4.9074,
      "step": 10475
    },
    {
      "epoch": 1.4346754313886607,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007725847033759,
      "loss": 4.8901,
      "step": 10476
    },
    {
      "epoch": 1.4348123801698165,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007725561400303288,
      "loss": 4.7892,
      "step": 10477
    },
    {
      "epoch": 1.4349493289509723,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0007725275740116579,
      "loss": 4.7658,
      "step": 10478
    },
    {
      "epoch": 1.435086277732128,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007724990053201242,
      "loss": 4.8732,
      "step": 10479
    },
    {
      "epoch": 1.435223226513284,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007724704339559644,
      "loss": 4.8263,
      "step": 10480
    },
    {
      "epoch": 1.4353601752944398,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007724418599194152,
      "loss": 4.8105,
      "step": 10481
    },
    {
      "epoch": 1.4354971240755958,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007724132832107136,
      "loss": 4.843,
      "step": 10482
    },
    {
      "epoch": 1.4356340728567516,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007723847038300963,
      "loss": 4.9789,
      "step": 10483
    },
    {
      "epoch": 1.4357710216379074,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007723561217778,
      "loss": 4.8384,
      "step": 10484
    },
    {
      "epoch": 1.4359079704190632,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007723275370540618,
      "loss": 4.8595,
      "step": 10485
    },
    {
      "epoch": 1.4360449192002191,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007722989496591184,
      "loss": 4.894,
      "step": 10486
    },
    {
      "epoch": 1.436181867981375,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007722703595932069,
      "loss": 4.8655,
      "step": 10487
    },
    {
      "epoch": 1.436318816762531,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007722417668565642,
      "loss": 4.8322,
      "step": 10488
    },
    {
      "epoch": 1.4364557655436867,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000772213171449427,
      "loss": 4.8961,
      "step": 10489
    },
    {
      "epoch": 1.4365927143248425,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007721845733720327,
      "loss": 4.8316,
      "step": 10490
    },
    {
      "epoch": 1.4367296631059983,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007721559726246178,
      "loss": 4.8776,
      "step": 10491
    },
    {
      "epoch": 1.4368666118871543,
      "grad_norm": 0.46875,
      "learning_rate": 0.0007721273692074198,
      "loss": 4.7561,
      "step": 10492
    },
    {
      "epoch": 1.43700356066831,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007720987631206754,
      "loss": 4.9617,
      "step": 10493
    },
    {
      "epoch": 1.437140509449466,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007720701543646219,
      "loss": 4.8722,
      "step": 10494
    },
    {
      "epoch": 1.4372774582306218,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007720415429394962,
      "loss": 4.8248,
      "step": 10495
    },
    {
      "epoch": 1.4374144070117776,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007720129288455355,
      "loss": 4.7829,
      "step": 10496
    },
    {
      "epoch": 1.4375513557929334,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007719843120829772,
      "loss": 4.7974,
      "step": 10497
    },
    {
      "epoch": 1.4376883045740894,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007719556926520577,
      "loss": 4.9141,
      "step": 10498
    },
    {
      "epoch": 1.4378252533552451,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000771927070553015,
      "loss": 4.8337,
      "step": 10499
    },
    {
      "epoch": 1.437962202136401,
      "grad_norm": 0.375,
      "learning_rate": 0.0007718984457860858,
      "loss": 4.7952,
      "step": 10500
    },
    {
      "epoch": 1.438099150917557,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007718698183515077,
      "loss": 4.8454,
      "step": 10501
    },
    {
      "epoch": 1.4382360996987127,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007718411882495175,
      "loss": 4.9664,
      "step": 10502
    },
    {
      "epoch": 1.4383730484798685,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007718125554803527,
      "loss": 4.7237,
      "step": 10503
    },
    {
      "epoch": 1.4385099972610242,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007717839200442506,
      "loss": 4.8163,
      "step": 10504
    },
    {
      "epoch": 1.4386469460421802,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007717552819414485,
      "loss": 4.7963,
      "step": 10505
    },
    {
      "epoch": 1.438783894823336,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007717266411721837,
      "loss": 4.8503,
      "step": 10506
    },
    {
      "epoch": 1.438920843604492,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007716979977366935,
      "loss": 4.9064,
      "step": 10507
    },
    {
      "epoch": 1.4390577923856478,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007716693516352154,
      "loss": 4.8827,
      "step": 10508
    },
    {
      "epoch": 1.4391947411668036,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007716407028679868,
      "loss": 4.9244,
      "step": 10509
    },
    {
      "epoch": 1.4393316899479593,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007716120514352449,
      "loss": 4.8174,
      "step": 10510
    },
    {
      "epoch": 1.4394686387291153,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007715833973372273,
      "loss": 4.8113,
      "step": 10511
    },
    {
      "epoch": 1.4396055875102711,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007715547405741716,
      "loss": 4.7146,
      "step": 10512
    },
    {
      "epoch": 1.4397425362914271,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007715260811463151,
      "loss": 4.7851,
      "step": 10513
    },
    {
      "epoch": 1.439879485072583,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007714974190538953,
      "loss": 4.858,
      "step": 10514
    },
    {
      "epoch": 1.4400164338537387,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007714687542971498,
      "loss": 4.7158,
      "step": 10515
    },
    {
      "epoch": 1.4401533826348945,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007714400868763162,
      "loss": 4.8265,
      "step": 10516
    },
    {
      "epoch": 1.4402903314160505,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007714114167916319,
      "loss": 4.847,
      "step": 10517
    },
    {
      "epoch": 1.4404272801972062,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007713827440433347,
      "loss": 4.883,
      "step": 10518
    },
    {
      "epoch": 1.4405642289783622,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000771354068631662,
      "loss": 4.8631,
      "step": 10519
    },
    {
      "epoch": 1.440701177759518,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007713253905568516,
      "loss": 4.8686,
      "step": 10520
    },
    {
      "epoch": 1.4408381265406738,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007712967098191411,
      "loss": 4.9094,
      "step": 10521
    },
    {
      "epoch": 1.4409750753218296,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007712680264187683,
      "loss": 4.8036,
      "step": 10522
    },
    {
      "epoch": 1.4411120241029856,
      "grad_norm": 0.375,
      "learning_rate": 0.0007712393403559708,
      "loss": 4.9736,
      "step": 10523
    },
    {
      "epoch": 1.4412489728841413,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007712106516309862,
      "loss": 4.7646,
      "step": 10524
    },
    {
      "epoch": 1.441385921665297,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007711819602440525,
      "loss": 4.7918,
      "step": 10525
    },
    {
      "epoch": 1.441522870446453,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007711532661954073,
      "loss": 4.8701,
      "step": 10526
    },
    {
      "epoch": 1.4416598192276089,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007711245694852886,
      "loss": 4.8173,
      "step": 10527
    },
    {
      "epoch": 1.4417967680087647,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007710958701139339,
      "loss": 4.8462,
      "step": 10528
    },
    {
      "epoch": 1.4419337167899204,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007710671680815813,
      "loss": 4.8902,
      "step": 10529
    },
    {
      "epoch": 1.4420706655710764,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007710384633884686,
      "loss": 4.8955,
      "step": 10530
    },
    {
      "epoch": 1.4422076143522322,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007710097560348336,
      "loss": 4.8607,
      "step": 10531
    },
    {
      "epoch": 1.4423445631333882,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007709810460209142,
      "loss": 4.8152,
      "step": 10532
    },
    {
      "epoch": 1.442481511914544,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007709523333469486,
      "loss": 4.772,
      "step": 10533
    },
    {
      "epoch": 1.4426184606956998,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007709236180131743,
      "loss": 4.895,
      "step": 10534
    },
    {
      "epoch": 1.4427554094768555,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007708949000198295,
      "loss": 4.8441,
      "step": 10535
    },
    {
      "epoch": 1.4428923582580115,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007708661793671524,
      "loss": 4.8753,
      "step": 10536
    },
    {
      "epoch": 1.4430293070391673,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007708374560553805,
      "loss": 4.7921,
      "step": 10537
    },
    {
      "epoch": 1.4431662558203233,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007708087300847525,
      "loss": 4.8736,
      "step": 10538
    },
    {
      "epoch": 1.443303204601479,
      "grad_norm": 0.375,
      "learning_rate": 0.000770780001455506,
      "loss": 4.8746,
      "step": 10539
    },
    {
      "epoch": 1.4434401533826349,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007707512701678791,
      "loss": 4.9001,
      "step": 10540
    },
    {
      "epoch": 1.4435771021637906,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007707225362221101,
      "loss": 4.8225,
      "step": 10541
    },
    {
      "epoch": 1.4437140509449466,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007706937996184369,
      "loss": 4.8325,
      "step": 10542
    },
    {
      "epoch": 1.4438509997261024,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007706650603570979,
      "loss": 4.811,
      "step": 10543
    },
    {
      "epoch": 1.4439879485072584,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007706363184383311,
      "loss": 4.876,
      "step": 10544
    },
    {
      "epoch": 1.4441248972884142,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007706075738623746,
      "loss": 4.899,
      "step": 10545
    },
    {
      "epoch": 1.44426184606957,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007705788266294669,
      "loss": 4.8216,
      "step": 10546
    },
    {
      "epoch": 1.4443987948507258,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000770550076739846,
      "loss": 4.8671,
      "step": 10547
    },
    {
      "epoch": 1.4445357436318818,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007705213241937502,
      "loss": 4.8493,
      "step": 10548
    },
    {
      "epoch": 1.4446726924130375,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007704925689914179,
      "loss": 4.8194,
      "step": 10549
    },
    {
      "epoch": 1.4448096411941933,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007704638111330872,
      "loss": 4.8192,
      "step": 10550
    },
    {
      "epoch": 1.4449465899753493,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007704350506189967,
      "loss": 4.8616,
      "step": 10551
    },
    {
      "epoch": 1.445083538756505,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007704062874493843,
      "loss": 4.9364,
      "step": 10552
    },
    {
      "epoch": 1.4452204875376609,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007703775216244889,
      "loss": 4.813,
      "step": 10553
    },
    {
      "epoch": 1.4453574363188166,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007703487531445486,
      "loss": 4.8681,
      "step": 10554
    },
    {
      "epoch": 1.4454943850999726,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007703199820098018,
      "loss": 4.9274,
      "step": 10555
    },
    {
      "epoch": 1.4456313338811284,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000770291208220487,
      "loss": 4.9747,
      "step": 10556
    },
    {
      "epoch": 1.4457682826622844,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007702624317768427,
      "loss": 4.8459,
      "step": 10557
    },
    {
      "epoch": 1.4459052314434402,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007702336526791071,
      "loss": 4.8358,
      "step": 10558
    },
    {
      "epoch": 1.446042180224596,
      "grad_norm": 0.375,
      "learning_rate": 0.000770204870927519,
      "loss": 4.8006,
      "step": 10559
    },
    {
      "epoch": 1.4461791290057517,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007701760865223169,
      "loss": 4.8465,
      "step": 10560
    },
    {
      "epoch": 1.4463160777869077,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007701472994637393,
      "loss": 4.8952,
      "step": 10561
    },
    {
      "epoch": 1.4464530265680635,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007701185097520247,
      "loss": 4.888,
      "step": 10562
    },
    {
      "epoch": 1.4465899753492195,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007700897173874117,
      "loss": 4.8348,
      "step": 10563
    },
    {
      "epoch": 1.4467269241303753,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000770060922370139,
      "loss": 4.8103,
      "step": 10564
    },
    {
      "epoch": 1.446863872911531,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007700321247004452,
      "loss": 4.925,
      "step": 10565
    },
    {
      "epoch": 1.4470008216926868,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007700033243785688,
      "loss": 4.8793,
      "step": 10566
    },
    {
      "epoch": 1.4471377704738428,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007699745214047487,
      "loss": 4.874,
      "step": 10567
    },
    {
      "epoch": 1.4472747192549986,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007699457157792234,
      "loss": 4.8683,
      "step": 10568
    },
    {
      "epoch": 1.4474116680361544,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007699169075022318,
      "loss": 4.8708,
      "step": 10569
    },
    {
      "epoch": 1.4475486168173104,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007698880965740125,
      "loss": 4.7811,
      "step": 10570
    },
    {
      "epoch": 1.4476855655984662,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007698592829948043,
      "loss": 4.8351,
      "step": 10571
    },
    {
      "epoch": 1.447822514379622,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007698304667648461,
      "loss": 4.8557,
      "step": 10572
    },
    {
      "epoch": 1.447959463160778,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007698016478843765,
      "loss": 4.7922,
      "step": 10573
    },
    {
      "epoch": 1.4480964119419337,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007697728263536346,
      "loss": 4.8254,
      "step": 10574
    },
    {
      "epoch": 1.4482333607230895,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007697440021728591,
      "loss": 4.8461,
      "step": 10575
    },
    {
      "epoch": 1.4483703095042455,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007697151753422887,
      "loss": 4.8686,
      "step": 10576
    },
    {
      "epoch": 1.4485072582854013,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007696863458621625,
      "loss": 4.8943,
      "step": 10577
    },
    {
      "epoch": 1.448644207066557,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007696575137327193,
      "loss": 4.9045,
      "step": 10578
    },
    {
      "epoch": 1.4487811558477128,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007696286789541983,
      "loss": 4.855,
      "step": 10579
    },
    {
      "epoch": 1.4489181046288688,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007695998415268381,
      "loss": 4.8557,
      "step": 10580
    },
    {
      "epoch": 1.4490550534100246,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007695710014508779,
      "loss": 4.8514,
      "step": 10581
    },
    {
      "epoch": 1.4491920021911806,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007695421587265568,
      "loss": 4.9232,
      "step": 10582
    },
    {
      "epoch": 1.4493289509723364,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007695133133541135,
      "loss": 4.8184,
      "step": 10583
    },
    {
      "epoch": 1.4494658997534922,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007694844653337873,
      "loss": 4.8115,
      "step": 10584
    },
    {
      "epoch": 1.449602848534648,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007694556146658172,
      "loss": 4.9297,
      "step": 10585
    },
    {
      "epoch": 1.449739797315804,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007694267613504424,
      "loss": 4.823,
      "step": 10586
    },
    {
      "epoch": 1.4498767460969597,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007693979053879017,
      "loss": 4.8834,
      "step": 10587
    },
    {
      "epoch": 1.4500136948781157,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007693690467784346,
      "loss": 4.9134,
      "step": 10588
    },
    {
      "epoch": 1.4501506436592715,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007693401855222801,
      "loss": 4.8651,
      "step": 10589
    },
    {
      "epoch": 1.4502875924404273,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007693113216196775,
      "loss": 4.8734,
      "step": 10590
    },
    {
      "epoch": 1.450424541221583,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007692824550708657,
      "loss": 4.8712,
      "step": 10591
    },
    {
      "epoch": 1.450561490002739,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007692535858760842,
      "loss": 4.8751,
      "step": 10592
    },
    {
      "epoch": 1.4506984387838948,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007692247140355721,
      "loss": 4.8466,
      "step": 10593
    },
    {
      "epoch": 1.4508353875650506,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007691958395495688,
      "loss": 4.854,
      "step": 10594
    },
    {
      "epoch": 1.4509723363462066,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007691669624183134,
      "loss": 4.874,
      "step": 10595
    },
    {
      "epoch": 1.4511092851273624,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007691380826420455,
      "loss": 4.8929,
      "step": 10596
    },
    {
      "epoch": 1.4512462339085181,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007691092002210042,
      "loss": 4.8186,
      "step": 10597
    },
    {
      "epoch": 1.4513831826896741,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000769080315155429,
      "loss": 4.9262,
      "step": 10598
    },
    {
      "epoch": 1.45152013147083,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000769051427445559,
      "loss": 4.8449,
      "step": 10599
    },
    {
      "epoch": 1.4516570802519857,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007690225370916339,
      "loss": 4.8369,
      "step": 10600
    },
    {
      "epoch": 1.4517940290331417,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000768993644093893,
      "loss": 4.8125,
      "step": 10601
    },
    {
      "epoch": 1.4519309778142975,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007689647484525758,
      "loss": 4.7693,
      "step": 10602
    },
    {
      "epoch": 1.4520679265954533,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007689358501679216,
      "loss": 4.7463,
      "step": 10603
    },
    {
      "epoch": 1.452204875376609,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007689069492401702,
      "loss": 4.9063,
      "step": 10604
    },
    {
      "epoch": 1.452341824157765,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007688780456695608,
      "loss": 4.9346,
      "step": 10605
    },
    {
      "epoch": 1.4524787729389208,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007688491394563329,
      "loss": 4.9357,
      "step": 10606
    },
    {
      "epoch": 1.4526157217200768,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007688202306007265,
      "loss": 4.8319,
      "step": 10607
    },
    {
      "epoch": 1.4527526705012326,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007687913191029808,
      "loss": 4.9268,
      "step": 10608
    },
    {
      "epoch": 1.4528896192823884,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007687624049633354,
      "loss": 4.7934,
      "step": 10609
    },
    {
      "epoch": 1.4530265680635441,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007687334881820299,
      "loss": 4.827,
      "step": 10610
    },
    {
      "epoch": 1.4531635168447001,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007687045687593042,
      "loss": 4.7981,
      "step": 10611
    },
    {
      "epoch": 1.453300465625856,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007686756466953978,
      "loss": 4.8764,
      "step": 10612
    },
    {
      "epoch": 1.453437414407012,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007686467219905504,
      "loss": 4.8964,
      "step": 10613
    },
    {
      "epoch": 1.4535743631881677,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007686177946450016,
      "loss": 4.8271,
      "step": 10614
    },
    {
      "epoch": 1.4537113119693235,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007685888646589912,
      "loss": 4.7801,
      "step": 10615
    },
    {
      "epoch": 1.4538482607504792,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000768559932032759,
      "loss": 4.8067,
      "step": 10616
    },
    {
      "epoch": 1.4539852095316352,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007685309967665448,
      "loss": 4.8625,
      "step": 10617
    },
    {
      "epoch": 1.454122158312791,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007685020588605882,
      "loss": 4.8255,
      "step": 10618
    },
    {
      "epoch": 1.4542591070939468,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007684731183151292,
      "loss": 4.8547,
      "step": 10619
    },
    {
      "epoch": 1.4543960558751028,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007684441751304075,
      "loss": 4.8812,
      "step": 10620
    },
    {
      "epoch": 1.4545330046562586,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007684152293066631,
      "loss": 4.8567,
      "step": 10621
    },
    {
      "epoch": 1.4546699534374143,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000768386280844136,
      "loss": 4.8874,
      "step": 10622
    },
    {
      "epoch": 1.4548069022185701,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007683573297430658,
      "loss": 4.7924,
      "step": 10623
    },
    {
      "epoch": 1.4549438509997261,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007683283760036924,
      "loss": 4.8351,
      "step": 10624
    },
    {
      "epoch": 1.455080799780882,
      "grad_norm": 0.359375,
      "learning_rate": 0.000768299419626256,
      "loss": 4.8535,
      "step": 10625
    },
    {
      "epoch": 1.455217748562038,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007682704606109964,
      "loss": 4.8536,
      "step": 10626
    },
    {
      "epoch": 1.4553546973431937,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007682414989581537,
      "loss": 4.8612,
      "step": 10627
    },
    {
      "epoch": 1.4554916461243494,
      "grad_norm": 0.359375,
      "learning_rate": 0.000768212534667968,
      "loss": 4.8615,
      "step": 10628
    },
    {
      "epoch": 1.4556285949055052,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000768183567740679,
      "loss": 4.8961,
      "step": 10629
    },
    {
      "epoch": 1.4557655436866612,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000768154598176527,
      "loss": 4.8664,
      "step": 10630
    },
    {
      "epoch": 1.455902492467817,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007681256259757522,
      "loss": 4.9142,
      "step": 10631
    },
    {
      "epoch": 1.456039441248973,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007680966511385944,
      "loss": 4.8233,
      "step": 10632
    },
    {
      "epoch": 1.4561763900301288,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007680676736652939,
      "loss": 4.8035,
      "step": 10633
    },
    {
      "epoch": 1.4563133388112846,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007680386935560908,
      "loss": 4.8125,
      "step": 10634
    },
    {
      "epoch": 1.4564502875924403,
      "grad_norm": 0.390625,
      "learning_rate": 0.000768009710811225,
      "loss": 4.8631,
      "step": 10635
    },
    {
      "epoch": 1.4565872363735963,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007679807254309374,
      "loss": 4.8321,
      "step": 10636
    },
    {
      "epoch": 1.456724185154752,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007679517374154673,
      "loss": 4.8487,
      "step": 10637
    },
    {
      "epoch": 1.456861133935908,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007679227467650557,
      "loss": 4.8558,
      "step": 10638
    },
    {
      "epoch": 1.4569980827170639,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007678937534799425,
      "loss": 4.8324,
      "step": 10639
    },
    {
      "epoch": 1.4571350314982197,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007678647575603679,
      "loss": 4.8198,
      "step": 10640
    },
    {
      "epoch": 1.4572719802793754,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007678357590065725,
      "loss": 4.7922,
      "step": 10641
    },
    {
      "epoch": 1.4574089290605314,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007678067578187962,
      "loss": 4.8549,
      "step": 10642
    },
    {
      "epoch": 1.4575458778416872,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007677777539972796,
      "loss": 4.8484,
      "step": 10643
    },
    {
      "epoch": 1.457682826622843,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007677487475422632,
      "loss": 4.8593,
      "step": 10644
    },
    {
      "epoch": 1.457819775403999,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000767719738453987,
      "loss": 4.9023,
      "step": 10645
    },
    {
      "epoch": 1.4579567241851548,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007676907267326918,
      "loss": 4.8552,
      "step": 10646
    },
    {
      "epoch": 1.4580936729663105,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007676617123786178,
      "loss": 4.9043,
      "step": 10647
    },
    {
      "epoch": 1.4582306217474663,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007676326953920055,
      "loss": 4.835,
      "step": 10648
    },
    {
      "epoch": 1.4583675705286223,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007676036757730954,
      "loss": 4.9209,
      "step": 10649
    },
    {
      "epoch": 1.458504519309778,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007675746535221278,
      "loss": 4.8716,
      "step": 10650
    },
    {
      "epoch": 1.458641468090934,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007675456286393435,
      "loss": 4.7617,
      "step": 10651
    },
    {
      "epoch": 1.4587784168720899,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007675166011249829,
      "loss": 4.9056,
      "step": 10652
    },
    {
      "epoch": 1.4589153656532456,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007674875709792866,
      "loss": 4.8529,
      "step": 10653
    },
    {
      "epoch": 1.4590523144344014,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007674585382024951,
      "loss": 4.7779,
      "step": 10654
    },
    {
      "epoch": 1.4591892632155574,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000767429502794849,
      "loss": 4.8753,
      "step": 10655
    },
    {
      "epoch": 1.4593262119967132,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000767400464756589,
      "loss": 4.8118,
      "step": 10656
    },
    {
      "epoch": 1.4594631607778692,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007673714240879557,
      "loss": 4.8909,
      "step": 10657
    },
    {
      "epoch": 1.459600109559025,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00076734238078919,
      "loss": 4.7682,
      "step": 10658
    },
    {
      "epoch": 1.4597370583401807,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007673133348605321,
      "loss": 4.8575,
      "step": 10659
    },
    {
      "epoch": 1.4598740071213365,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000767284286302223,
      "loss": 4.8719,
      "step": 10660
    },
    {
      "epoch": 1.4600109559024925,
      "grad_norm": 0.375,
      "learning_rate": 0.0007672552351145035,
      "loss": 4.8065,
      "step": 10661
    },
    {
      "epoch": 1.4601479046836483,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007672261812976143,
      "loss": 4.8546,
      "step": 10662
    },
    {
      "epoch": 1.4602848534648043,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007671971248517962,
      "loss": 4.8818,
      "step": 10663
    },
    {
      "epoch": 1.46042180224596,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007671680657772897,
      "loss": 4.8017,
      "step": 10664
    },
    {
      "epoch": 1.4605587510271159,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000767139004074336,
      "loss": 4.9038,
      "step": 10665
    },
    {
      "epoch": 1.4606956998082716,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007671099397431757,
      "loss": 4.8316,
      "step": 10666
    },
    {
      "epoch": 1.4608326485894276,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007670808727840501,
      "loss": 4.9236,
      "step": 10667
    },
    {
      "epoch": 1.4609695973705834,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007670518031971994,
      "loss": 4.9303,
      "step": 10668
    },
    {
      "epoch": 1.4611065461517392,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000767022730982865,
      "loss": 4.8964,
      "step": 10669
    },
    {
      "epoch": 1.4612434949328952,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007669936561412877,
      "loss": 4.8047,
      "step": 10670
    },
    {
      "epoch": 1.461380443714051,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007669645786727082,
      "loss": 4.9072,
      "step": 10671
    },
    {
      "epoch": 1.4615173924952067,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007669354985773681,
      "loss": 4.7679,
      "step": 10672
    },
    {
      "epoch": 1.4616543412763625,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007669064158555078,
      "loss": 4.8851,
      "step": 10673
    },
    {
      "epoch": 1.4617912900575185,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007668773305073686,
      "loss": 4.8543,
      "step": 10674
    },
    {
      "epoch": 1.4619282388386743,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007668482425331913,
      "loss": 4.904,
      "step": 10675
    },
    {
      "epoch": 1.4620651876198303,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007668191519332172,
      "loss": 4.8637,
      "step": 10676
    },
    {
      "epoch": 1.462202136400986,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007667900587076873,
      "loss": 4.8403,
      "step": 10677
    },
    {
      "epoch": 1.4623390851821418,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007667609628568427,
      "loss": 4.8381,
      "step": 10678
    },
    {
      "epoch": 1.4624760339632976,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007667318643809245,
      "loss": 4.9062,
      "step": 10679
    },
    {
      "epoch": 1.4626129827444536,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007667027632801738,
      "loss": 4.8923,
      "step": 10680
    },
    {
      "epoch": 1.4627499315256094,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000766673659554832,
      "loss": 4.7923,
      "step": 10681
    },
    {
      "epoch": 1.4628868803067654,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007666445532051399,
      "loss": 4.9242,
      "step": 10682
    },
    {
      "epoch": 1.4630238290879212,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007666154442313391,
      "loss": 4.8052,
      "step": 10683
    },
    {
      "epoch": 1.463160777869077,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007665863326336707,
      "loss": 4.8239,
      "step": 10684
    },
    {
      "epoch": 1.4632977266502327,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007665572184123758,
      "loss": 4.8594,
      "step": 10685
    },
    {
      "epoch": 1.4634346754313887,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007665281015676959,
      "loss": 4.8466,
      "step": 10686
    },
    {
      "epoch": 1.4635716242125445,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0007664989820998722,
      "loss": 4.8289,
      "step": 10687
    },
    {
      "epoch": 1.4637085729937003,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000766469860009146,
      "loss": 4.7368,
      "step": 10688
    },
    {
      "epoch": 1.4638455217748563,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007664407352957586,
      "loss": 4.9168,
      "step": 10689
    },
    {
      "epoch": 1.463982470556012,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007664116079599515,
      "loss": 4.8481,
      "step": 10690
    },
    {
      "epoch": 1.4641194193371678,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007663824780019659,
      "loss": 4.8508,
      "step": 10691
    },
    {
      "epoch": 1.4642563681183238,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007663533454220434,
      "loss": 4.8452,
      "step": 10692
    },
    {
      "epoch": 1.4643933168994796,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007663242102204254,
      "loss": 4.8851,
      "step": 10693
    },
    {
      "epoch": 1.4645302656806354,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007662950723973531,
      "loss": 4.8204,
      "step": 10694
    },
    {
      "epoch": 1.4646672144617914,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007662659319530682,
      "loss": 4.8442,
      "step": 10695
    },
    {
      "epoch": 1.4648041632429472,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007662367888878124,
      "loss": 4.8579,
      "step": 10696
    },
    {
      "epoch": 1.464941112024103,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007662076432018267,
      "loss": 4.9113,
      "step": 10697
    },
    {
      "epoch": 1.4650780608052587,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007661784948953531,
      "loss": 4.8611,
      "step": 10698
    },
    {
      "epoch": 1.4652150095864147,
      "grad_norm": 0.40625,
      "learning_rate": 0.000766149343968633,
      "loss": 4.8339,
      "step": 10699
    },
    {
      "epoch": 1.4653519583675705,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007661201904219079,
      "loss": 4.8642,
      "step": 10700
    },
    {
      "epoch": 1.4654889071487265,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007660910342554194,
      "loss": 4.8052,
      "step": 10701
    },
    {
      "epoch": 1.4656258559298823,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000766061875469409,
      "loss": 4.8552,
      "step": 10702
    },
    {
      "epoch": 1.465762804711038,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007660327140641189,
      "loss": 4.841,
      "step": 10703
    },
    {
      "epoch": 1.4658997534921938,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007660035500397901,
      "loss": 4.8249,
      "step": 10704
    },
    {
      "epoch": 1.4660367022733498,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007659743833966647,
      "loss": 4.8675,
      "step": 10705
    },
    {
      "epoch": 1.4661736510545056,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007659452141349842,
      "loss": 4.8303,
      "step": 10706
    },
    {
      "epoch": 1.4663105998356616,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007659160422549906,
      "loss": 4.8658,
      "step": 10707
    },
    {
      "epoch": 1.4664475486168174,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007658868677569254,
      "loss": 4.8265,
      "step": 10708
    },
    {
      "epoch": 1.4665844973979731,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007658576906410305,
      "loss": 4.8422,
      "step": 10709
    },
    {
      "epoch": 1.466721446179129,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007658285109075476,
      "loss": 4.8164,
      "step": 10710
    },
    {
      "epoch": 1.466858394960285,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007657993285567185,
      "loss": 4.8065,
      "step": 10711
    },
    {
      "epoch": 1.4669953437414407,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007657701435887852,
      "loss": 4.8414,
      "step": 10712
    },
    {
      "epoch": 1.4671322925225965,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007657409560039894,
      "loss": 4.8631,
      "step": 10713
    },
    {
      "epoch": 1.4672692413037525,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000765711765802573,
      "loss": 4.8179,
      "step": 10714
    },
    {
      "epoch": 1.4674061900849082,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007656825729847782,
      "loss": 4.77,
      "step": 10715
    },
    {
      "epoch": 1.467543138866064,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007656533775508465,
      "loss": 4.9137,
      "step": 10716
    },
    {
      "epoch": 1.46768008764722,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007656241795010202,
      "loss": 4.8988,
      "step": 10717
    },
    {
      "epoch": 1.4678170364283758,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000765594978835541,
      "loss": 4.8591,
      "step": 10718
    },
    {
      "epoch": 1.4679539852095316,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000765565775554651,
      "loss": 4.8484,
      "step": 10719
    },
    {
      "epoch": 1.4680909339906876,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007655365696585923,
      "loss": 4.9035,
      "step": 10720
    },
    {
      "epoch": 1.4682278827718434,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007655073611476068,
      "loss": 4.7971,
      "step": 10721
    },
    {
      "epoch": 1.4683648315529991,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007654781500219367,
      "loss": 4.873,
      "step": 10722
    },
    {
      "epoch": 1.468501780334155,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007654489362818238,
      "loss": 4.8395,
      "step": 10723
    },
    {
      "epoch": 1.468638729115311,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007654197199275107,
      "loss": 4.859,
      "step": 10724
    },
    {
      "epoch": 1.4687756778964667,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0007653905009592391,
      "loss": 4.8058,
      "step": 10725
    },
    {
      "epoch": 1.4689126266776227,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007653612793772511,
      "loss": 4.8734,
      "step": 10726
    },
    {
      "epoch": 1.4690495754587785,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007653320551817891,
      "loss": 4.9088,
      "step": 10727
    },
    {
      "epoch": 1.4691865242399342,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007653028283730953,
      "loss": 4.9411,
      "step": 10728
    },
    {
      "epoch": 1.46932347302109,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007652735989514118,
      "loss": 4.89,
      "step": 10729
    },
    {
      "epoch": 1.469460421802246,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007652443669169808,
      "loss": 4.8186,
      "step": 10730
    },
    {
      "epoch": 1.4695973705834018,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007652151322700448,
      "loss": 4.9116,
      "step": 10731
    },
    {
      "epoch": 1.4697343193645578,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007651858950108458,
      "loss": 4.8956,
      "step": 10732
    },
    {
      "epoch": 1.4698712681457136,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000765156655139626,
      "loss": 4.7684,
      "step": 10733
    },
    {
      "epoch": 1.4700082169268693,
      "grad_norm": 0.375,
      "learning_rate": 0.0007651274126566282,
      "loss": 4.8219,
      "step": 10734
    },
    {
      "epoch": 1.4701451657080251,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007650981675620942,
      "loss": 4.8762,
      "step": 10735
    },
    {
      "epoch": 1.4702821144891811,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0007650689198562666,
      "loss": 4.7939,
      "step": 10736
    },
    {
      "epoch": 1.470419063270337,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000765039669539388,
      "loss": 4.853,
      "step": 10737
    },
    {
      "epoch": 1.4705560120514927,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0007650104166117004,
      "loss": 4.8556,
      "step": 10738
    },
    {
      "epoch": 1.4706929608326487,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007649811610734465,
      "loss": 4.8437,
      "step": 10739
    },
    {
      "epoch": 1.4708299096138044,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007649519029248686,
      "loss": 4.8089,
      "step": 10740
    },
    {
      "epoch": 1.4709668583949602,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007649226421662093,
      "loss": 4.9555,
      "step": 10741
    },
    {
      "epoch": 1.471103807176116,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000764893378797711,
      "loss": 4.8297,
      "step": 10742
    },
    {
      "epoch": 1.471240755957272,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007648641128196163,
      "loss": 4.8691,
      "step": 10743
    },
    {
      "epoch": 1.4713777047384278,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007648348442321675,
      "loss": 4.7623,
      "step": 10744
    },
    {
      "epoch": 1.4715146535195838,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007648055730356076,
      "loss": 4.8261,
      "step": 10745
    },
    {
      "epoch": 1.4716516023007395,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007647762992301786,
      "loss": 4.9266,
      "step": 10746
    },
    {
      "epoch": 1.4717885510818953,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007647470228161234,
      "loss": 4.9024,
      "step": 10747
    },
    {
      "epoch": 1.471925499863051,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007647177437936849,
      "loss": 4.8779,
      "step": 10748
    },
    {
      "epoch": 1.472062448644207,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007646884621631052,
      "loss": 4.8828,
      "step": 10749
    },
    {
      "epoch": 1.4721993974253629,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007646591779246275,
      "loss": 4.9241,
      "step": 10750
    },
    {
      "epoch": 1.4723363462065189,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007646298910784939,
      "loss": 4.7767,
      "step": 10751
    },
    {
      "epoch": 1.4724732949876747,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007646006016249476,
      "loss": 4.9455,
      "step": 10752
    },
    {
      "epoch": 1.4726102437688304,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000764571309564231,
      "loss": 4.9107,
      "step": 10753
    },
    {
      "epoch": 1.4727471925499862,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007645420148965872,
      "loss": 4.7706,
      "step": 10754
    },
    {
      "epoch": 1.4728841413311422,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007645127176222587,
      "loss": 4.8839,
      "step": 10755
    },
    {
      "epoch": 1.473021090112298,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007644834177414884,
      "loss": 4.8177,
      "step": 10756
    },
    {
      "epoch": 1.473158038893454,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007644541152545191,
      "loss": 4.7874,
      "step": 10757
    },
    {
      "epoch": 1.4732949876746098,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007644248101615935,
      "loss": 4.8221,
      "step": 10758
    },
    {
      "epoch": 1.4734319364557655,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007643955024629546,
      "loss": 4.7853,
      "step": 10759
    },
    {
      "epoch": 1.4735688852369213,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007643661921588452,
      "loss": 4.8176,
      "step": 10760
    },
    {
      "epoch": 1.4737058340180773,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007643368792495084,
      "loss": 4.8532,
      "step": 10761
    },
    {
      "epoch": 1.473842782799233,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007643075637351869,
      "loss": 4.8728,
      "step": 10762
    },
    {
      "epoch": 1.4739797315803889,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007642782456161238,
      "loss": 4.8148,
      "step": 10763
    },
    {
      "epoch": 1.4741166803615449,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007642489248925619,
      "loss": 4.7992,
      "step": 10764
    },
    {
      "epoch": 1.4742536291427006,
      "grad_norm": 0.375,
      "learning_rate": 0.0007642196015647442,
      "loss": 4.8052,
      "step": 10765
    },
    {
      "epoch": 1.4743905779238564,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000764190275632914,
      "loss": 4.7924,
      "step": 10766
    },
    {
      "epoch": 1.4745275267050122,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007641609470973139,
      "loss": 4.8499,
      "step": 10767
    },
    {
      "epoch": 1.4746644754861682,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007641316159581874,
      "loss": 4.7942,
      "step": 10768
    },
    {
      "epoch": 1.474801424267324,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007641022822157772,
      "loss": 4.9032,
      "step": 10769
    },
    {
      "epoch": 1.47493837304848,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007640729458703266,
      "loss": 4.7957,
      "step": 10770
    },
    {
      "epoch": 1.4750753218296357,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007640436069220786,
      "loss": 4.7534,
      "step": 10771
    },
    {
      "epoch": 1.4752122706107915,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007640142653712765,
      "loss": 4.7928,
      "step": 10772
    },
    {
      "epoch": 1.4753492193919473,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007639849212181633,
      "loss": 4.874,
      "step": 10773
    },
    {
      "epoch": 1.4754861681731033,
      "grad_norm": 0.375,
      "learning_rate": 0.0007639555744629821,
      "loss": 4.796,
      "step": 10774
    },
    {
      "epoch": 1.475623116954259,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007639262251059765,
      "loss": 4.887,
      "step": 10775
    },
    {
      "epoch": 1.475760065735415,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007638968731473893,
      "loss": 4.8393,
      "step": 10776
    },
    {
      "epoch": 1.4758970145165708,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007638675185874639,
      "loss": 4.7922,
      "step": 10777
    },
    {
      "epoch": 1.4760339632977266,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007638381614264437,
      "loss": 4.9825,
      "step": 10778
    },
    {
      "epoch": 1.4761709120788824,
      "grad_norm": 0.375,
      "learning_rate": 0.0007638088016645718,
      "loss": 4.8368,
      "step": 10779
    },
    {
      "epoch": 1.4763078608600384,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007637794393020916,
      "loss": 4.9296,
      "step": 10780
    },
    {
      "epoch": 1.4764448096411942,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0007637500743392465,
      "loss": 4.821,
      "step": 10781
    },
    {
      "epoch": 1.4765817584223502,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007637207067762796,
      "loss": 4.8761,
      "step": 10782
    },
    {
      "epoch": 1.476718707203506,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0007636913366134344,
      "loss": 4.8521,
      "step": 10783
    },
    {
      "epoch": 1.4768556559846617,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007636619638509546,
      "loss": 4.8107,
      "step": 10784
    },
    {
      "epoch": 1.4769926047658175,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007636325884890832,
      "loss": 4.8734,
      "step": 10785
    },
    {
      "epoch": 1.4771295535469735,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007636032105280638,
      "loss": 4.7616,
      "step": 10786
    },
    {
      "epoch": 1.4772665023281293,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007635738299681399,
      "loss": 4.8698,
      "step": 10787
    },
    {
      "epoch": 1.477403451109285,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007635444468095549,
      "loss": 4.8415,
      "step": 10788
    },
    {
      "epoch": 1.477540399890441,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007635150610525524,
      "loss": 4.7673,
      "step": 10789
    },
    {
      "epoch": 1.4776773486715968,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007634856726973759,
      "loss": 4.8023,
      "step": 10790
    },
    {
      "epoch": 1.4778142974527526,
      "grad_norm": 0.375,
      "learning_rate": 0.000763456281744269,
      "loss": 4.8411,
      "step": 10791
    },
    {
      "epoch": 1.4779512462339084,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000763426888193475,
      "loss": 4.8316,
      "step": 10792
    },
    {
      "epoch": 1.4780881950150644,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007633974920452379,
      "loss": 4.8496,
      "step": 10793
    },
    {
      "epoch": 1.4782251437962202,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000763368093299801,
      "loss": 4.7399,
      "step": 10794
    },
    {
      "epoch": 1.4783620925773762,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007633386919574082,
      "loss": 4.8178,
      "step": 10795
    },
    {
      "epoch": 1.478499041358532,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007633092880183027,
      "loss": 4.8349,
      "step": 10796
    },
    {
      "epoch": 1.4786359901396877,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007632798814827286,
      "loss": 4.8103,
      "step": 10797
    },
    {
      "epoch": 1.4787729389208435,
      "grad_norm": 0.375,
      "learning_rate": 0.0007632504723509296,
      "loss": 4.8807,
      "step": 10798
    },
    {
      "epoch": 1.4789098877019995,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007632210606231491,
      "loss": 4.9096,
      "step": 10799
    },
    {
      "epoch": 1.4790468364831553,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007631916462996312,
      "loss": 4.8638,
      "step": 10800
    },
    {
      "epoch": 1.4791837852643113,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007631622293806193,
      "loss": 4.8228,
      "step": 10801
    },
    {
      "epoch": 1.479320734045467,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007631328098663576,
      "loss": 4.8661,
      "step": 10802
    },
    {
      "epoch": 1.4794576828266228,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007631033877570896,
      "loss": 4.9377,
      "step": 10803
    },
    {
      "epoch": 1.4795946316077786,
      "grad_norm": 0.375,
      "learning_rate": 0.0007630739630530591,
      "loss": 4.8123,
      "step": 10804
    },
    {
      "epoch": 1.4797315803889346,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007630445357545101,
      "loss": 4.854,
      "step": 10805
    },
    {
      "epoch": 1.4798685291700904,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007630151058616866,
      "loss": 4.7623,
      "step": 10806
    },
    {
      "epoch": 1.4800054779512462,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007629856733748323,
      "loss": 4.7541,
      "step": 10807
    },
    {
      "epoch": 1.4801424267324021,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007629562382941911,
      "loss": 4.8493,
      "step": 10808
    },
    {
      "epoch": 1.480279375513558,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000762926800620007,
      "loss": 4.8667,
      "step": 10809
    },
    {
      "epoch": 1.4804163242947137,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007628973603525239,
      "loss": 4.9085,
      "step": 10810
    },
    {
      "epoch": 1.4805532730758697,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007628679174919859,
      "loss": 4.8341,
      "step": 10811
    },
    {
      "epoch": 1.4806902218570255,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007628384720386366,
      "loss": 4.8184,
      "step": 10812
    },
    {
      "epoch": 1.4808271706381813,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007628090239927207,
      "loss": 4.7931,
      "step": 10813
    },
    {
      "epoch": 1.4809641194193373,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007627795733544818,
      "loss": 4.823,
      "step": 10814
    },
    {
      "epoch": 1.481101068200493,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000762750120124164,
      "loss": 4.8554,
      "step": 10815
    },
    {
      "epoch": 1.4812380169816488,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007627206643020114,
      "loss": 4.876,
      "step": 10816
    },
    {
      "epoch": 1.4813749657628046,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007626912058882682,
      "loss": 4.8619,
      "step": 10817
    },
    {
      "epoch": 1.4815119145439606,
      "grad_norm": 0.375,
      "learning_rate": 0.0007626617448831783,
      "loss": 4.8149,
      "step": 10818
    },
    {
      "epoch": 1.4816488633251164,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007626322812869863,
      "loss": 4.9129,
      "step": 10819
    },
    {
      "epoch": 1.4817858121062724,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007626028150999359,
      "loss": 4.8228,
      "step": 10820
    },
    {
      "epoch": 1.4819227608874281,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007625733463222715,
      "loss": 4.8901,
      "step": 10821
    },
    {
      "epoch": 1.482059709668584,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007625438749542373,
      "loss": 4.8451,
      "step": 10822
    },
    {
      "epoch": 1.4821966584497397,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007625144009960775,
      "loss": 4.8967,
      "step": 10823
    },
    {
      "epoch": 1.4823336072308957,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007624849244480365,
      "loss": 4.8542,
      "step": 10824
    },
    {
      "epoch": 1.4824705560120515,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007624554453103583,
      "loss": 4.8347,
      "step": 10825
    },
    {
      "epoch": 1.4826075047932075,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007624259635832877,
      "loss": 4.8543,
      "step": 10826
    },
    {
      "epoch": 1.4827444535743632,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007623964792670685,
      "loss": 4.8951,
      "step": 10827
    },
    {
      "epoch": 1.482881402355519,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007623669923619451,
      "loss": 4.8631,
      "step": 10828
    },
    {
      "epoch": 1.4830183511366748,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007623375028681621,
      "loss": 4.9102,
      "step": 10829
    },
    {
      "epoch": 1.4831552999178308,
      "grad_norm": 0.375,
      "learning_rate": 0.0007623080107859638,
      "loss": 4.8247,
      "step": 10830
    },
    {
      "epoch": 1.4832922486989866,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007622785161155946,
      "loss": 4.9188,
      "step": 10831
    },
    {
      "epoch": 1.4834291974801423,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000762249018857299,
      "loss": 4.7986,
      "step": 10832
    },
    {
      "epoch": 1.4835661462612983,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007622195190113213,
      "loss": 4.8842,
      "step": 10833
    },
    {
      "epoch": 1.4837030950424541,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007621900165779059,
      "loss": 4.8568,
      "step": 10834
    },
    {
      "epoch": 1.48384004382361,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007621605115572976,
      "loss": 4.9127,
      "step": 10835
    },
    {
      "epoch": 1.483976992604766,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007621310039497408,
      "loss": 4.8728,
      "step": 10836
    },
    {
      "epoch": 1.4841139413859217,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007621014937554799,
      "loss": 4.8677,
      "step": 10837
    },
    {
      "epoch": 1.4842508901670775,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007620719809747597,
      "loss": 4.7594,
      "step": 10838
    },
    {
      "epoch": 1.4843878389482335,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007620424656078243,
      "loss": 4.8576,
      "step": 10839
    },
    {
      "epoch": 1.4845247877293892,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000762012947654919,
      "loss": 4.9335,
      "step": 10840
    },
    {
      "epoch": 1.484661736510545,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007619834271162878,
      "loss": 4.8182,
      "step": 10841
    },
    {
      "epoch": 1.4847986852917008,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007619539039921758,
      "loss": 4.8647,
      "step": 10842
    },
    {
      "epoch": 1.4849356340728568,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007619243782828273,
      "loss": 4.8646,
      "step": 10843
    },
    {
      "epoch": 1.4850725828540126,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007618948499884873,
      "loss": 4.8104,
      "step": 10844
    },
    {
      "epoch": 1.4852095316351686,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007618653191094002,
      "loss": 4.8626,
      "step": 10845
    },
    {
      "epoch": 1.4853464804163243,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007618357856458111,
      "loss": 4.8478,
      "step": 10846
    },
    {
      "epoch": 1.48548342919748,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007618062495979643,
      "loss": 4.8591,
      "step": 10847
    },
    {
      "epoch": 1.4856203779786359,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007617767109661049,
      "loss": 4.8398,
      "step": 10848
    },
    {
      "epoch": 1.4857573267597919,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007617471697504776,
      "loss": 4.7788,
      "step": 10849
    },
    {
      "epoch": 1.4858942755409477,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007617176259513273,
      "loss": 4.8204,
      "step": 10850
    },
    {
      "epoch": 1.4860312243221037,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007616880795688987,
      "loss": 4.8921,
      "step": 10851
    },
    {
      "epoch": 1.4861681731032594,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007616585306034367,
      "loss": 4.825,
      "step": 10852
    },
    {
      "epoch": 1.4863051218844152,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007616289790551862,
      "loss": 4.7936,
      "step": 10853
    },
    {
      "epoch": 1.486442070665571,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007615994249243921,
      "loss": 4.8759,
      "step": 10854
    },
    {
      "epoch": 1.486579019446727,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007615698682112993,
      "loss": 4.9325,
      "step": 10855
    },
    {
      "epoch": 1.4867159682278828,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007615403089161528,
      "loss": 4.8346,
      "step": 10856
    },
    {
      "epoch": 1.4868529170090385,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007615107470391975,
      "loss": 4.8763,
      "step": 10857
    },
    {
      "epoch": 1.4869898657901945,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007614811825806784,
      "loss": 4.9113,
      "step": 10858
    },
    {
      "epoch": 1.4871268145713503,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007614516155408404,
      "loss": 4.8247,
      "step": 10859
    },
    {
      "epoch": 1.487263763352506,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007614220459199289,
      "loss": 4.8022,
      "step": 10860
    },
    {
      "epoch": 1.4874007121336619,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007613924737181884,
      "loss": 4.8327,
      "step": 10861
    },
    {
      "epoch": 1.4875376609148179,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007613628989358646,
      "loss": 4.7698,
      "step": 10862
    },
    {
      "epoch": 1.4876746096959736,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007613333215732019,
      "loss": 4.9367,
      "step": 10863
    },
    {
      "epoch": 1.4878115584771296,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000761303741630446,
      "loss": 4.852,
      "step": 10864
    },
    {
      "epoch": 1.4879485072582854,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007612741591078419,
      "loss": 4.7858,
      "step": 10865
    },
    {
      "epoch": 1.4880854560394412,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007612445740056345,
      "loss": 4.8795,
      "step": 10866
    },
    {
      "epoch": 1.488222404820597,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007612149863240692,
      "loss": 4.8152,
      "step": 10867
    },
    {
      "epoch": 1.488359353601753,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007611853960633911,
      "loss": 4.7935,
      "step": 10868
    },
    {
      "epoch": 1.4884963023829088,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007611558032238456,
      "loss": 4.8531,
      "step": 10869
    },
    {
      "epoch": 1.4886332511640648,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007611262078056777,
      "loss": 4.8814,
      "step": 10870
    },
    {
      "epoch": 1.4887701999452205,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007610966098091328,
      "loss": 4.8624,
      "step": 10871
    },
    {
      "epoch": 1.4889071487263763,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007610670092344561,
      "loss": 4.8715,
      "step": 10872
    },
    {
      "epoch": 1.489044097507532,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007610374060818932,
      "loss": 4.8698,
      "step": 10873
    },
    {
      "epoch": 1.489181046288688,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000761007800351689,
      "loss": 4.8528,
      "step": 10874
    },
    {
      "epoch": 1.4893179950698439,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007609781920440891,
      "loss": 4.8695,
      "step": 10875
    },
    {
      "epoch": 1.4894549438509999,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007609485811593388,
      "loss": 4.8826,
      "step": 10876
    },
    {
      "epoch": 1.4895918926321556,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007609189676976835,
      "loss": 4.8883,
      "step": 10877
    },
    {
      "epoch": 1.4897288414133114,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007608893516593688,
      "loss": 4.8764,
      "step": 10878
    },
    {
      "epoch": 1.4898657901944672,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007608597330446397,
      "loss": 4.8738,
      "step": 10879
    },
    {
      "epoch": 1.4900027389756232,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007608301118537421,
      "loss": 4.8509,
      "step": 10880
    },
    {
      "epoch": 1.490139687756779,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007608004880869212,
      "loss": 4.8329,
      "step": 10881
    },
    {
      "epoch": 1.4902766365379347,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007607708617444227,
      "loss": 4.8496,
      "step": 10882
    },
    {
      "epoch": 1.4904135853190907,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007607412328264918,
      "loss": 4.8311,
      "step": 10883
    },
    {
      "epoch": 1.4905505341002465,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007607116013333745,
      "loss": 4.8744,
      "step": 10884
    },
    {
      "epoch": 1.4906874828814023,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007606819672653159,
      "loss": 4.7994,
      "step": 10885
    },
    {
      "epoch": 1.490824431662558,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007606523306225621,
      "loss": 4.8722,
      "step": 10886
    },
    {
      "epoch": 1.490961380443714,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007606226914053582,
      "loss": 4.8249,
      "step": 10887
    },
    {
      "epoch": 1.4910983292248698,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007605930496139502,
      "loss": 4.8748,
      "step": 10888
    },
    {
      "epoch": 1.4912352780060258,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007605634052485835,
      "loss": 4.9596,
      "step": 10889
    },
    {
      "epoch": 1.4913722267871816,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007605337583095037,
      "loss": 4.7747,
      "step": 10890
    },
    {
      "epoch": 1.4915091755683374,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0007605041087969569,
      "loss": 4.7636,
      "step": 10891
    },
    {
      "epoch": 1.4916461243494932,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007604744567111882,
      "loss": 4.7843,
      "step": 10892
    },
    {
      "epoch": 1.4917830731306492,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007604448020524441,
      "loss": 4.8253,
      "step": 10893
    },
    {
      "epoch": 1.491920021911805,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007604151448209698,
      "loss": 4.9777,
      "step": 10894
    },
    {
      "epoch": 1.492056970692961,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007603854850170112,
      "loss": 4.8752,
      "step": 10895
    },
    {
      "epoch": 1.4921939194741167,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007603558226408141,
      "loss": 4.8586,
      "step": 10896
    },
    {
      "epoch": 1.4923308682552725,
      "grad_norm": 0.375,
      "learning_rate": 0.0007603261576926243,
      "loss": 4.9223,
      "step": 10897
    },
    {
      "epoch": 1.4924678170364283,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007602964901726879,
      "loss": 4.9016,
      "step": 10898
    },
    {
      "epoch": 1.4926047658175843,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007602668200812503,
      "loss": 4.8466,
      "step": 10899
    },
    {
      "epoch": 1.49274171459874,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007602371474185578,
      "loss": 4.8637,
      "step": 10900
    },
    {
      "epoch": 1.492878663379896,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000760207472184856,
      "loss": 4.8394,
      "step": 10901
    },
    {
      "epoch": 1.4930156121610518,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007601777943803911,
      "loss": 4.8763,
      "step": 10902
    },
    {
      "epoch": 1.4931525609422076,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007601481140054088,
      "loss": 4.9247,
      "step": 10903
    },
    {
      "epoch": 1.4932895097233634,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007601184310601552,
      "loss": 4.8128,
      "step": 10904
    },
    {
      "epoch": 1.4934264585045194,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007600887455448764,
      "loss": 4.7746,
      "step": 10905
    },
    {
      "epoch": 1.4935634072856752,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000760059057459818,
      "loss": 4.816,
      "step": 10906
    },
    {
      "epoch": 1.493700356066831,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007600293668052267,
      "loss": 4.75,
      "step": 10907
    },
    {
      "epoch": 1.493837304847987,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007599996735813477,
      "loss": 4.7618,
      "step": 10908
    },
    {
      "epoch": 1.4939742536291427,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007599699777884279,
      "loss": 4.8044,
      "step": 10909
    },
    {
      "epoch": 1.4941112024102985,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007599402794267129,
      "loss": 4.8394,
      "step": 10910
    },
    {
      "epoch": 1.4942481511914543,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007599105784964488,
      "loss": 4.787,
      "step": 10911
    },
    {
      "epoch": 1.4943850999726103,
      "grad_norm": 0.375,
      "learning_rate": 0.0007598808749978821,
      "loss": 4.7938,
      "step": 10912
    },
    {
      "epoch": 1.494522048753766,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007598511689312587,
      "loss": 4.882,
      "step": 10913
    },
    {
      "epoch": 1.494658997534922,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007598214602968249,
      "loss": 4.8714,
      "step": 10914
    },
    {
      "epoch": 1.4947959463160778,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007597917490948266,
      "loss": 4.8168,
      "step": 10915
    },
    {
      "epoch": 1.4949328950972336,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007597620353255104,
      "loss": 4.9012,
      "step": 10916
    },
    {
      "epoch": 1.4950698438783894,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007597323189891224,
      "loss": 4.759,
      "step": 10917
    },
    {
      "epoch": 1.4952067926595454,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007597026000859089,
      "loss": 4.8031,
      "step": 10918
    },
    {
      "epoch": 1.4953437414407011,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007596728786161161,
      "loss": 4.758,
      "step": 10919
    },
    {
      "epoch": 1.4954806902218571,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007596431545799905,
      "loss": 4.8218,
      "step": 10920
    },
    {
      "epoch": 1.495617639003013,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007596134279777781,
      "loss": 4.8819,
      "step": 10921
    },
    {
      "epoch": 1.4957545877841687,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007595836988097257,
      "loss": 4.8863,
      "step": 10922
    },
    {
      "epoch": 1.4958915365653245,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007595539670760791,
      "loss": 4.8385,
      "step": 10923
    },
    {
      "epoch": 1.4960284853464805,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007595242327770852,
      "loss": 4.8598,
      "step": 10924
    },
    {
      "epoch": 1.4961654341276363,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007594944959129901,
      "loss": 4.8951,
      "step": 10925
    },
    {
      "epoch": 1.4963023829087923,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007594647564840405,
      "loss": 4.9008,
      "step": 10926
    },
    {
      "epoch": 1.496439331689948,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007594350144904826,
      "loss": 4.862,
      "step": 10927
    },
    {
      "epoch": 1.4965762804711038,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007594052699325631,
      "loss": 4.8567,
      "step": 10928
    },
    {
      "epoch": 1.4967132292522596,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007593755228105284,
      "loss": 4.906,
      "step": 10929
    },
    {
      "epoch": 1.4968501780334156,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007593457731246249,
      "loss": 4.8864,
      "step": 10930
    },
    {
      "epoch": 1.4969871268145714,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007593160208750994,
      "loss": 4.8439,
      "step": 10931
    },
    {
      "epoch": 1.4971240755957271,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0007592862660621983,
      "loss": 4.7913,
      "step": 10932
    },
    {
      "epoch": 1.4972610243768831,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007592565086861682,
      "loss": 4.8381,
      "step": 10933
    },
    {
      "epoch": 1.497397973158039,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007592267487472558,
      "loss": 4.8637,
      "step": 10934
    },
    {
      "epoch": 1.4975349219391947,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007591969862457075,
      "loss": 4.7954,
      "step": 10935
    },
    {
      "epoch": 1.4976718707203505,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007591672211817701,
      "loss": 4.8581,
      "step": 10936
    },
    {
      "epoch": 1.4978088195015065,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007591374535556904,
      "loss": 4.8598,
      "step": 10937
    },
    {
      "epoch": 1.4979457682826622,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007591076833677148,
      "loss": 4.7834,
      "step": 10938
    },
    {
      "epoch": 1.4980827170638182,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007590779106180904,
      "loss": 4.8886,
      "step": 10939
    },
    {
      "epoch": 1.498219665844974,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007590481353070636,
      "loss": 4.8084,
      "step": 10940
    },
    {
      "epoch": 1.4983566146261298,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007590183574348811,
      "loss": 4.8548,
      "step": 10941
    },
    {
      "epoch": 1.4984935634072856,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00075898857700179,
      "loss": 4.8213,
      "step": 10942
    },
    {
      "epoch": 1.4986305121884416,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000758958794008037,
      "loss": 4.8171,
      "step": 10943
    },
    {
      "epoch": 1.4987674609695973,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007589290084538687,
      "loss": 4.8791,
      "step": 10944
    },
    {
      "epoch": 1.4989044097507533,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000758899220339532,
      "loss": 4.8362,
      "step": 10945
    },
    {
      "epoch": 1.4990413585319091,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007588694296652741,
      "loss": 4.8586,
      "step": 10946
    },
    {
      "epoch": 1.499178307313065,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007588396364313416,
      "loss": 4.8256,
      "step": 10947
    },
    {
      "epoch": 1.4993152560942207,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0007588098406379813,
      "loss": 4.8442,
      "step": 10948
    },
    {
      "epoch": 1.4994522048753767,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007587800422854405,
      "loss": 4.8403,
      "step": 10949
    },
    {
      "epoch": 1.4995891536565324,
      "grad_norm": 0.46875,
      "learning_rate": 0.0007587502413739657,
      "loss": 4.8537,
      "step": 10950
    },
    {
      "epoch": 1.4997261024376882,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007587204379038041,
      "loss": 4.8112,
      "step": 10951
    },
    {
      "epoch": 1.4998630512188442,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007586906318752027,
      "loss": 4.8657,
      "step": 10952
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007586608232884085,
      "loss": 4.9847,
      "step": 10953
    },
    {
      "epoch": 1.5001369487811558,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007586310121436685,
      "loss": 4.8395,
      "step": 10954
    },
    {
      "epoch": 1.5002738975623116,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007586011984412297,
      "loss": 4.7819,
      "step": 10955
    },
    {
      "epoch": 1.5004108463434676,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007585713821813393,
      "loss": 4.8779,
      "step": 10956
    },
    {
      "epoch": 1.5005477951246236,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007585415633642443,
      "loss": 4.8737,
      "step": 10957
    },
    {
      "epoch": 1.5006847439057793,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007585117419901919,
      "loss": 4.8515,
      "step": 10958
    },
    {
      "epoch": 1.500821692686935,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000758481918059429,
      "loss": 4.8199,
      "step": 10959
    },
    {
      "epoch": 1.5009586414680909,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007584520915722031,
      "loss": 4.7908,
      "step": 10960
    },
    {
      "epoch": 1.5010955902492467,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007584222625287612,
      "loss": 4.8842,
      "step": 10961
    },
    {
      "epoch": 1.5012325390304027,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007583924309293506,
      "loss": 4.877,
      "step": 10962
    },
    {
      "epoch": 1.5013694878115584,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007583625967742183,
      "loss": 4.8772,
      "step": 10963
    },
    {
      "epoch": 1.5015064365927144,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007583327600636116,
      "loss": 4.8651,
      "step": 10964
    },
    {
      "epoch": 1.5016433853738702,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007583029207977781,
      "loss": 4.8396,
      "step": 10965
    },
    {
      "epoch": 1.501780334155026,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007582730789769646,
      "loss": 4.8981,
      "step": 10966
    },
    {
      "epoch": 1.5019172829361818,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007582432346014188,
      "loss": 4.8804,
      "step": 10967
    },
    {
      "epoch": 1.5020542317173378,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007582133876713877,
      "loss": 4.8611,
      "step": 10968
    },
    {
      "epoch": 1.5021911804984935,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007581835381871191,
      "loss": 4.913,
      "step": 10969
    },
    {
      "epoch": 1.5023281292796495,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007581536861488598,
      "loss": 4.857,
      "step": 10970
    },
    {
      "epoch": 1.5024650780608053,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007581238315568575,
      "loss": 4.8165,
      "step": 10971
    },
    {
      "epoch": 1.502602026841961,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007580939744113598,
      "loss": 4.9196,
      "step": 10972
    },
    {
      "epoch": 1.5027389756231169,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007580641147126137,
      "loss": 4.8989,
      "step": 10973
    },
    {
      "epoch": 1.5028759244042726,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000758034252460867,
      "loss": 4.9413,
      "step": 10974
    },
    {
      "epoch": 1.5030128731854286,
      "grad_norm": 0.375,
      "learning_rate": 0.000758004387656367,
      "loss": 4.8883,
      "step": 10975
    },
    {
      "epoch": 1.5031498219665846,
      "grad_norm": 0.375,
      "learning_rate": 0.0007579745202993612,
      "loss": 4.9624,
      "step": 10976
    },
    {
      "epoch": 1.5032867707477404,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007579446503900972,
      "loss": 4.8446,
      "step": 10977
    },
    {
      "epoch": 1.5034237195288962,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007579147779288225,
      "loss": 4.8933,
      "step": 10978
    },
    {
      "epoch": 1.503560668310052,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007578849029157848,
      "loss": 4.8127,
      "step": 10979
    },
    {
      "epoch": 1.5036976170912077,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007578550253512313,
      "loss": 4.8715,
      "step": 10980
    },
    {
      "epoch": 1.5038345658723637,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007578251452354099,
      "loss": 4.8579,
      "step": 10981
    },
    {
      "epoch": 1.5039715146535197,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007577952625685684,
      "loss": 4.8904,
      "step": 10982
    },
    {
      "epoch": 1.5041084634346755,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000757765377350954,
      "loss": 4.9062,
      "step": 10983
    },
    {
      "epoch": 1.5042454122158313,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007577354895828146,
      "loss": 4.8451,
      "step": 10984
    },
    {
      "epoch": 1.504382360996987,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007577055992643979,
      "loss": 4.8389,
      "step": 10985
    },
    {
      "epoch": 1.5045193097781429,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007576757063959516,
      "loss": 4.8152,
      "step": 10986
    },
    {
      "epoch": 1.5046562585592989,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007576458109777234,
      "loss": 4.8072,
      "step": 10987
    },
    {
      "epoch": 1.5047932073404546,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007576159130099609,
      "loss": 4.8912,
      "step": 10988
    },
    {
      "epoch": 1.5049301561216106,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007575860124929122,
      "loss": 4.7985,
      "step": 10989
    },
    {
      "epoch": 1.5050671049027664,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000757556109426825,
      "loss": 4.7564,
      "step": 10990
    },
    {
      "epoch": 1.5052040536839222,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007575262038119468,
      "loss": 4.8869,
      "step": 10991
    },
    {
      "epoch": 1.505341002465078,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007574962956485259,
      "loss": 4.8875,
      "step": 10992
    },
    {
      "epoch": 1.505477951246234,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007574663849368098,
      "loss": 4.8981,
      "step": 10993
    },
    {
      "epoch": 1.5056149000273897,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007574364716770465,
      "loss": 4.8673,
      "step": 10994
    },
    {
      "epoch": 1.5057518488085457,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007574065558694839,
      "loss": 4.8696,
      "step": 10995
    },
    {
      "epoch": 1.5058887975897015,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00075737663751437,
      "loss": 4.7938,
      "step": 10996
    },
    {
      "epoch": 1.5060257463708573,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007573467166119526,
      "loss": 4.8565,
      "step": 10997
    },
    {
      "epoch": 1.506162695152013,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007573167931624797,
      "loss": 4.8139,
      "step": 10998
    },
    {
      "epoch": 1.5062996439331688,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007572868671661993,
      "loss": 4.7679,
      "step": 10999
    },
    {
      "epoch": 1.5064365927143248,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007572569386233595,
      "loss": 4.8873,
      "step": 11000
    },
    {
      "epoch": 1.5065735414954808,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007572270075342081,
      "loss": 4.802,
      "step": 11001
    },
    {
      "epoch": 1.5067104902766366,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007571970738989934,
      "loss": 4.8905,
      "step": 11002
    },
    {
      "epoch": 1.5068474390577924,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007571671377179633,
      "loss": 4.937,
      "step": 11003
    },
    {
      "epoch": 1.5069843878389482,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007571371989913659,
      "loss": 4.862,
      "step": 11004
    },
    {
      "epoch": 1.507121336620104,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007571072577194495,
      "loss": 4.8836,
      "step": 11005
    },
    {
      "epoch": 1.50725828540126,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000757077313902462,
      "loss": 4.8551,
      "step": 11006
    },
    {
      "epoch": 1.5073952341824157,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007570473675406514,
      "loss": 4.8597,
      "step": 11007
    },
    {
      "epoch": 1.5075321829635717,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007570174186342664,
      "loss": 4.7289,
      "step": 11008
    },
    {
      "epoch": 1.5076691317447275,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007569874671835547,
      "loss": 4.7591,
      "step": 11009
    },
    {
      "epoch": 1.5078060805258833,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007569575131887649,
      "loss": 4.7727,
      "step": 11010
    },
    {
      "epoch": 1.507943029307039,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007569275566501449,
      "loss": 4.8439,
      "step": 11011
    },
    {
      "epoch": 1.508079978088195,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007568975975679431,
      "loss": 4.9064,
      "step": 11012
    },
    {
      "epoch": 1.5082169268693508,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007568676359424078,
      "loss": 4.7961,
      "step": 11013
    },
    {
      "epoch": 1.5083538756505068,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007568376717737873,
      "loss": 4.8159,
      "step": 11014
    },
    {
      "epoch": 1.5084908244316626,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007568077050623299,
      "loss": 4.8924,
      "step": 11015
    },
    {
      "epoch": 1.5086277732128184,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007567777358082838,
      "loss": 4.8433,
      "step": 11016
    },
    {
      "epoch": 1.5087647219939742,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007567477640118978,
      "loss": 4.8514,
      "step": 11017
    },
    {
      "epoch": 1.5089016707751302,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007567177896734198,
      "loss": 4.8579,
      "step": 11018
    },
    {
      "epoch": 1.509038619556286,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007566878127930984,
      "loss": 4.895,
      "step": 11019
    },
    {
      "epoch": 1.509175568337442,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007566578333711821,
      "loss": 4.8535,
      "step": 11020
    },
    {
      "epoch": 1.5093125171185977,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007566278514079192,
      "loss": 4.7804,
      "step": 11021
    },
    {
      "epoch": 1.5094494658997535,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007565978669035581,
      "loss": 4.7825,
      "step": 11022
    },
    {
      "epoch": 1.5095864146809093,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007565678798583475,
      "loss": 4.8692,
      "step": 11023
    },
    {
      "epoch": 1.509723363462065,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007565378902725359,
      "loss": 4.8365,
      "step": 11024
    },
    {
      "epoch": 1.509860312243221,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007565078981463717,
      "loss": 4.7751,
      "step": 11025
    },
    {
      "epoch": 1.509997261024377,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007564779034801036,
      "loss": 4.8244,
      "step": 11026
    },
    {
      "epoch": 1.5101342098055328,
      "grad_norm": 0.375,
      "learning_rate": 0.0007564479062739801,
      "loss": 4.9129,
      "step": 11027
    },
    {
      "epoch": 1.5102711585866886,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007564179065282496,
      "loss": 4.8835,
      "step": 11028
    },
    {
      "epoch": 1.5104081073678444,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007563879042431609,
      "loss": 4.8736,
      "step": 11029
    },
    {
      "epoch": 1.5105450561490001,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007563578994189628,
      "loss": 4.8072,
      "step": 11030
    },
    {
      "epoch": 1.5106820049301561,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007563278920559038,
      "loss": 4.8288,
      "step": 11031
    },
    {
      "epoch": 1.510818953711312,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007562978821542324,
      "loss": 4.7891,
      "step": 11032
    },
    {
      "epoch": 1.510955902492468,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007562678697141975,
      "loss": 4.884,
      "step": 11033
    },
    {
      "epoch": 1.5110928512736237,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007562378547360479,
      "loss": 4.9042,
      "step": 11034
    },
    {
      "epoch": 1.5112298000547795,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007562078372200321,
      "loss": 4.8287,
      "step": 11035
    },
    {
      "epoch": 1.5113667488359352,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007561778171663989,
      "loss": 4.8134,
      "step": 11036
    },
    {
      "epoch": 1.5115036976170912,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007561477945753974,
      "loss": 4.8194,
      "step": 11037
    },
    {
      "epoch": 1.511640646398247,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007561177694472761,
      "loss": 4.8434,
      "step": 11038
    },
    {
      "epoch": 1.511777595179403,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007560877417822838,
      "loss": 4.8428,
      "step": 11039
    },
    {
      "epoch": 1.5119145439605588,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007560577115806696,
      "loss": 4.7483,
      "step": 11040
    },
    {
      "epoch": 1.5120514927417146,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007560276788426821,
      "loss": 4.8207,
      "step": 11041
    },
    {
      "epoch": 1.5121884415228704,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007559976435685704,
      "loss": 4.8209,
      "step": 11042
    },
    {
      "epoch": 1.5123253903040261,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007559676057585832,
      "loss": 4.8395,
      "step": 11043
    },
    {
      "epoch": 1.5124623390851821,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007559375654129697,
      "loss": 4.7756,
      "step": 11044
    },
    {
      "epoch": 1.5125992878663381,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007559075225319785,
      "loss": 4.8937,
      "step": 11045
    },
    {
      "epoch": 1.512736236647494,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007558774771158588,
      "loss": 4.8679,
      "step": 11046
    },
    {
      "epoch": 1.5128731854286497,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007558474291648597,
      "loss": 4.8571,
      "step": 11047
    },
    {
      "epoch": 1.5130101342098055,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00075581737867923,
      "loss": 4.8617,
      "step": 11048
    },
    {
      "epoch": 1.5131470829909612,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007557873256592188,
      "loss": 4.814,
      "step": 11049
    },
    {
      "epoch": 1.5132840317721172,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007557572701050752,
      "loss": 4.8879,
      "step": 11050
    },
    {
      "epoch": 1.5134209805532732,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007557272120170482,
      "loss": 4.8447,
      "step": 11051
    },
    {
      "epoch": 1.513557929334429,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007556971513953868,
      "loss": 4.9056,
      "step": 11052
    },
    {
      "epoch": 1.5136948781155848,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007556670882403405,
      "loss": 4.9084,
      "step": 11053
    },
    {
      "epoch": 1.5138318268967406,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007556370225521582,
      "loss": 4.863,
      "step": 11054
    },
    {
      "epoch": 1.5139687756778963,
      "grad_norm": 0.34375,
      "learning_rate": 0.000755606954331089,
      "loss": 4.8741,
      "step": 11055
    },
    {
      "epoch": 1.5141057244590523,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007555768835773821,
      "loss": 4.85,
      "step": 11056
    },
    {
      "epoch": 1.5142426732402081,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007555468102912868,
      "loss": 4.8198,
      "step": 11057
    },
    {
      "epoch": 1.5143796220213641,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007555167344730523,
      "loss": 4.7804,
      "step": 11058
    },
    {
      "epoch": 1.51451657080252,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007554866561229278,
      "loss": 4.8476,
      "step": 11059
    },
    {
      "epoch": 1.5146535195836757,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007554565752411626,
      "loss": 4.825,
      "step": 11060
    },
    {
      "epoch": 1.5147904683648314,
      "grad_norm": 0.375,
      "learning_rate": 0.000755426491828006,
      "loss": 4.7577,
      "step": 11061
    },
    {
      "epoch": 1.5149274171459874,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007553964058837072,
      "loss": 4.8004,
      "step": 11062
    },
    {
      "epoch": 1.5150643659271432,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007553663174085155,
      "loss": 4.7576,
      "step": 11063
    },
    {
      "epoch": 1.5152013147082992,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007553362264026805,
      "loss": 4.8357,
      "step": 11064
    },
    {
      "epoch": 1.515338263489455,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007553061328664514,
      "loss": 4.8239,
      "step": 11065
    },
    {
      "epoch": 1.5154752122706108,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007552760368000778,
      "loss": 4.7696,
      "step": 11066
    },
    {
      "epoch": 1.5156121610517665,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007552459382038088,
      "loss": 4.8354,
      "step": 11067
    },
    {
      "epoch": 1.5157491098329223,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007552158370778939,
      "loss": 4.8592,
      "step": 11068
    },
    {
      "epoch": 1.5158860586140783,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007551857334225828,
      "loss": 4.865,
      "step": 11069
    },
    {
      "epoch": 1.5160230073952343,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007551556272381248,
      "loss": 4.7944,
      "step": 11070
    },
    {
      "epoch": 1.51615995617639,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007551255185247693,
      "loss": 4.8412,
      "step": 11071
    },
    {
      "epoch": 1.5162969049575459,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007550954072827658,
      "loss": 4.8281,
      "step": 11072
    },
    {
      "epoch": 1.5164338537387017,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007550652935123641,
      "loss": 4.7822,
      "step": 11073
    },
    {
      "epoch": 1.5165708025198574,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007550351772138135,
      "loss": 4.7939,
      "step": 11074
    },
    {
      "epoch": 1.5167077513010134,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000755005058387364,
      "loss": 4.8212,
      "step": 11075
    },
    {
      "epoch": 1.5168447000821694,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007549749370332647,
      "loss": 4.8166,
      "step": 11076
    },
    {
      "epoch": 1.5169816488633252,
      "grad_norm": 0.375,
      "learning_rate": 0.0007549448131517654,
      "loss": 4.8174,
      "step": 11077
    },
    {
      "epoch": 1.517118597644481,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007549146867431158,
      "loss": 4.7635,
      "step": 11078
    },
    {
      "epoch": 1.5172555464256368,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007548845578075657,
      "loss": 4.827,
      "step": 11079
    },
    {
      "epoch": 1.5173924952067925,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007548544263453644,
      "loss": 4.8119,
      "step": 11080
    },
    {
      "epoch": 1.5175294439879485,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007548242923567619,
      "loss": 4.8786,
      "step": 11081
    },
    {
      "epoch": 1.5176663927691043,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007547941558420078,
      "loss": 4.8367,
      "step": 11082
    },
    {
      "epoch": 1.5178033415502603,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007547640168013519,
      "loss": 4.9167,
      "step": 11083
    },
    {
      "epoch": 1.517940290331416,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007547338752350441,
      "loss": 4.7742,
      "step": 11084
    },
    {
      "epoch": 1.5180772391125719,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007547037311433339,
      "loss": 4.8865,
      "step": 11085
    },
    {
      "epoch": 1.5182141878937276,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007546735845264714,
      "loss": 4.852,
      "step": 11086
    },
    {
      "epoch": 1.5183511366748836,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007546434353847062,
      "loss": 4.8077,
      "step": 11087
    },
    {
      "epoch": 1.5184880854560394,
      "grad_norm": 0.375,
      "learning_rate": 0.0007546132837182885,
      "loss": 4.8334,
      "step": 11088
    },
    {
      "epoch": 1.5186250342371954,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007545831295274676,
      "loss": 4.8911,
      "step": 11089
    },
    {
      "epoch": 1.5187619830183512,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007545529728124939,
      "loss": 4.814,
      "step": 11090
    },
    {
      "epoch": 1.518898931799507,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007545228135736171,
      "loss": 4.8994,
      "step": 11091
    },
    {
      "epoch": 1.5190358805806627,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007544926518110871,
      "loss": 4.7729,
      "step": 11092
    },
    {
      "epoch": 1.5191728293618185,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007544624875251543,
      "loss": 4.8918,
      "step": 11093
    },
    {
      "epoch": 1.5193097781429745,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007544323207160678,
      "loss": 4.7955,
      "step": 11094
    },
    {
      "epoch": 1.5194467269241305,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007544021513840785,
      "loss": 4.8788,
      "step": 11095
    },
    {
      "epoch": 1.5195836757052863,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000754371979529436,
      "loss": 4.8052,
      "step": 11096
    },
    {
      "epoch": 1.519720624486442,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007543418051523901,
      "loss": 4.8707,
      "step": 11097
    },
    {
      "epoch": 1.5198575732675979,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007543116282531915,
      "loss": 4.8194,
      "step": 11098
    },
    {
      "epoch": 1.5199945220487536,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007542814488320896,
      "loss": 4.8179,
      "step": 11099
    },
    {
      "epoch": 1.5201314708299096,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000754251266889335,
      "loss": 4.8207,
      "step": 11100
    },
    {
      "epoch": 1.5202684196110656,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007542210824251777,
      "loss": 4.82,
      "step": 11101
    },
    {
      "epoch": 1.5204053683922214,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007541908954398676,
      "loss": 4.9136,
      "step": 11102
    },
    {
      "epoch": 1.5205423171733772,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007541607059336553,
      "loss": 4.8083,
      "step": 11103
    },
    {
      "epoch": 1.520679265954533,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007541305139067906,
      "loss": 4.843,
      "step": 11104
    },
    {
      "epoch": 1.5208162147356887,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007541003193595239,
      "loss": 4.836,
      "step": 11105
    },
    {
      "epoch": 1.5209531635168447,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007540701222921054,
      "loss": 4.8595,
      "step": 11106
    },
    {
      "epoch": 1.5210901122980005,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007540399227047854,
      "loss": 4.7981,
      "step": 11107
    },
    {
      "epoch": 1.5212270610791565,
      "grad_norm": 0.390625,
      "learning_rate": 0.000754009720597814,
      "loss": 4.7786,
      "step": 11108
    },
    {
      "epoch": 1.5213640098603123,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007539795159714417,
      "loss": 4.7866,
      "step": 11109
    },
    {
      "epoch": 1.521500958641468,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007539493088259187,
      "loss": 4.8559,
      "step": 11110
    },
    {
      "epoch": 1.5216379074226238,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007539190991614952,
      "loss": 4.8517,
      "step": 11111
    },
    {
      "epoch": 1.5217748562037798,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007538888869784218,
      "loss": 4.8759,
      "step": 11112
    },
    {
      "epoch": 1.5219118049849356,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007538586722769489,
      "loss": 4.9103,
      "step": 11113
    },
    {
      "epoch": 1.5220487537660916,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007538284550573267,
      "loss": 4.8201,
      "step": 11114
    },
    {
      "epoch": 1.5221857025472474,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007537982353198056,
      "loss": 4.9091,
      "step": 11115
    },
    {
      "epoch": 1.5223226513284032,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007537680130646362,
      "loss": 4.8353,
      "step": 11116
    },
    {
      "epoch": 1.522459600109559,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007537377882920691,
      "loss": 4.8092,
      "step": 11117
    },
    {
      "epoch": 1.5225965488907147,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007537075610023542,
      "loss": 4.9258,
      "step": 11118
    },
    {
      "epoch": 1.5227334976718707,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007536773311957425,
      "loss": 4.8771,
      "step": 11119
    },
    {
      "epoch": 1.5228704464530267,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007536470988724844,
      "loss": 4.8622,
      "step": 11120
    },
    {
      "epoch": 1.5230073952341825,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007536168640328306,
      "loss": 4.8953,
      "step": 11121
    },
    {
      "epoch": 1.5231443440153383,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007535866266770313,
      "loss": 4.831,
      "step": 11122
    },
    {
      "epoch": 1.523281292796494,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007535563868053374,
      "loss": 4.8349,
      "step": 11123
    },
    {
      "epoch": 1.5234182415776498,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007535261444179994,
      "loss": 4.8235,
      "step": 11124
    },
    {
      "epoch": 1.5235551903588058,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007534958995152678,
      "loss": 4.8111,
      "step": 11125
    },
    {
      "epoch": 1.5236921391399618,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007534656520973935,
      "loss": 4.8498,
      "step": 11126
    },
    {
      "epoch": 1.5238290879211176,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007534354021646268,
      "loss": 4.8519,
      "step": 11127
    },
    {
      "epoch": 1.5239660367022734,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007534051497172188,
      "loss": 4.9441,
      "step": 11128
    },
    {
      "epoch": 1.5241029854834292,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007533748947554199,
      "loss": 4.877,
      "step": 11129
    },
    {
      "epoch": 1.524239934264585,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007533446372794812,
      "loss": 4.8301,
      "step": 11130
    },
    {
      "epoch": 1.524376883045741,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007533143772896531,
      "loss": 4.8526,
      "step": 11131
    },
    {
      "epoch": 1.5245138318268967,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007532841147861864,
      "loss": 4.8319,
      "step": 11132
    },
    {
      "epoch": 1.5246507806080527,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000753253849769332,
      "loss": 4.7874,
      "step": 11133
    },
    {
      "epoch": 1.5247877293892085,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007532235822393407,
      "loss": 4.8494,
      "step": 11134
    },
    {
      "epoch": 1.5249246781703643,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007531933121964633,
      "loss": 4.8123,
      "step": 11135
    },
    {
      "epoch": 1.52506162695152,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007531630396409507,
      "loss": 4.7853,
      "step": 11136
    },
    {
      "epoch": 1.525198575732676,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007531327645730537,
      "loss": 4.755,
      "step": 11137
    },
    {
      "epoch": 1.5253355245138318,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007531024869930233,
      "loss": 4.7921,
      "step": 11138
    },
    {
      "epoch": 1.5254724732949878,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007530722069011104,
      "loss": 4.817,
      "step": 11139
    },
    {
      "epoch": 1.5256094220761436,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000753041924297566,
      "loss": 4.8925,
      "step": 11140
    },
    {
      "epoch": 1.5257463708572994,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007530116391826407,
      "loss": 4.775,
      "step": 11141
    },
    {
      "epoch": 1.5258833196384551,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007529813515565859,
      "loss": 4.8702,
      "step": 11142
    },
    {
      "epoch": 1.526020268419611,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007529510614196524,
      "loss": 4.8869,
      "step": 11143
    },
    {
      "epoch": 1.526157217200767,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007529207687720913,
      "loss": 4.8312,
      "step": 11144
    },
    {
      "epoch": 1.526294165981923,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007528904736141535,
      "loss": 4.8609,
      "step": 11145
    },
    {
      "epoch": 1.5264311147630787,
      "grad_norm": 0.375,
      "learning_rate": 0.0007528601759460903,
      "loss": 4.8649,
      "step": 11146
    },
    {
      "epoch": 1.5265680635442345,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007528298757681527,
      "loss": 4.8265,
      "step": 11147
    },
    {
      "epoch": 1.5267050123253902,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007527995730805915,
      "loss": 4.8232,
      "step": 11148
    },
    {
      "epoch": 1.526841961106546,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007527692678836583,
      "loss": 4.7895,
      "step": 11149
    },
    {
      "epoch": 1.526978909887702,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000752738960177604,
      "loss": 4.8081,
      "step": 11150
    },
    {
      "epoch": 1.5271158586688578,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007527086499626798,
      "loss": 4.8006,
      "step": 11151
    },
    {
      "epoch": 1.5272528074500138,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007526783372391369,
      "loss": 4.8151,
      "step": 11152
    },
    {
      "epoch": 1.5273897562311696,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007526480220072263,
      "loss": 4.8287,
      "step": 11153
    },
    {
      "epoch": 1.5275267050123253,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007526177042671996,
      "loss": 4.8345,
      "step": 11154
    },
    {
      "epoch": 1.5276636537934811,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000752587384019308,
      "loss": 4.8719,
      "step": 11155
    },
    {
      "epoch": 1.5278006025746371,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007525570612638025,
      "loss": 4.7959,
      "step": 11156
    },
    {
      "epoch": 1.527937551355793,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007525267360009345,
      "loss": 4.877,
      "step": 11157
    },
    {
      "epoch": 1.528074500136949,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007524964082309553,
      "loss": 4.8196,
      "step": 11158
    },
    {
      "epoch": 1.5282114489181047,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007524660779541164,
      "loss": 4.876,
      "step": 11159
    },
    {
      "epoch": 1.5283483976992605,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007524357451706691,
      "loss": 4.8558,
      "step": 11160
    },
    {
      "epoch": 1.5284853464804162,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007524054098808645,
      "loss": 4.8343,
      "step": 11161
    },
    {
      "epoch": 1.528622295261572,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007523750720849544,
      "loss": 4.9224,
      "step": 11162
    },
    {
      "epoch": 1.528759244042728,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007523447317831899,
      "loss": 4.8553,
      "step": 11163
    },
    {
      "epoch": 1.528896192823884,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007523143889758227,
      "loss": 4.7851,
      "step": 11164
    },
    {
      "epoch": 1.5290331416050398,
      "grad_norm": 0.40625,
      "learning_rate": 0.000752284043663104,
      "loss": 4.8305,
      "step": 11165
    },
    {
      "epoch": 1.5291700903861956,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007522536958452855,
      "loss": 4.8955,
      "step": 11166
    },
    {
      "epoch": 1.5293070391673513,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007522233455226187,
      "loss": 4.8918,
      "step": 11167
    },
    {
      "epoch": 1.529443987948507,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007521929926953547,
      "loss": 4.7641,
      "step": 11168
    },
    {
      "epoch": 1.529580936729663,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007521626373637455,
      "loss": 4.8759,
      "step": 11169
    },
    {
      "epoch": 1.529717885510819,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007521322795280426,
      "loss": 4.7405,
      "step": 11170
    },
    {
      "epoch": 1.5298548342919749,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007521019191884974,
      "loss": 4.7909,
      "step": 11171
    },
    {
      "epoch": 1.5299917830731307,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007520715563453617,
      "loss": 4.7651,
      "step": 11172
    },
    {
      "epoch": 1.5301287318542864,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000752041190998887,
      "loss": 4.803,
      "step": 11173
    },
    {
      "epoch": 1.5302656806354422,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000752010823149325,
      "loss": 4.8096,
      "step": 11174
    },
    {
      "epoch": 1.5304026294165982,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007519804527969272,
      "loss": 4.8241,
      "step": 11175
    },
    {
      "epoch": 1.530539578197754,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007519500799419456,
      "loss": 4.8151,
      "step": 11176
    },
    {
      "epoch": 1.53067652697891,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007519197045846317,
      "loss": 4.827,
      "step": 11177
    },
    {
      "epoch": 1.5308134757600658,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007518893267252373,
      "loss": 4.8273,
      "step": 11178
    },
    {
      "epoch": 1.5309504245412215,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007518589463640141,
      "loss": 4.8671,
      "step": 11179
    },
    {
      "epoch": 1.5310873733223773,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007518285635012137,
      "loss": 4.8841,
      "step": 11180
    },
    {
      "epoch": 1.5312243221035333,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007517981781370883,
      "loss": 4.7745,
      "step": 11181
    },
    {
      "epoch": 1.531361270884689,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007517677902718895,
      "loss": 4.8252,
      "step": 11182
    },
    {
      "epoch": 1.531498219665845,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007517373999058689,
      "loss": 4.8137,
      "step": 11183
    },
    {
      "epoch": 1.5316351684470009,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007517070070392787,
      "loss": 4.8869,
      "step": 11184
    },
    {
      "epoch": 1.5317721172281566,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007516766116723707,
      "loss": 4.7872,
      "step": 11185
    },
    {
      "epoch": 1.5319090660093124,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007516462138053965,
      "loss": 4.823,
      "step": 11186
    },
    {
      "epoch": 1.5320460147904682,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007516158134386084,
      "loss": 4.8817,
      "step": 11187
    },
    {
      "epoch": 1.5321829635716242,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007515854105722582,
      "loss": 4.8795,
      "step": 11188
    },
    {
      "epoch": 1.5323199123527802,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007515550052065978,
      "loss": 4.8903,
      "step": 11189
    },
    {
      "epoch": 1.532456861133936,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007515245973418792,
      "loss": 4.8163,
      "step": 11190
    },
    {
      "epoch": 1.5325938099150918,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007514941869783545,
      "loss": 4.8433,
      "step": 11191
    },
    {
      "epoch": 1.5327307586962475,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007514637741162756,
      "loss": 4.8331,
      "step": 11192
    },
    {
      "epoch": 1.5328677074774033,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007514333587558946,
      "loss": 4.8675,
      "step": 11193
    },
    {
      "epoch": 1.5330046562585593,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007514029408974634,
      "loss": 4.892,
      "step": 11194
    },
    {
      "epoch": 1.5331416050397153,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0007513725205412343,
      "loss": 4.8024,
      "step": 11195
    },
    {
      "epoch": 1.533278553820871,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007513420976874594,
      "loss": 4.8165,
      "step": 11196
    },
    {
      "epoch": 1.5334155026020269,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007513116723363905,
      "loss": 4.7902,
      "step": 11197
    },
    {
      "epoch": 1.5335524513831826,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007512812444882801,
      "loss": 4.8578,
      "step": 11198
    },
    {
      "epoch": 1.5336894001643384,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007512508141433802,
      "loss": 4.8796,
      "step": 11199
    },
    {
      "epoch": 1.5338263489454944,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007512203813019432,
      "loss": 4.8709,
      "step": 11200
    },
    {
      "epoch": 1.5339632977266502,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007511899459642209,
      "loss": 4.7712,
      "step": 11201
    },
    {
      "epoch": 1.5341002465078062,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007511595081304659,
      "loss": 4.8404,
      "step": 11202
    },
    {
      "epoch": 1.534237195288962,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007511290678009301,
      "loss": 4.8319,
      "step": 11203
    },
    {
      "epoch": 1.5343741440701177,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007510986249758662,
      "loss": 4.8182,
      "step": 11204
    },
    {
      "epoch": 1.5345110928512735,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000751068179655526,
      "loss": 4.8412,
      "step": 11205
    },
    {
      "epoch": 1.5346480416324295,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007510377318401622,
      "loss": 4.8234,
      "step": 11206
    },
    {
      "epoch": 1.5347849904135853,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000751007281530027,
      "loss": 4.7996,
      "step": 11207
    },
    {
      "epoch": 1.5349219391947413,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007509768287253726,
      "loss": 4.8328,
      "step": 11208
    },
    {
      "epoch": 1.535058887975897,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007509463734264517,
      "loss": 4.7445,
      "step": 11209
    },
    {
      "epoch": 1.5351958367570528,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007509159156335164,
      "loss": 4.859,
      "step": 11210
    },
    {
      "epoch": 1.5353327855382086,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007508854553468191,
      "loss": 4.8847,
      "step": 11211
    },
    {
      "epoch": 1.5354697343193644,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007508549925666125,
      "loss": 4.761,
      "step": 11212
    },
    {
      "epoch": 1.5356066831005204,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007508245272931489,
      "loss": 4.8623,
      "step": 11213
    },
    {
      "epoch": 1.5357436318816764,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007507940595266806,
      "loss": 4.9207,
      "step": 11214
    },
    {
      "epoch": 1.5358805806628322,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007507635892674603,
      "loss": 4.8635,
      "step": 11215
    },
    {
      "epoch": 1.536017529443988,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007507331165157406,
      "loss": 4.8305,
      "step": 11216
    },
    {
      "epoch": 1.5361544782251437,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007507026412717736,
      "loss": 4.9709,
      "step": 11217
    },
    {
      "epoch": 1.5362914270062995,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0007506721635358125,
      "loss": 4.7837,
      "step": 11218
    },
    {
      "epoch": 1.5364283757874555,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007506416833081094,
      "loss": 4.8146,
      "step": 11219
    },
    {
      "epoch": 1.5365653245686115,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000750611200588917,
      "loss": 4.9037,
      "step": 11220
    },
    {
      "epoch": 1.5367022733497673,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007505807153784879,
      "loss": 4.8087,
      "step": 11221
    },
    {
      "epoch": 1.536839222130923,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007505502276770749,
      "loss": 4.7666,
      "step": 11222
    },
    {
      "epoch": 1.5369761709120788,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007505197374849305,
      "loss": 4.8238,
      "step": 11223
    },
    {
      "epoch": 1.5371131196932346,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007504892448023073,
      "loss": 4.8103,
      "step": 11224
    },
    {
      "epoch": 1.5372500684743906,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007504587496294584,
      "loss": 4.8163,
      "step": 11225
    },
    {
      "epoch": 1.5373870172555464,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007504282519666361,
      "loss": 4.7963,
      "step": 11226
    },
    {
      "epoch": 1.5375239660367024,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007503977518140933,
      "loss": 4.866,
      "step": 11227
    },
    {
      "epoch": 1.5376609148178582,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007503672491720826,
      "loss": 4.7979,
      "step": 11228
    },
    {
      "epoch": 1.537797863599014,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007503367440408572,
      "loss": 4.8252,
      "step": 11229
    },
    {
      "epoch": 1.5379348123801697,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007503062364206696,
      "loss": 4.8535,
      "step": 11230
    },
    {
      "epoch": 1.5380717611613257,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007502757263117725,
      "loss": 4.8974,
      "step": 11231
    },
    {
      "epoch": 1.5382087099424815,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007502452137144189,
      "loss": 4.8435,
      "step": 11232
    },
    {
      "epoch": 1.5383456587236375,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007502146986288617,
      "loss": 4.7766,
      "step": 11233
    },
    {
      "epoch": 1.5384826075047933,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007501841810553539,
      "loss": 4.863,
      "step": 11234
    },
    {
      "epoch": 1.538619556285949,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007501536609941481,
      "loss": 4.8365,
      "step": 11235
    },
    {
      "epoch": 1.5387565050671048,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007501231384454975,
      "loss": 4.8818,
      "step": 11236
    },
    {
      "epoch": 1.5388934538482606,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007500926134096548,
      "loss": 4.8611,
      "step": 11237
    },
    {
      "epoch": 1.5390304026294166,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007500620858868734,
      "loss": 4.9053,
      "step": 11238
    },
    {
      "epoch": 1.5391673514105726,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007500315558774056,
      "loss": 4.8749,
      "step": 11239
    },
    {
      "epoch": 1.5393043001917284,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000750001023381505,
      "loss": 4.8734,
      "step": 11240
    },
    {
      "epoch": 1.5394412489728841,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007499704883994244,
      "loss": 4.9034,
      "step": 11241
    },
    {
      "epoch": 1.53957819775404,
      "grad_norm": 0.390625,
      "learning_rate": 0.000749939950931417,
      "loss": 4.9483,
      "step": 11242
    },
    {
      "epoch": 1.5397151465351957,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007499094109777356,
      "loss": 4.8378,
      "step": 11243
    },
    {
      "epoch": 1.5398520953163517,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007498788685386334,
      "loss": 4.8921,
      "step": 11244
    },
    {
      "epoch": 1.5399890440975077,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007498483236143637,
      "loss": 4.8846,
      "step": 11245
    },
    {
      "epoch": 1.5401259928786635,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007498177762051794,
      "loss": 4.8097,
      "step": 11246
    },
    {
      "epoch": 1.5402629416598193,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007497872263113338,
      "loss": 4.8344,
      "step": 11247
    },
    {
      "epoch": 1.540399890440975,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007497566739330801,
      "loss": 4.8967,
      "step": 11248
    },
    {
      "epoch": 1.5405368392221308,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007497261190706713,
      "loss": 4.8928,
      "step": 11249
    },
    {
      "epoch": 1.5406737880032868,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007496955617243607,
      "loss": 4.8622,
      "step": 11250
    },
    {
      "epoch": 1.5408107367844426,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007496650018944017,
      "loss": 4.8287,
      "step": 11251
    },
    {
      "epoch": 1.5409476855655986,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007496344395810473,
      "loss": 4.8711,
      "step": 11252
    },
    {
      "epoch": 1.5410846343467544,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007496038747845511,
      "loss": 4.848,
      "step": 11253
    },
    {
      "epoch": 1.5412215831279101,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007495733075051661,
      "loss": 4.7866,
      "step": 11254
    },
    {
      "epoch": 1.541358531909066,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007495427377431455,
      "loss": 4.77,
      "step": 11255
    },
    {
      "epoch": 1.541495480690222,
      "grad_norm": 0.375,
      "learning_rate": 0.000749512165498743,
      "loss": 4.7814,
      "step": 11256
    },
    {
      "epoch": 1.5416324294713777,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007494815907722118,
      "loss": 4.7887,
      "step": 11257
    },
    {
      "epoch": 1.5417693782525337,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007494510135638054,
      "loss": 4.7574,
      "step": 11258
    },
    {
      "epoch": 1.5419063270336895,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007494204338737771,
      "loss": 4.8118,
      "step": 11259
    },
    {
      "epoch": 1.5420432758148452,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007493898517023801,
      "loss": 4.8052,
      "step": 11260
    },
    {
      "epoch": 1.542180224596001,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007493592670498683,
      "loss": 4.7711,
      "step": 11261
    },
    {
      "epoch": 1.5423171733771568,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007493286799164948,
      "loss": 4.8658,
      "step": 11262
    },
    {
      "epoch": 1.5424541221583128,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007492980903025132,
      "loss": 4.7902,
      "step": 11263
    },
    {
      "epoch": 1.5425910709394688,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000749267498208177,
      "loss": 4.8125,
      "step": 11264
    },
    {
      "epoch": 1.5427280197206246,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007492369036337397,
      "loss": 4.771,
      "step": 11265
    },
    {
      "epoch": 1.5428649685017803,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007492063065794551,
      "loss": 4.8981,
      "step": 11266
    },
    {
      "epoch": 1.5430019172829361,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007491757070455764,
      "loss": 4.7127,
      "step": 11267
    },
    {
      "epoch": 1.543138866064092,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007491451050323573,
      "loss": 4.8865,
      "step": 11268
    },
    {
      "epoch": 1.543275814845248,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007491145005400513,
      "loss": 4.8182,
      "step": 11269
    },
    {
      "epoch": 1.5434127636264037,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007490838935689123,
      "loss": 4.8181,
      "step": 11270
    },
    {
      "epoch": 1.5435497124075597,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007490532841191938,
      "loss": 4.8174,
      "step": 11271
    },
    {
      "epoch": 1.5436866611887154,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007490226721911495,
      "loss": 4.8457,
      "step": 11272
    },
    {
      "epoch": 1.5438236099698712,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000748992057785033,
      "loss": 4.8586,
      "step": 11273
    },
    {
      "epoch": 1.543960558751027,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007489614409010981,
      "loss": 4.7225,
      "step": 11274
    },
    {
      "epoch": 1.544097507532183,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007489308215395987,
      "loss": 4.8807,
      "step": 11275
    },
    {
      "epoch": 1.5442344563133388,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000748900199700788,
      "loss": 4.8752,
      "step": 11276
    },
    {
      "epoch": 1.5443714050944948,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007488695753849204,
      "loss": 4.8548,
      "step": 11277
    },
    {
      "epoch": 1.5445083538756506,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007488389485922493,
      "loss": 4.8042,
      "step": 11278
    },
    {
      "epoch": 1.5446453026568063,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007488083193230285,
      "loss": 4.8332,
      "step": 11279
    },
    {
      "epoch": 1.544782251437962,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007487776875775121,
      "loss": 4.8629,
      "step": 11280
    },
    {
      "epoch": 1.544919200219118,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007487470533559538,
      "loss": 4.8097,
      "step": 11281
    },
    {
      "epoch": 1.5450561490002739,
      "grad_norm": 0.375,
      "learning_rate": 0.0007487164166586075,
      "loss": 4.8758,
      "step": 11282
    },
    {
      "epoch": 1.5451930977814299,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000748685777485727,
      "loss": 4.7598,
      "step": 11283
    },
    {
      "epoch": 1.5453300465625857,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007486551358375663,
      "loss": 4.9689,
      "step": 11284
    },
    {
      "epoch": 1.5454669953437414,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007486244917143793,
      "loss": 4.8263,
      "step": 11285
    },
    {
      "epoch": 1.5456039441248972,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007485938451164202,
      "loss": 4.8534,
      "step": 11286
    },
    {
      "epoch": 1.545740892906053,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007485631960439425,
      "loss": 4.7691,
      "step": 11287
    },
    {
      "epoch": 1.545877841687209,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0007485325444972005,
      "loss": 4.7853,
      "step": 11288
    },
    {
      "epoch": 1.546014790468365,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007485018904764483,
      "loss": 4.7076,
      "step": 11289
    },
    {
      "epoch": 1.5461517392495208,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007484712339819396,
      "loss": 4.7557,
      "step": 11290
    },
    {
      "epoch": 1.5462886880306765,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007484405750139288,
      "loss": 4.7995,
      "step": 11291
    },
    {
      "epoch": 1.5464256368118323,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007484099135726698,
      "loss": 4.9041,
      "step": 11292
    },
    {
      "epoch": 1.546562585592988,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007483792496584167,
      "loss": 4.8545,
      "step": 11293
    },
    {
      "epoch": 1.546699534374144,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007483485832714237,
      "loss": 4.8279,
      "step": 11294
    },
    {
      "epoch": 1.5468364831552999,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007483179144119449,
      "loss": 4.8713,
      "step": 11295
    },
    {
      "epoch": 1.5469734319364559,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007482872430802345,
      "loss": 4.8321,
      "step": 11296
    },
    {
      "epoch": 1.5471103807176116,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007482565692765466,
      "loss": 4.8523,
      "step": 11297
    },
    {
      "epoch": 1.5472473294987674,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007482258930011354,
      "loss": 4.9298,
      "step": 11298
    },
    {
      "epoch": 1.5473842782799232,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007481952142542552,
      "loss": 4.9241,
      "step": 11299
    },
    {
      "epoch": 1.5475212270610792,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007481645330361602,
      "loss": 4.8295,
      "step": 11300
    },
    {
      "epoch": 1.547658175842235,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007481338493471047,
      "loss": 4.8645,
      "step": 11301
    },
    {
      "epoch": 1.547795124623391,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007481031631873429,
      "loss": 4.7953,
      "step": 11302
    },
    {
      "epoch": 1.5479320734045467,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007480724745571293,
      "loss": 4.9326,
      "step": 11303
    },
    {
      "epoch": 1.5480690221857025,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007480417834567178,
      "loss": 4.8502,
      "step": 11304
    },
    {
      "epoch": 1.5482059709668583,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007480110898863632,
      "loss": 4.8206,
      "step": 11305
    },
    {
      "epoch": 1.548342919748014,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007479803938463196,
      "loss": 4.7812,
      "step": 11306
    },
    {
      "epoch": 1.54847986852917,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007479496953368415,
      "loss": 4.7561,
      "step": 11307
    },
    {
      "epoch": 1.548616817310326,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007479189943581834,
      "loss": 4.8375,
      "step": 11308
    },
    {
      "epoch": 1.5487537660914819,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007478882909105995,
      "loss": 4.7826,
      "step": 11309
    },
    {
      "epoch": 1.5488907148726376,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007478575849943444,
      "loss": 4.8931,
      "step": 11310
    },
    {
      "epoch": 1.5490276636537934,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007478268766096724,
      "loss": 4.8724,
      "step": 11311
    },
    {
      "epoch": 1.5491646124349492,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007477961657568382,
      "loss": 4.8292,
      "step": 11312
    },
    {
      "epoch": 1.5493015612161052,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007477654524360962,
      "loss": 4.7803,
      "step": 11313
    },
    {
      "epoch": 1.5494385099972612,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000747734736647701,
      "loss": 4.7306,
      "step": 11314
    },
    {
      "epoch": 1.549575458778417,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000747704018391907,
      "loss": 4.8861,
      "step": 11315
    },
    {
      "epoch": 1.5497124075595727,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007476732976689688,
      "loss": 4.7585,
      "step": 11316
    },
    {
      "epoch": 1.5498493563407285,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007476425744791412,
      "loss": 4.8018,
      "step": 11317
    },
    {
      "epoch": 1.5499863051218843,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007476118488226786,
      "loss": 4.8196,
      "step": 11318
    },
    {
      "epoch": 1.5501232539030403,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007475811206998356,
      "loss": 4.8716,
      "step": 11319
    },
    {
      "epoch": 1.550260202684196,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000747550390110867,
      "loss": 4.7011,
      "step": 11320
    },
    {
      "epoch": 1.550397151465352,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007475196570560276,
      "loss": 4.8682,
      "step": 11321
    },
    {
      "epoch": 1.5505341002465078,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007474889215355717,
      "loss": 4.8385,
      "step": 11322
    },
    {
      "epoch": 1.5506710490276636,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007474581835497544,
      "loss": 4.891,
      "step": 11323
    },
    {
      "epoch": 1.5508079978088194,
      "grad_norm": 0.390625,
      "learning_rate": 0.00074742744309883,
      "loss": 4.8009,
      "step": 11324
    },
    {
      "epoch": 1.5509449465899754,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007473967001830536,
      "loss": 4.813,
      "step": 11325
    },
    {
      "epoch": 1.5510818953711312,
      "grad_norm": 0.359375,
      "learning_rate": 0.00074736595480268,
      "loss": 4.7653,
      "step": 11326
    },
    {
      "epoch": 1.5512188441522872,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007473352069579637,
      "loss": 4.8601,
      "step": 11327
    },
    {
      "epoch": 1.551355792933443,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007473044566491599,
      "loss": 4.8606,
      "step": 11328
    },
    {
      "epoch": 1.5514927417145987,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000747273703876523,
      "loss": 4.8225,
      "step": 11329
    },
    {
      "epoch": 1.5516296904957545,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007472429486403082,
      "loss": 4.8514,
      "step": 11330
    },
    {
      "epoch": 1.5517666392769103,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007472121909407703,
      "loss": 4.8785,
      "step": 11331
    },
    {
      "epoch": 1.5519035880580663,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007471814307781641,
      "loss": 4.8278,
      "step": 11332
    },
    {
      "epoch": 1.5520405368392223,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007471506681527444,
      "loss": 4.7782,
      "step": 11333
    },
    {
      "epoch": 1.552177485620378,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007471199030647665,
      "loss": 4.8085,
      "step": 11334
    },
    {
      "epoch": 1.5523144344015338,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007470891355144852,
      "loss": 4.8366,
      "step": 11335
    },
    {
      "epoch": 1.5524513831826896,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0007470583655021553,
      "loss": 4.8585,
      "step": 11336
    },
    {
      "epoch": 1.5525883319638454,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007470275930280321,
      "loss": 4.7866,
      "step": 11337
    },
    {
      "epoch": 1.5527252807450014,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007469968180923703,
      "loss": 4.8879,
      "step": 11338
    },
    {
      "epoch": 1.5528622295261574,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007469660406954253,
      "loss": 4.9044,
      "step": 11339
    },
    {
      "epoch": 1.5529991783073132,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007469352608374517,
      "loss": 4.8182,
      "step": 11340
    },
    {
      "epoch": 1.553136127088469,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000746904478518705,
      "loss": 4.7888,
      "step": 11341
    },
    {
      "epoch": 1.5532730758696247,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007468736937394401,
      "loss": 4.8384,
      "step": 11342
    },
    {
      "epoch": 1.5534100246507805,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007468429064999121,
      "loss": 4.8095,
      "step": 11343
    },
    {
      "epoch": 1.5535469734319365,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007468121168003761,
      "loss": 4.861,
      "step": 11344
    },
    {
      "epoch": 1.5536839222130923,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007467813246410876,
      "loss": 4.8579,
      "step": 11345
    },
    {
      "epoch": 1.5538208709942483,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007467505300223016,
      "loss": 4.8516,
      "step": 11346
    },
    {
      "epoch": 1.553957819775404,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007467197329442731,
      "loss": 4.874,
      "step": 11347
    },
    {
      "epoch": 1.5540947685565598,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007466889334072575,
      "loss": 4.8049,
      "step": 11348
    },
    {
      "epoch": 1.5542317173377156,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007466581314115099,
      "loss": 4.9053,
      "step": 11349
    },
    {
      "epoch": 1.5543686661188716,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007466273269572858,
      "loss": 4.8832,
      "step": 11350
    },
    {
      "epoch": 1.5545056149000274,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007465965200448403,
      "loss": 4.8573,
      "step": 11351
    },
    {
      "epoch": 1.5546425636811834,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007465657106744287,
      "loss": 4.7661,
      "step": 11352
    },
    {
      "epoch": 1.5547795124623391,
      "grad_norm": 0.375,
      "learning_rate": 0.0007465348988463066,
      "loss": 4.8372,
      "step": 11353
    },
    {
      "epoch": 1.554916461243495,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007465040845607289,
      "loss": 4.9064,
      "step": 11354
    },
    {
      "epoch": 1.5550534100246507,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007464732678179514,
      "loss": 4.8108,
      "step": 11355
    },
    {
      "epoch": 1.5551903588058065,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007464424486182292,
      "loss": 4.8812,
      "step": 11356
    },
    {
      "epoch": 1.5553273075869625,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007464116269618176,
      "loss": 4.8463,
      "step": 11357
    },
    {
      "epoch": 1.5554642563681185,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007463808028489725,
      "loss": 4.8025,
      "step": 11358
    },
    {
      "epoch": 1.5556012051492742,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007463499762799489,
      "loss": 4.7938,
      "step": 11359
    },
    {
      "epoch": 1.55573815393043,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007463191472550025,
      "loss": 4.8539,
      "step": 11360
    },
    {
      "epoch": 1.5558751027115858,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007462883157743888,
      "loss": 4.8623,
      "step": 11361
    },
    {
      "epoch": 1.5560120514927416,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007462574818383631,
      "loss": 4.83,
      "step": 11362
    },
    {
      "epoch": 1.5561490002738976,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007462266454471811,
      "loss": 4.8333,
      "step": 11363
    },
    {
      "epoch": 1.5562859490550536,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007461958066010983,
      "loss": 4.7814,
      "step": 11364
    },
    {
      "epoch": 1.5564228978362094,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007461649653003702,
      "loss": 4.9113,
      "step": 11365
    },
    {
      "epoch": 1.5565598466173651,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007461341215452526,
      "loss": 4.7906,
      "step": 11366
    },
    {
      "epoch": 1.556696795398521,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007461032753360008,
      "loss": 4.7888,
      "step": 11367
    },
    {
      "epoch": 1.5568337441796767,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007460724266728707,
      "loss": 4.8523,
      "step": 11368
    },
    {
      "epoch": 1.5569706929608327,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007460415755561179,
      "loss": 4.8085,
      "step": 11369
    },
    {
      "epoch": 1.5571076417419885,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000746010721985998,
      "loss": 4.885,
      "step": 11370
    },
    {
      "epoch": 1.5572445905231445,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007459798659627666,
      "loss": 4.8059,
      "step": 11371
    },
    {
      "epoch": 1.5573815393043002,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007459490074866794,
      "loss": 4.7955,
      "step": 11372
    },
    {
      "epoch": 1.557518488085456,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007459181465579926,
      "loss": 4.8066,
      "step": 11373
    },
    {
      "epoch": 1.5576554368666118,
      "grad_norm": 0.375,
      "learning_rate": 0.0007458872831769615,
      "loss": 4.8107,
      "step": 11374
    },
    {
      "epoch": 1.5577923856477678,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007458564173438418,
      "loss": 4.798,
      "step": 11375
    },
    {
      "epoch": 1.5579293344289236,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007458255490588897,
      "loss": 4.7915,
      "step": 11376
    },
    {
      "epoch": 1.5580662832100796,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007457946783223607,
      "loss": 4.8009,
      "step": 11377
    },
    {
      "epoch": 1.5582032319912353,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007457638051345106,
      "loss": 4.8343,
      "step": 11378
    },
    {
      "epoch": 1.5583401807723911,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007457329294955955,
      "loss": 4.893,
      "step": 11379
    },
    {
      "epoch": 1.558477129553547,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007457020514058711,
      "loss": 4.8684,
      "step": 11380
    },
    {
      "epoch": 1.5586140783347027,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007456711708655934,
      "loss": 4.791,
      "step": 11381
    },
    {
      "epoch": 1.5587510271158587,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007456402878750182,
      "loss": 4.886,
      "step": 11382
    },
    {
      "epoch": 1.5588879758970147,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007456094024344014,
      "loss": 4.8096,
      "step": 11383
    },
    {
      "epoch": 1.5590249246781704,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007455785145439992,
      "loss": 4.8615,
      "step": 11384
    },
    {
      "epoch": 1.5591618734593262,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007455476242040674,
      "loss": 4.7985,
      "step": 11385
    },
    {
      "epoch": 1.559298822240482,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007455167314148619,
      "loss": 4.8317,
      "step": 11386
    },
    {
      "epoch": 1.5594357710216378,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007454858361766388,
      "loss": 4.8299,
      "step": 11387
    },
    {
      "epoch": 1.5595727198027938,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0007454549384896543,
      "loss": 4.8424,
      "step": 11388
    },
    {
      "epoch": 1.5597096685839495,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007454240383541643,
      "loss": 4.8331,
      "step": 11389
    },
    {
      "epoch": 1.5598466173651055,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007453931357704249,
      "loss": 4.8612,
      "step": 11390
    },
    {
      "epoch": 1.5599835661462613,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007453622307386922,
      "loss": 4.8273,
      "step": 11391
    },
    {
      "epoch": 1.560120514927417,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007453313232592222,
      "loss": 4.7202,
      "step": 11392
    },
    {
      "epoch": 1.5602574637085729,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007453004133322714,
      "loss": 4.8313,
      "step": 11393
    },
    {
      "epoch": 1.5603944124897289,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007452695009580955,
      "loss": 4.7828,
      "step": 11394
    },
    {
      "epoch": 1.5605313612708847,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007452385861369509,
      "loss": 4.8513,
      "step": 11395
    },
    {
      "epoch": 1.5606683100520407,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007452076688690938,
      "loss": 4.8479,
      "step": 11396
    },
    {
      "epoch": 1.5608052588331964,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007451767491547805,
      "loss": 4.8461,
      "step": 11397
    },
    {
      "epoch": 1.5609422076143522,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007451458269942669,
      "loss": 4.8518,
      "step": 11398
    },
    {
      "epoch": 1.561079156395508,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007451149023878097,
      "loss": 4.8723,
      "step": 11399
    },
    {
      "epoch": 1.561216105176664,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000745083975335665,
      "loss": 4.8258,
      "step": 11400
    },
    {
      "epoch": 1.5613530539578198,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000745053045838089,
      "loss": 4.7806,
      "step": 11401
    },
    {
      "epoch": 1.5614900027389758,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007450221138953383,
      "loss": 4.8497,
      "step": 11402
    },
    {
      "epoch": 1.5616269515201315,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007449911795076688,
      "loss": 4.7994,
      "step": 11403
    },
    {
      "epoch": 1.5617639003012873,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007449602426753371,
      "loss": 4.8289,
      "step": 11404
    },
    {
      "epoch": 1.561900849082443,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007449293033985997,
      "loss": 4.823,
      "step": 11405
    },
    {
      "epoch": 1.5620377978635989,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007448983616777129,
      "loss": 4.9021,
      "step": 11406
    },
    {
      "epoch": 1.5621747466447549,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007448674175129328,
      "loss": 4.8031,
      "step": 11407
    },
    {
      "epoch": 1.5623116954259109,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007448364709045166,
      "loss": 4.8191,
      "step": 11408
    },
    {
      "epoch": 1.5624486442070666,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007448055218527201,
      "loss": 4.8233,
      "step": 11409
    },
    {
      "epoch": 1.5625855929882224,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007447745703578001,
      "loss": 4.8491,
      "step": 11410
    },
    {
      "epoch": 1.5627225417693782,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007447436164200128,
      "loss": 4.8283,
      "step": 11411
    },
    {
      "epoch": 1.562859490550534,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000744712660039615,
      "loss": 4.8223,
      "step": 11412
    },
    {
      "epoch": 1.56299643933169,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007446817012168632,
      "loss": 4.8106,
      "step": 11413
    },
    {
      "epoch": 1.5631333881128457,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007446507399520139,
      "loss": 4.8018,
      "step": 11414
    },
    {
      "epoch": 1.5632703368940017,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007446197762453236,
      "loss": 4.8192,
      "step": 11415
    },
    {
      "epoch": 1.5634072856751575,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007445888100970491,
      "loss": 4.8103,
      "step": 11416
    },
    {
      "epoch": 1.5635442344563133,
      "grad_norm": 0.40625,
      "learning_rate": 0.000744557841507447,
      "loss": 4.8014,
      "step": 11417
    },
    {
      "epoch": 1.563681183237469,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007445268704767739,
      "loss": 4.817,
      "step": 11418
    },
    {
      "epoch": 1.563818132018625,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007444958970052863,
      "loss": 4.879,
      "step": 11419
    },
    {
      "epoch": 1.5639550807997809,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007444649210932411,
      "loss": 4.7616,
      "step": 11420
    },
    {
      "epoch": 1.5640920295809368,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000744433942740895,
      "loss": 4.8764,
      "step": 11421
    },
    {
      "epoch": 1.5642289783620926,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007444029619485046,
      "loss": 4.8145,
      "step": 11422
    },
    {
      "epoch": 1.5643659271432484,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007443719787163266,
      "loss": 4.7647,
      "step": 11423
    },
    {
      "epoch": 1.5645028759244042,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007443409930446179,
      "loss": 4.8133,
      "step": 11424
    },
    {
      "epoch": 1.56463982470556,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007443100049336353,
      "loss": 4.8059,
      "step": 11425
    },
    {
      "epoch": 1.564776773486716,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007442790143836357,
      "loss": 4.84,
      "step": 11426
    },
    {
      "epoch": 1.564913722267872,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007442480213948755,
      "loss": 4.743,
      "step": 11427
    },
    {
      "epoch": 1.5650506710490277,
      "grad_norm": 0.46875,
      "learning_rate": 0.0007442170259676121,
      "loss": 4.82,
      "step": 11428
    },
    {
      "epoch": 1.5651876198301835,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007441860281021019,
      "loss": 4.8052,
      "step": 11429
    },
    {
      "epoch": 1.5653245686113393,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000744155027798602,
      "loss": 4.8291,
      "step": 11430
    },
    {
      "epoch": 1.565461517392495,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007441240250573695,
      "loss": 4.7816,
      "step": 11431
    },
    {
      "epoch": 1.565598466173651,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007440930198786609,
      "loss": 4.8024,
      "step": 11432
    },
    {
      "epoch": 1.565735414954807,
      "grad_norm": 0.375,
      "learning_rate": 0.0007440620122627336,
      "loss": 4.8673,
      "step": 11433
    },
    {
      "epoch": 1.5658723637359628,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007440310022098441,
      "loss": 4.8382,
      "step": 11434
    },
    {
      "epoch": 1.5660093125171186,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007439999897202497,
      "loss": 4.7452,
      "step": 11435
    },
    {
      "epoch": 1.5661462612982744,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007439689747942073,
      "loss": 4.8252,
      "step": 11436
    },
    {
      "epoch": 1.5662832100794302,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007439379574319738,
      "loss": 4.862,
      "step": 11437
    },
    {
      "epoch": 1.5664201588605862,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007439069376338068,
      "loss": 4.826,
      "step": 11438
    },
    {
      "epoch": 1.566557107641742,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007438759153999626,
      "loss": 4.8082,
      "step": 11439
    },
    {
      "epoch": 1.566694056422898,
      "grad_norm": 0.421875,
      "learning_rate": 0.000743844890730699,
      "loss": 4.8082,
      "step": 11440
    },
    {
      "epoch": 1.5668310052040537,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007438138636262725,
      "loss": 4.8979,
      "step": 11441
    },
    {
      "epoch": 1.5669679539852095,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007437828340869407,
      "loss": 4.7469,
      "step": 11442
    },
    {
      "epoch": 1.5671049027663653,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007437518021129603,
      "loss": 4.9049,
      "step": 11443
    },
    {
      "epoch": 1.5672418515475213,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007437207677045888,
      "loss": 4.8106,
      "step": 11444
    },
    {
      "epoch": 1.567378800328677,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007436897308620834,
      "loss": 4.9199,
      "step": 11445
    },
    {
      "epoch": 1.567515749109833,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007436586915857013,
      "loss": 4.8187,
      "step": 11446
    },
    {
      "epoch": 1.5676526978909888,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007436276498756994,
      "loss": 4.8976,
      "step": 11447
    },
    {
      "epoch": 1.5677896466721446,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007435966057323354,
      "loss": 4.8658,
      "step": 11448
    },
    {
      "epoch": 1.5679265954533004,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007435655591558663,
      "loss": 4.9954,
      "step": 11449
    },
    {
      "epoch": 1.5680635442344562,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007435345101465495,
      "loss": 4.8157,
      "step": 11450
    },
    {
      "epoch": 1.5682004930156122,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007435034587046422,
      "loss": 4.8684,
      "step": 11451
    },
    {
      "epoch": 1.5683374417967682,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007434724048304017,
      "loss": 4.8828,
      "step": 11452
    },
    {
      "epoch": 1.568474390577924,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007434413485240857,
      "loss": 4.779,
      "step": 11453
    },
    {
      "epoch": 1.5686113393590797,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007434102897859511,
      "loss": 4.8644,
      "step": 11454
    },
    {
      "epoch": 1.5687482881402355,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007433792286162557,
      "loss": 4.6992,
      "step": 11455
    },
    {
      "epoch": 1.5688852369213913,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007433481650152566,
      "loss": 4.7833,
      "step": 11456
    },
    {
      "epoch": 1.5690221857025473,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007433170989832113,
      "loss": 4.8428,
      "step": 11457
    },
    {
      "epoch": 1.5691591344837033,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007432860305203775,
      "loss": 4.8453,
      "step": 11458
    },
    {
      "epoch": 1.569296083264859,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007432549596270123,
      "loss": 4.7764,
      "step": 11459
    },
    {
      "epoch": 1.5694330320460148,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007432238863033736,
      "loss": 4.9006,
      "step": 11460
    },
    {
      "epoch": 1.5695699808271706,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007431928105497184,
      "loss": 4.8566,
      "step": 11461
    },
    {
      "epoch": 1.5697069296083264,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007431617323663046,
      "loss": 4.8655,
      "step": 11462
    },
    {
      "epoch": 1.5698438783894824,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007431306517533898,
      "loss": 4.9032,
      "step": 11463
    },
    {
      "epoch": 1.5699808271706381,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007430995687112314,
      "loss": 4.9157,
      "step": 11464
    },
    {
      "epoch": 1.5701177759517941,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000743068483240087,
      "loss": 4.8382,
      "step": 11465
    },
    {
      "epoch": 1.57025472473295,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007430373953402141,
      "loss": 4.7859,
      "step": 11466
    },
    {
      "epoch": 1.5703916735141057,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007430063050118706,
      "loss": 4.877,
      "step": 11467
    },
    {
      "epoch": 1.5705286222952615,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007429752122553141,
      "loss": 4.7979,
      "step": 11468
    },
    {
      "epoch": 1.5706655710764175,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007429441170708021,
      "loss": 4.8602,
      "step": 11469
    },
    {
      "epoch": 1.5708025198575732,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007429130194585925,
      "loss": 4.9464,
      "step": 11470
    },
    {
      "epoch": 1.5709394686387292,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007428819194189428,
      "loss": 4.8803,
      "step": 11471
    },
    {
      "epoch": 1.571076417419885,
      "grad_norm": 0.359375,
      "learning_rate": 0.000742850816952111,
      "loss": 4.8278,
      "step": 11472
    },
    {
      "epoch": 1.5712133662010408,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007428197120583545,
      "loss": 4.8717,
      "step": 11473
    },
    {
      "epoch": 1.5713503149821966,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007427886047379314,
      "loss": 4.7953,
      "step": 11474
    },
    {
      "epoch": 1.5714872637633523,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007427574949910994,
      "loss": 4.84,
      "step": 11475
    },
    {
      "epoch": 1.5716242125445083,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007427263828181162,
      "loss": 4.8097,
      "step": 11476
    },
    {
      "epoch": 1.5717611613256643,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007426952682192396,
      "loss": 4.7469,
      "step": 11477
    },
    {
      "epoch": 1.5718981101068201,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007426641511947277,
      "loss": 4.7978,
      "step": 11478
    },
    {
      "epoch": 1.572035058887976,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007426330317448381,
      "loss": 4.869,
      "step": 11479
    },
    {
      "epoch": 1.5721720076691317,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007426019098698291,
      "loss": 4.7494,
      "step": 11480
    },
    {
      "epoch": 1.5723089564502875,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007425707855699582,
      "loss": 4.7676,
      "step": 11481
    },
    {
      "epoch": 1.5724459052314435,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007425396588454836,
      "loss": 4.8189,
      "step": 11482
    },
    {
      "epoch": 1.5725828540125995,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000742508529696663,
      "loss": 4.8034,
      "step": 11483
    },
    {
      "epoch": 1.5727198027937552,
      "grad_norm": 0.375,
      "learning_rate": 0.0007424773981237544,
      "loss": 4.879,
      "step": 11484
    },
    {
      "epoch": 1.572856751574911,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0007424462641270162,
      "loss": 4.7645,
      "step": 11485
    },
    {
      "epoch": 1.5729937003560668,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007424151277067058,
      "loss": 4.8339,
      "step": 11486
    },
    {
      "epoch": 1.5731306491372226,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007423839888630818,
      "loss": 4.8375,
      "step": 11487
    },
    {
      "epoch": 1.5732675979183786,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007423528475964021,
      "loss": 4.7909,
      "step": 11488
    },
    {
      "epoch": 1.5734045466995343,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007423217039069245,
      "loss": 4.8536,
      "step": 11489
    },
    {
      "epoch": 1.5735414954806903,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007422905577949072,
      "loss": 4.7984,
      "step": 11490
    },
    {
      "epoch": 1.573678444261846,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007422594092606086,
      "loss": 4.9229,
      "step": 11491
    },
    {
      "epoch": 1.5738153930430019,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007422282583042865,
      "loss": 4.8925,
      "step": 11492
    },
    {
      "epoch": 1.5739523418241577,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007421971049261992,
      "loss": 4.7372,
      "step": 11493
    },
    {
      "epoch": 1.5740892906053137,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007421659491266049,
      "loss": 4.8071,
      "step": 11494
    },
    {
      "epoch": 1.5742262393864694,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007421347909057619,
      "loss": 4.8298,
      "step": 11495
    },
    {
      "epoch": 1.5743631881676254,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007421036302639281,
      "loss": 4.8163,
      "step": 11496
    },
    {
      "epoch": 1.5745001369487812,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007420724672013619,
      "loss": 4.9117,
      "step": 11497
    },
    {
      "epoch": 1.574637085729937,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007420413017183217,
      "loss": 4.8769,
      "step": 11498
    },
    {
      "epoch": 1.5747740345110928,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007420101338150656,
      "loss": 4.8067,
      "step": 11499
    },
    {
      "epoch": 1.5749109832922485,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007419789634918519,
      "loss": 4.7027,
      "step": 11500
    },
    {
      "epoch": 1.5750479320734045,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007419477907489388,
      "loss": 4.8794,
      "step": 11501
    },
    {
      "epoch": 1.5751848808545605,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000741916615586585,
      "loss": 4.8418,
      "step": 11502
    },
    {
      "epoch": 1.5753218296357163,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007418854380050485,
      "loss": 4.8563,
      "step": 11503
    },
    {
      "epoch": 1.575458778416872,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000741854258004588,
      "loss": 4.867,
      "step": 11504
    },
    {
      "epoch": 1.5755957271980279,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007418230755854616,
      "loss": 4.8453,
      "step": 11505
    },
    {
      "epoch": 1.5757326759791837,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007417918907479279,
      "loss": 4.884,
      "step": 11506
    },
    {
      "epoch": 1.5758696247603396,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007417607034922452,
      "loss": 4.772,
      "step": 11507
    },
    {
      "epoch": 1.5760065735414954,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007417295138186721,
      "loss": 4.7501,
      "step": 11508
    },
    {
      "epoch": 1.5761435223226514,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000741698321727467,
      "loss": 4.7639,
      "step": 11509
    },
    {
      "epoch": 1.5762804711038072,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007416671272188883,
      "loss": 4.8041,
      "step": 11510
    },
    {
      "epoch": 1.576417419884963,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007416359302931947,
      "loss": 4.7802,
      "step": 11511
    },
    {
      "epoch": 1.5765543686661188,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007416047309506447,
      "loss": 4.7905,
      "step": 11512
    },
    {
      "epoch": 1.5766913174472748,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007415735291914966,
      "loss": 4.8036,
      "step": 11513
    },
    {
      "epoch": 1.5768282662284305,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007415423250160094,
      "loss": 4.8372,
      "step": 11514
    },
    {
      "epoch": 1.5769652150095865,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007415111184244415,
      "loss": 4.8124,
      "step": 11515
    },
    {
      "epoch": 1.5771021637907423,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007414799094170514,
      "loss": 4.9194,
      "step": 11516
    },
    {
      "epoch": 1.577239112571898,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007414486979940979,
      "loss": 4.7868,
      "step": 11517
    },
    {
      "epoch": 1.5773760613530539,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007414174841558394,
      "loss": 4.8629,
      "step": 11518
    },
    {
      "epoch": 1.5775130101342099,
      "grad_norm": 0.359375,
      "learning_rate": 0.000741386267902535,
      "loss": 4.7763,
      "step": 11519
    },
    {
      "epoch": 1.5776499589153656,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007413550492344431,
      "loss": 4.8195,
      "step": 11520
    },
    {
      "epoch": 1.5777869076965216,
      "grad_norm": 0.375,
      "learning_rate": 0.0007413238281518224,
      "loss": 4.7791,
      "step": 11521
    },
    {
      "epoch": 1.5779238564776774,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007412926046549318,
      "loss": 4.8546,
      "step": 11522
    },
    {
      "epoch": 1.5780608052588332,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00074126137874403,
      "loss": 4.8329,
      "step": 11523
    },
    {
      "epoch": 1.578197754039989,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007412301504193757,
      "loss": 4.8533,
      "step": 11524
    },
    {
      "epoch": 1.5783347028211447,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007411989196812276,
      "loss": 4.7972,
      "step": 11525
    },
    {
      "epoch": 1.5784716516023007,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007411676865298448,
      "loss": 4.8205,
      "step": 11526
    },
    {
      "epoch": 1.5786086003834567,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007411364509654862,
      "loss": 4.8561,
      "step": 11527
    },
    {
      "epoch": 1.5787455491646125,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007411052129884102,
      "loss": 4.7988,
      "step": 11528
    },
    {
      "epoch": 1.5788824979457683,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000741073972598876,
      "loss": 4.885,
      "step": 11529
    },
    {
      "epoch": 1.579019446726924,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007410427297971424,
      "loss": 4.9643,
      "step": 11530
    },
    {
      "epoch": 1.5791563955080798,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007410114845834684,
      "loss": 4.8254,
      "step": 11531
    },
    {
      "epoch": 1.5792933442892358,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007409802369581128,
      "loss": 4.8152,
      "step": 11532
    },
    {
      "epoch": 1.5794302930703916,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007409489869213346,
      "loss": 4.8211,
      "step": 11533
    },
    {
      "epoch": 1.5795672418515476,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000740917734473393,
      "loss": 4.7699,
      "step": 11534
    },
    {
      "epoch": 1.5797041906327034,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007408864796145467,
      "loss": 4.85,
      "step": 11535
    },
    {
      "epoch": 1.5798411394138592,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007408552223450549,
      "loss": 4.792,
      "step": 11536
    },
    {
      "epoch": 1.579978088195015,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007408239626651766,
      "loss": 4.7822,
      "step": 11537
    },
    {
      "epoch": 1.580115036976171,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007407927005751707,
      "loss": 4.8167,
      "step": 11538
    },
    {
      "epoch": 1.5802519857573267,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007407614360752964,
      "loss": 4.9194,
      "step": 11539
    },
    {
      "epoch": 1.5803889345384827,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007407301691658128,
      "loss": 4.8065,
      "step": 11540
    },
    {
      "epoch": 1.5805258833196385,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007406988998469791,
      "loss": 4.7744,
      "step": 11541
    },
    {
      "epoch": 1.5806628321007943,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007406676281190542,
      "loss": 4.8306,
      "step": 11542
    },
    {
      "epoch": 1.58079978088195,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007406363539822976,
      "loss": 4.8149,
      "step": 11543
    },
    {
      "epoch": 1.5809367296631058,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007406050774369682,
      "loss": 4.8211,
      "step": 11544
    },
    {
      "epoch": 1.5810736784442618,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007405737984833252,
      "loss": 4.8182,
      "step": 11545
    },
    {
      "epoch": 1.5812106272254178,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007405425171216279,
      "loss": 4.8687,
      "step": 11546
    },
    {
      "epoch": 1.5813475760065736,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007405112333521357,
      "loss": 4.6624,
      "step": 11547
    },
    {
      "epoch": 1.5814845247877294,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007404799471751074,
      "loss": 4.8309,
      "step": 11548
    },
    {
      "epoch": 1.5816214735688852,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007404486585908028,
      "loss": 4.8219,
      "step": 11549
    },
    {
      "epoch": 1.581758422350041,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000740417367599481,
      "loss": 4.7913,
      "step": 11550
    },
    {
      "epoch": 1.581895371131197,
      "grad_norm": 0.375,
      "learning_rate": 0.0007403860742014009,
      "loss": 4.8717,
      "step": 11551
    },
    {
      "epoch": 1.582032319912353,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007403547783968226,
      "loss": 4.921,
      "step": 11552
    },
    {
      "epoch": 1.5821692686935087,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007403234801860049,
      "loss": 4.8194,
      "step": 11553
    },
    {
      "epoch": 1.5823062174746645,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007402921795692074,
      "loss": 4.8397,
      "step": 11554
    },
    {
      "epoch": 1.5824431662558203,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007402608765466893,
      "loss": 4.7935,
      "step": 11555
    },
    {
      "epoch": 1.582580115036976,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007402295711187103,
      "loss": 4.8522,
      "step": 11556
    },
    {
      "epoch": 1.582717063818132,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007401982632855296,
      "loss": 4.8058,
      "step": 11557
    },
    {
      "epoch": 1.5828540125992878,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007401669530474067,
      "loss": 4.9158,
      "step": 11558
    },
    {
      "epoch": 1.5829909613804438,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007401356404046011,
      "loss": 4.8206,
      "step": 11559
    },
    {
      "epoch": 1.5831279101615996,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007401043253573724,
      "loss": 4.9218,
      "step": 11560
    },
    {
      "epoch": 1.5832648589427554,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007400730079059801,
      "loss": 4.8647,
      "step": 11561
    },
    {
      "epoch": 1.5834018077239111,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007400416880506837,
      "loss": 4.7679,
      "step": 11562
    },
    {
      "epoch": 1.5835387565050671,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007400103657917425,
      "loss": 4.8022,
      "step": 11563
    },
    {
      "epoch": 1.583675705286223,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007399790411294165,
      "loss": 4.9094,
      "step": 11564
    },
    {
      "epoch": 1.583812654067379,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000739947714063965,
      "loss": 4.7726,
      "step": 11565
    },
    {
      "epoch": 1.5839496028485347,
      "grad_norm": 0.375,
      "learning_rate": 0.0007399163845956476,
      "loss": 4.7418,
      "step": 11566
    },
    {
      "epoch": 1.5840865516296905,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007398850527247241,
      "loss": 4.8813,
      "step": 11567
    },
    {
      "epoch": 1.5842235004108463,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007398537184514541,
      "loss": 4.8807,
      "step": 11568
    },
    {
      "epoch": 1.584360449192002,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007398223817760973,
      "loss": 4.8027,
      "step": 11569
    },
    {
      "epoch": 1.584497397973158,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007397910426989134,
      "loss": 4.8453,
      "step": 11570
    },
    {
      "epoch": 1.584634346754314,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000739759701220162,
      "loss": 4.7915,
      "step": 11571
    },
    {
      "epoch": 1.5847712955354698,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000739728357340103,
      "loss": 4.9141,
      "step": 11572
    },
    {
      "epoch": 1.5849082443166256,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007396970110589958,
      "loss": 4.8362,
      "step": 11573
    },
    {
      "epoch": 1.5850451930977814,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007396656623771008,
      "loss": 4.806,
      "step": 11574
    },
    {
      "epoch": 1.5851821418789371,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007396343112946773,
      "loss": 4.837,
      "step": 11575
    },
    {
      "epoch": 1.5853190906600931,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007396029578119851,
      "loss": 4.83,
      "step": 11576
    },
    {
      "epoch": 1.5854560394412491,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007395716019292844,
      "loss": 4.895,
      "step": 11577
    },
    {
      "epoch": 1.585592988222405,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007395402436468346,
      "loss": 4.7651,
      "step": 11578
    },
    {
      "epoch": 1.5857299370035607,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007395088829648961,
      "loss": 4.8187,
      "step": 11579
    },
    {
      "epoch": 1.5858668857847165,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007394775198837282,
      "loss": 4.9054,
      "step": 11580
    },
    {
      "epoch": 1.5860038345658722,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007394461544035913,
      "loss": 4.7695,
      "step": 11581
    },
    {
      "epoch": 1.5861407833470282,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007394147865247452,
      "loss": 4.8176,
      "step": 11582
    },
    {
      "epoch": 1.586277732128184,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007393834162474496,
      "loss": 4.8,
      "step": 11583
    },
    {
      "epoch": 1.58641468090934,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007393520435719647,
      "loss": 4.8142,
      "step": 11584
    },
    {
      "epoch": 1.5865516296904958,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007393206684985506,
      "loss": 4.8457,
      "step": 11585
    },
    {
      "epoch": 1.5866885784716516,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007392892910274672,
      "loss": 4.833,
      "step": 11586
    },
    {
      "epoch": 1.5868255272528073,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007392579111589743,
      "loss": 4.7946,
      "step": 11587
    },
    {
      "epoch": 1.5869624760339633,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007392265288933324,
      "loss": 4.9285,
      "step": 11588
    },
    {
      "epoch": 1.5870994248151191,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007391951442308011,
      "loss": 4.8907,
      "step": 11589
    },
    {
      "epoch": 1.5872363735962751,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000739163757171641,
      "loss": 4.8842,
      "step": 11590
    },
    {
      "epoch": 1.587373322377431,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007391323677161117,
      "loss": 4.7878,
      "step": 11591
    },
    {
      "epoch": 1.5875102711585867,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007391009758644737,
      "loss": 4.8186,
      "step": 11592
    },
    {
      "epoch": 1.5876472199397424,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007390695816169871,
      "loss": 4.7051,
      "step": 11593
    },
    {
      "epoch": 1.5877841687208982,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007390381849739119,
      "loss": 4.8747,
      "step": 11594
    },
    {
      "epoch": 1.5879211175020542,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007390067859355084,
      "loss": 4.8122,
      "step": 11595
    },
    {
      "epoch": 1.5880580662832102,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007389753845020367,
      "loss": 4.9048,
      "step": 11596
    },
    {
      "epoch": 1.588195015064366,
      "grad_norm": 0.375,
      "learning_rate": 0.0007389439806737572,
      "loss": 4.8836,
      "step": 11597
    },
    {
      "epoch": 1.5883319638455218,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007389125744509301,
      "loss": 4.7357,
      "step": 11598
    },
    {
      "epoch": 1.5884689126266776,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007388811658338156,
      "loss": 4.7707,
      "step": 11599
    },
    {
      "epoch": 1.5886058614078333,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000738849754822674,
      "loss": 4.8921,
      "step": 11600
    },
    {
      "epoch": 1.5887428101889893,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007388183414177658,
      "loss": 4.7834,
      "step": 11601
    },
    {
      "epoch": 1.5888797589701453,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007387869256193511,
      "loss": 4.8602,
      "step": 11602
    },
    {
      "epoch": 1.589016707751301,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007387555074276904,
      "loss": 4.86,
      "step": 11603
    },
    {
      "epoch": 1.5891536565324569,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007387240868430439,
      "loss": 4.8812,
      "step": 11604
    },
    {
      "epoch": 1.5892906053136127,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007386926638656721,
      "loss": 4.8331,
      "step": 11605
    },
    {
      "epoch": 1.5894275540947684,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007386612384958354,
      "loss": 4.8517,
      "step": 11606
    },
    {
      "epoch": 1.5895645028759244,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007386298107337942,
      "loss": 4.7967,
      "step": 11607
    },
    {
      "epoch": 1.5897014516570802,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007385983805798092,
      "loss": 4.8517,
      "step": 11608
    },
    {
      "epoch": 1.5898384004382362,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007385669480341405,
      "loss": 4.7782,
      "step": 11609
    },
    {
      "epoch": 1.589975349219392,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007385355130970488,
      "loss": 4.7265,
      "step": 11610
    },
    {
      "epoch": 1.5901122980005478,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007385040757687944,
      "loss": 4.7432,
      "step": 11611
    },
    {
      "epoch": 1.5902492467817035,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007384726360496381,
      "loss": 4.7836,
      "step": 11612
    },
    {
      "epoch": 1.5903861955628595,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007384411939398403,
      "loss": 4.853,
      "step": 11613
    },
    {
      "epoch": 1.5905231443440153,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007384097494396617,
      "loss": 4.6797,
      "step": 11614
    },
    {
      "epoch": 1.5906600931251713,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007383783025493627,
      "loss": 4.8465,
      "step": 11615
    },
    {
      "epoch": 1.590797041906327,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007383468532692039,
      "loss": 4.8405,
      "step": 11616
    },
    {
      "epoch": 1.5909339906874829,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007383154015994461,
      "loss": 4.76,
      "step": 11617
    },
    {
      "epoch": 1.5910709394686386,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007382839475403497,
      "loss": 4.842,
      "step": 11618
    },
    {
      "epoch": 1.5912078882497944,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007382524910921758,
      "loss": 4.8671,
      "step": 11619
    },
    {
      "epoch": 1.5913448370309504,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007382210322551847,
      "loss": 4.7883,
      "step": 11620
    },
    {
      "epoch": 1.5914817858121064,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007381895710296373,
      "loss": 4.7837,
      "step": 11621
    },
    {
      "epoch": 1.5916187345932622,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007381581074157942,
      "loss": 4.8318,
      "step": 11622
    },
    {
      "epoch": 1.591755683374418,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007381266414139161,
      "loss": 4.8434,
      "step": 11623
    },
    {
      "epoch": 1.5918926321555738,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000738095173024264,
      "loss": 4.7559,
      "step": 11624
    },
    {
      "epoch": 1.5920295809367295,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007380637022470984,
      "loss": 4.785,
      "step": 11625
    },
    {
      "epoch": 1.5921665297178855,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007380322290826804,
      "loss": 4.7907,
      "step": 11626
    },
    {
      "epoch": 1.5923034784990415,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007380007535312706,
      "loss": 4.8548,
      "step": 11627
    },
    {
      "epoch": 1.5924404272801973,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007379692755931299,
      "loss": 4.8224,
      "step": 11628
    },
    {
      "epoch": 1.592577376061353,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007379377952685194,
      "loss": 4.8791,
      "step": 11629
    },
    {
      "epoch": 1.5927143248425089,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007379063125576995,
      "loss": 4.8904,
      "step": 11630
    },
    {
      "epoch": 1.5928512736236646,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007378748274609316,
      "loss": 4.7887,
      "step": 11631
    },
    {
      "epoch": 1.5929882224048206,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007378433399784764,
      "loss": 4.7609,
      "step": 11632
    },
    {
      "epoch": 1.5931251711859764,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007378118501105948,
      "loss": 4.8574,
      "step": 11633
    },
    {
      "epoch": 1.5932621199671324,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007377803578575477,
      "loss": 4.8299,
      "step": 11634
    },
    {
      "epoch": 1.5933990687482882,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007377488632195964,
      "loss": 4.769,
      "step": 11635
    },
    {
      "epoch": 1.593536017529444,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007377173661970015,
      "loss": 4.8142,
      "step": 11636
    },
    {
      "epoch": 1.5936729663105997,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007376858667900243,
      "loss": 4.8539,
      "step": 11637
    },
    {
      "epoch": 1.5938099150917557,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007376543649989259,
      "loss": 4.7921,
      "step": 11638
    },
    {
      "epoch": 1.5939468638729115,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007376228608239671,
      "loss": 4.8515,
      "step": 11639
    },
    {
      "epoch": 1.5940838126540675,
      "grad_norm": 0.375,
      "learning_rate": 0.0007375913542654091,
      "loss": 4.7922,
      "step": 11640
    },
    {
      "epoch": 1.5942207614352233,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007375598453235132,
      "loss": 4.8291,
      "step": 11641
    },
    {
      "epoch": 1.594357710216379,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007375283339985402,
      "loss": 4.9164,
      "step": 11642
    },
    {
      "epoch": 1.5944946589975348,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007374968202907513,
      "loss": 4.858,
      "step": 11643
    },
    {
      "epoch": 1.5946316077786906,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007374653042004078,
      "loss": 4.8204,
      "step": 11644
    },
    {
      "epoch": 1.5947685565598466,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007374337857277709,
      "loss": 4.8261,
      "step": 11645
    },
    {
      "epoch": 1.5949055053410026,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007374022648731017,
      "loss": 4.8734,
      "step": 11646
    },
    {
      "epoch": 1.5950424541221584,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007373707416366614,
      "loss": 4.7965,
      "step": 11647
    },
    {
      "epoch": 1.5951794029033142,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007373392160187114,
      "loss": 4.8269,
      "step": 11648
    },
    {
      "epoch": 1.59531635168447,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007373076880195127,
      "loss": 4.8151,
      "step": 11649
    },
    {
      "epoch": 1.5954533004656257,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007372761576393269,
      "loss": 4.8236,
      "step": 11650
    },
    {
      "epoch": 1.5955902492467817,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007372446248784149,
      "loss": 4.8587,
      "step": 11651
    },
    {
      "epoch": 1.5957271980279375,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007372130897370383,
      "loss": 4.8323,
      "step": 11652
    },
    {
      "epoch": 1.5958641468090935,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007371815522154585,
      "loss": 4.7878,
      "step": 11653
    },
    {
      "epoch": 1.5960010955902493,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007371500123139368,
      "loss": 4.806,
      "step": 11654
    },
    {
      "epoch": 1.596138044371405,
      "grad_norm": 0.328125,
      "learning_rate": 0.0007371184700327343,
      "loss": 4.8544,
      "step": 11655
    },
    {
      "epoch": 1.5962749931525608,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007370869253721128,
      "loss": 4.7956,
      "step": 11656
    },
    {
      "epoch": 1.5964119419337168,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007370553783323335,
      "loss": 4.8157,
      "step": 11657
    },
    {
      "epoch": 1.5965488907148726,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007370238289136578,
      "loss": 4.757,
      "step": 11658
    },
    {
      "epoch": 1.5966858394960286,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007369922771163473,
      "loss": 4.7906,
      "step": 11659
    },
    {
      "epoch": 1.5968227882771844,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007369607229406634,
      "loss": 4.9442,
      "step": 11660
    },
    {
      "epoch": 1.5969597370583402,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007369291663868677,
      "loss": 4.8468,
      "step": 11661
    },
    {
      "epoch": 1.597096685839496,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007368976074552216,
      "loss": 4.7442,
      "step": 11662
    },
    {
      "epoch": 1.5972336346206517,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007368660461459865,
      "loss": 4.6997,
      "step": 11663
    },
    {
      "epoch": 1.5973705834018077,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007368344824594243,
      "loss": 4.8684,
      "step": 11664
    },
    {
      "epoch": 1.5975075321829637,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007368029163957962,
      "loss": 4.8753,
      "step": 11665
    },
    {
      "epoch": 1.5976444809641195,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007367713479553643,
      "loss": 4.8407,
      "step": 11666
    },
    {
      "epoch": 1.5977814297452753,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007367397771383896,
      "loss": 4.7969,
      "step": 11667
    },
    {
      "epoch": 1.597918378526431,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007367082039451342,
      "loss": 4.8716,
      "step": 11668
    },
    {
      "epoch": 1.5980553273075868,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007366766283758595,
      "loss": 4.8265,
      "step": 11669
    },
    {
      "epoch": 1.5981922760887428,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007366450504308273,
      "loss": 4.796,
      "step": 11670
    },
    {
      "epoch": 1.5983292248698988,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0007366134701102994,
      "loss": 4.8509,
      "step": 11671
    },
    {
      "epoch": 1.5984661736510546,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007365818874145372,
      "loss": 4.8769,
      "step": 11672
    },
    {
      "epoch": 1.5986031224322104,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007365503023438028,
      "loss": 4.8077,
      "step": 11673
    },
    {
      "epoch": 1.5987400712133661,
      "grad_norm": 0.375,
      "learning_rate": 0.0007365187148983576,
      "loss": 4.8492,
      "step": 11674
    },
    {
      "epoch": 1.598877019994522,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007364871250784635,
      "loss": 4.7942,
      "step": 11675
    },
    {
      "epoch": 1.599013968775678,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007364555328843825,
      "loss": 4.8362,
      "step": 11676
    },
    {
      "epoch": 1.5991509175568337,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007364239383163761,
      "loss": 4.8449,
      "step": 11677
    },
    {
      "epoch": 1.5992878663379897,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007363923413747063,
      "loss": 4.8318,
      "step": 11678
    },
    {
      "epoch": 1.5994248151191455,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007363607420596349,
      "loss": 4.8198,
      "step": 11679
    },
    {
      "epoch": 1.5995617639003012,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000736329140371424,
      "loss": 4.8157,
      "step": 11680
    },
    {
      "epoch": 1.599698712681457,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000736297536310335,
      "loss": 4.838,
      "step": 11681
    },
    {
      "epoch": 1.599835661462613,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007362659298766303,
      "loss": 4.8431,
      "step": 11682
    },
    {
      "epoch": 1.5999726102437688,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007362343210705714,
      "loss": 4.7379,
      "step": 11683
    },
    {
      "epoch": 1.6001095590249248,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007362027098924207,
      "loss": 4.8145,
      "step": 11684
    },
    {
      "epoch": 1.6002465078060806,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007361710963424399,
      "loss": 4.8024,
      "step": 11685
    },
    {
      "epoch": 1.6003834565872364,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007361394804208909,
      "loss": 4.8012,
      "step": 11686
    },
    {
      "epoch": 1.6005204053683921,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000736107862128036,
      "loss": 4.843,
      "step": 11687
    },
    {
      "epoch": 1.600657354149548,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007360762414641369,
      "loss": 4.828,
      "step": 11688
    },
    {
      "epoch": 1.600794302930704,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000736044618429456,
      "loss": 4.7981,
      "step": 11689
    },
    {
      "epoch": 1.60093125171186,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007360129930242551,
      "loss": 4.7882,
      "step": 11690
    },
    {
      "epoch": 1.6010682004930157,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007359813652487964,
      "loss": 4.8872,
      "step": 11691
    },
    {
      "epoch": 1.6012051492741715,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000735949735103342,
      "loss": 4.8276,
      "step": 11692
    },
    {
      "epoch": 1.6013420980553272,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007359181025881539,
      "loss": 4.8317,
      "step": 11693
    },
    {
      "epoch": 1.601479046836483,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007358864677034945,
      "loss": 4.8466,
      "step": 11694
    },
    {
      "epoch": 1.601615995617639,
      "grad_norm": 0.375,
      "learning_rate": 0.0007358548304496258,
      "loss": 4.7909,
      "step": 11695
    },
    {
      "epoch": 1.601752944398795,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007358231908268099,
      "loss": 4.8598,
      "step": 11696
    },
    {
      "epoch": 1.6018898931799508,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007357915488353092,
      "loss": 4.8399,
      "step": 11697
    },
    {
      "epoch": 1.6020268419611066,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007357599044753858,
      "loss": 4.8434,
      "step": 11698
    },
    {
      "epoch": 1.6021637907422623,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007357282577473019,
      "loss": 4.8398,
      "step": 11699
    },
    {
      "epoch": 1.6023007395234181,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00073569660865132,
      "loss": 4.7395,
      "step": 11700
    },
    {
      "epoch": 1.6024376883045741,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007356649571877021,
      "loss": 4.8981,
      "step": 11701
    },
    {
      "epoch": 1.60257463708573,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007356333033567107,
      "loss": 4.8349,
      "step": 11702
    },
    {
      "epoch": 1.602711585866886,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000735601647158608,
      "loss": 4.9497,
      "step": 11703
    },
    {
      "epoch": 1.6028485346480417,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007355699885936564,
      "loss": 4.8728,
      "step": 11704
    },
    {
      "epoch": 1.6029854834291974,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007355383276621183,
      "loss": 4.797,
      "step": 11705
    },
    {
      "epoch": 1.6031224322103532,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007355066643642561,
      "loss": 4.8505,
      "step": 11706
    },
    {
      "epoch": 1.6032593809915092,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007354749987003319,
      "loss": 4.864,
      "step": 11707
    },
    {
      "epoch": 1.603396329772665,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007354433306706085,
      "loss": 4.8591,
      "step": 11708
    },
    {
      "epoch": 1.603533278553821,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007354116602753482,
      "loss": 4.8096,
      "step": 11709
    },
    {
      "epoch": 1.6036702273349768,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007353799875148136,
      "loss": 4.8214,
      "step": 11710
    },
    {
      "epoch": 1.6038071761161325,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007353483123892669,
      "loss": 4.8468,
      "step": 11711
    },
    {
      "epoch": 1.6039441248972883,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007353166348989707,
      "loss": 4.8275,
      "step": 11712
    },
    {
      "epoch": 1.604081073678444,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007352849550441877,
      "loss": 4.8189,
      "step": 11713
    },
    {
      "epoch": 1.6042180224596,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007352532728251803,
      "loss": 4.8254,
      "step": 11714
    },
    {
      "epoch": 1.604354971240756,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000735221588242211,
      "loss": 4.8598,
      "step": 11715
    },
    {
      "epoch": 1.6044919200219119,
      "grad_norm": 0.375,
      "learning_rate": 0.0007351899012955423,
      "loss": 4.8164,
      "step": 11716
    },
    {
      "epoch": 1.6046288688030677,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000735158211985437,
      "loss": 4.8185,
      "step": 11717
    },
    {
      "epoch": 1.6047658175842234,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007351265203121577,
      "loss": 4.895,
      "step": 11718
    },
    {
      "epoch": 1.6049027663653792,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007350948262759671,
      "loss": 4.7889,
      "step": 11719
    },
    {
      "epoch": 1.6050397151465352,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007350631298771275,
      "loss": 4.8872,
      "step": 11720
    },
    {
      "epoch": 1.6051766639276912,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007350314311159019,
      "loss": 4.8325,
      "step": 11721
    },
    {
      "epoch": 1.605313612708847,
      "grad_norm": 0.359375,
      "learning_rate": 0.000734999729992553,
      "loss": 4.802,
      "step": 11722
    },
    {
      "epoch": 1.6054505614900028,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007349680265073433,
      "loss": 4.8409,
      "step": 11723
    },
    {
      "epoch": 1.6055875102711585,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007349363206605357,
      "loss": 4.9125,
      "step": 11724
    },
    {
      "epoch": 1.6057244590523143,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007349046124523929,
      "loss": 4.7889,
      "step": 11725
    },
    {
      "epoch": 1.6058614078334703,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007348729018831778,
      "loss": 4.8073,
      "step": 11726
    },
    {
      "epoch": 1.605998356614626,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000734841188953153,
      "loss": 4.8742,
      "step": 11727
    },
    {
      "epoch": 1.606135305395782,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007348094736625815,
      "loss": 4.8354,
      "step": 11728
    },
    {
      "epoch": 1.6062722541769379,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007347777560117259,
      "loss": 4.8223,
      "step": 11729
    },
    {
      "epoch": 1.6064092029580936,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007347460360008492,
      "loss": 4.7975,
      "step": 11730
    },
    {
      "epoch": 1.6065461517392494,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007347143136302143,
      "loss": 4.8677,
      "step": 11731
    },
    {
      "epoch": 1.6066831005204054,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007346825889000839,
      "loss": 4.9189,
      "step": 11732
    },
    {
      "epoch": 1.6068200493015612,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007346508618107211,
      "loss": 4.8503,
      "step": 11733
    },
    {
      "epoch": 1.6069569980827172,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007346191323623889,
      "loss": 4.8492,
      "step": 11734
    },
    {
      "epoch": 1.607093946863873,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007345874005553499,
      "loss": 4.8526,
      "step": 11735
    },
    {
      "epoch": 1.6072308956450287,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007345556663898676,
      "loss": 4.8518,
      "step": 11736
    },
    {
      "epoch": 1.6073678444261845,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007345239298662045,
      "loss": 4.8358,
      "step": 11737
    },
    {
      "epoch": 1.6075047932073403,
      "grad_norm": 0.375,
      "learning_rate": 0.0007344921909846237,
      "loss": 4.8332,
      "step": 11738
    },
    {
      "epoch": 1.6076417419884963,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007344604497453885,
      "loss": 4.7851,
      "step": 11739
    },
    {
      "epoch": 1.6077786907696523,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007344287061487618,
      "loss": 4.8693,
      "step": 11740
    },
    {
      "epoch": 1.607915639550808,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007343969601950065,
      "loss": 4.841,
      "step": 11741
    },
    {
      "epoch": 1.6080525883319639,
      "grad_norm": 0.375,
      "learning_rate": 0.0007343652118843859,
      "loss": 4.8147,
      "step": 11742
    },
    {
      "epoch": 1.6081895371131196,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007343334612171631,
      "loss": 4.8185,
      "step": 11743
    },
    {
      "epoch": 1.6083264858942754,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007343017081936009,
      "loss": 4.8754,
      "step": 11744
    },
    {
      "epoch": 1.6084634346754314,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007342699528139629,
      "loss": 4.8215,
      "step": 11745
    },
    {
      "epoch": 1.6086003834565874,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007342381950785121,
      "loss": 4.8308,
      "step": 11746
    },
    {
      "epoch": 1.6087373322377432,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007342064349875116,
      "loss": 4.8735,
      "step": 11747
    },
    {
      "epoch": 1.608874281018899,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007341746725412245,
      "loss": 4.8362,
      "step": 11748
    },
    {
      "epoch": 1.6090112298000547,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007341429077399144,
      "loss": 4.8598,
      "step": 11749
    },
    {
      "epoch": 1.6091481785812105,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007341111405838441,
      "loss": 4.7719,
      "step": 11750
    },
    {
      "epoch": 1.6092851273623665,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007340793710732773,
      "loss": 4.7665,
      "step": 11751
    },
    {
      "epoch": 1.6094220761435223,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007340475992084769,
      "loss": 4.8965,
      "step": 11752
    },
    {
      "epoch": 1.6095590249246783,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007340158249897065,
      "loss": 4.8261,
      "step": 11753
    },
    {
      "epoch": 1.609695973705834,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007339840484172291,
      "loss": 4.8082,
      "step": 11754
    },
    {
      "epoch": 1.6098329224869898,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007339522694913082,
      "loss": 4.8791,
      "step": 11755
    },
    {
      "epoch": 1.6099698712681456,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007339204882122073,
      "loss": 4.7912,
      "step": 11756
    },
    {
      "epoch": 1.6101068200493016,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007338887045801897,
      "loss": 4.7731,
      "step": 11757
    },
    {
      "epoch": 1.6102437688304574,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007338569185955187,
      "loss": 4.8492,
      "step": 11758
    },
    {
      "epoch": 1.6103807176116134,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007338251302584579,
      "loss": 4.7584,
      "step": 11759
    },
    {
      "epoch": 1.6105176663927692,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007337933395692704,
      "loss": 4.8066,
      "step": 11760
    },
    {
      "epoch": 1.610654615173925,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007337615465282199,
      "loss": 4.815,
      "step": 11761
    },
    {
      "epoch": 1.6107915639550807,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00073372975113557,
      "loss": 4.8944,
      "step": 11762
    },
    {
      "epoch": 1.6109285127362365,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000733697953391584,
      "loss": 4.8393,
      "step": 11763
    },
    {
      "epoch": 1.6110654615173925,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007336661532965254,
      "loss": 4.777,
      "step": 11764
    },
    {
      "epoch": 1.6112024102985485,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007336343508506577,
      "loss": 4.772,
      "step": 11765
    },
    {
      "epoch": 1.6113393590797043,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007336025460542449,
      "loss": 4.864,
      "step": 11766
    },
    {
      "epoch": 1.61147630786086,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007335707389075499,
      "loss": 4.8133,
      "step": 11767
    },
    {
      "epoch": 1.6116132566420158,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007335389294108366,
      "loss": 4.9274,
      "step": 11768
    },
    {
      "epoch": 1.6117502054231716,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007335071175643688,
      "loss": 4.9402,
      "step": 11769
    },
    {
      "epoch": 1.6118871542043276,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007334753033684097,
      "loss": 4.8591,
      "step": 11770
    },
    {
      "epoch": 1.6120241029854834,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007334434868232236,
      "loss": 4.8929,
      "step": 11771
    },
    {
      "epoch": 1.6121610517666394,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007334116679290736,
      "loss": 4.8274,
      "step": 11772
    },
    {
      "epoch": 1.6122980005477952,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007333798466862235,
      "loss": 4.7708,
      "step": 11773
    },
    {
      "epoch": 1.612434949328951,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007333480230949371,
      "loss": 4.7762,
      "step": 11774
    },
    {
      "epoch": 1.6125718981101067,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007333161971554782,
      "loss": 4.8449,
      "step": 11775
    },
    {
      "epoch": 1.6127088468912627,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007332843688681104,
      "loss": 4.9238,
      "step": 11776
    },
    {
      "epoch": 1.6128457956724185,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007332525382330976,
      "loss": 4.8596,
      "step": 11777
    },
    {
      "epoch": 1.6129827444535745,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007332207052507035,
      "loss": 4.8527,
      "step": 11778
    },
    {
      "epoch": 1.6131196932347303,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007331888699211918,
      "loss": 4.7979,
      "step": 11779
    },
    {
      "epoch": 1.613256642015886,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007331570322448267,
      "loss": 4.8261,
      "step": 11780
    },
    {
      "epoch": 1.6133935907970418,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007331251922218715,
      "loss": 4.8555,
      "step": 11781
    },
    {
      "epoch": 1.6135305395781978,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007330933498525906,
      "loss": 4.7889,
      "step": 11782
    },
    {
      "epoch": 1.6136674883593536,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007330615051372475,
      "loss": 4.8179,
      "step": 11783
    },
    {
      "epoch": 1.6138044371405096,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007330296580761064,
      "loss": 4.8574,
      "step": 11784
    },
    {
      "epoch": 1.6139413859216654,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007329978086694311,
      "loss": 4.8932,
      "step": 11785
    },
    {
      "epoch": 1.6140783347028211,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007329659569174853,
      "loss": 4.854,
      "step": 11786
    },
    {
      "epoch": 1.614215283483977,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007329341028205333,
      "loss": 4.8658,
      "step": 11787
    },
    {
      "epoch": 1.6143522322651327,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000732902246378839,
      "loss": 4.7894,
      "step": 11788
    },
    {
      "epoch": 1.6144891810462887,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007328703875926665,
      "loss": 4.8905,
      "step": 11789
    },
    {
      "epoch": 1.6146261298274447,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007328385264622795,
      "loss": 4.7666,
      "step": 11790
    },
    {
      "epoch": 1.6147630786086005,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007328066629879422,
      "loss": 4.8469,
      "step": 11791
    },
    {
      "epoch": 1.6149000273897562,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007327747971699189,
      "loss": 4.7875,
      "step": 11792
    },
    {
      "epoch": 1.615036976170912,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007327429290084733,
      "loss": 4.8649,
      "step": 11793
    },
    {
      "epoch": 1.6151739249520678,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007327110585038698,
      "loss": 4.8517,
      "step": 11794
    },
    {
      "epoch": 1.6153108737332238,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007326791856563723,
      "loss": 4.8236,
      "step": 11795
    },
    {
      "epoch": 1.6154478225143796,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007326473104662451,
      "loss": 4.7823,
      "step": 11796
    },
    {
      "epoch": 1.6155847712955356,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0007326154329337522,
      "loss": 4.8398,
      "step": 11797
    },
    {
      "epoch": 1.6157217200766913,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000732583553059158,
      "loss": 4.7212,
      "step": 11798
    },
    {
      "epoch": 1.6158586688578471,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007325516708427264,
      "loss": 4.8142,
      "step": 11799
    },
    {
      "epoch": 1.615995617639003,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007325197862847218,
      "loss": 4.8401,
      "step": 11800
    },
    {
      "epoch": 1.616132566420159,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0007324878993854086,
      "loss": 4.8425,
      "step": 11801
    },
    {
      "epoch": 1.6162695152013147,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007324560101450506,
      "loss": 4.7825,
      "step": 11802
    },
    {
      "epoch": 1.6164064639824707,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0007324241185639126,
      "loss": 4.8175,
      "step": 11803
    },
    {
      "epoch": 1.6165434127636265,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007323922246422584,
      "loss": 4.7886,
      "step": 11804
    },
    {
      "epoch": 1.6166803615447822,
      "grad_norm": 0.5625,
      "learning_rate": 0.0007323603283803527,
      "loss": 4.8059,
      "step": 11805
    },
    {
      "epoch": 1.616817310325938,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007323284297784594,
      "loss": 4.8615,
      "step": 11806
    },
    {
      "epoch": 1.6169542591070938,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0007322965288368436,
      "loss": 4.7482,
      "step": 11807
    },
    {
      "epoch": 1.6170912078882498,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007322646255557688,
      "loss": 4.8491,
      "step": 11808
    },
    {
      "epoch": 1.6172281566694058,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0007322327199355001,
      "loss": 4.7417,
      "step": 11809
    },
    {
      "epoch": 1.6173651054505616,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007322008119763012,
      "loss": 4.7654,
      "step": 11810
    },
    {
      "epoch": 1.6175020542317173,
      "grad_norm": 0.46875,
      "learning_rate": 0.0007321689016784373,
      "loss": 4.8492,
      "step": 11811
    },
    {
      "epoch": 1.6176390030128731,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007321369890421723,
      "loss": 4.8344,
      "step": 11812
    },
    {
      "epoch": 1.6177759517940289,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0007321050740677708,
      "loss": 4.8986,
      "step": 11813
    },
    {
      "epoch": 1.6179129005751849,
      "grad_norm": 0.375,
      "learning_rate": 0.0007320731567554973,
      "loss": 4.8279,
      "step": 11814
    },
    {
      "epoch": 1.6180498493563409,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0007320412371056165,
      "loss": 4.8499,
      "step": 11815
    },
    {
      "epoch": 1.6181867981374967,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007320093151183927,
      "loss": 4.8217,
      "step": 11816
    },
    {
      "epoch": 1.6183237469186524,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007319773907940905,
      "loss": 4.8101,
      "step": 11817
    },
    {
      "epoch": 1.6184606956998082,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0007319454641329746,
      "loss": 4.814,
      "step": 11818
    },
    {
      "epoch": 1.618597644480964,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007319135351353092,
      "loss": 4.9107,
      "step": 11819
    },
    {
      "epoch": 1.61873459326212,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0007318816038013593,
      "loss": 4.8633,
      "step": 11820
    },
    {
      "epoch": 1.6188715420432758,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007318496701313894,
      "loss": 4.8352,
      "step": 11821
    },
    {
      "epoch": 1.6190084908244318,
      "grad_norm": 0.4609375,
      "learning_rate": 0.000731817734125664,
      "loss": 4.79,
      "step": 11822
    },
    {
      "epoch": 1.6191454396055875,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000731785795784448,
      "loss": 4.7527,
      "step": 11823
    },
    {
      "epoch": 1.6192823883867433,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007317538551080059,
      "loss": 4.8105,
      "step": 11824
    },
    {
      "epoch": 1.619419337167899,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007317219120966025,
      "loss": 4.9157,
      "step": 11825
    },
    {
      "epoch": 1.619556285949055,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007316899667505026,
      "loss": 4.8821,
      "step": 11826
    },
    {
      "epoch": 1.6196932347302109,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007316580190699707,
      "loss": 4.8133,
      "step": 11827
    },
    {
      "epoch": 1.6198301835113669,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007316260690552717,
      "loss": 4.8501,
      "step": 11828
    },
    {
      "epoch": 1.6199671322925226,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007315941167066702,
      "loss": 4.823,
      "step": 11829
    },
    {
      "epoch": 1.6201040810736784,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007315621620244314,
      "loss": 4.8467,
      "step": 11830
    },
    {
      "epoch": 1.6202410298548342,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007315302050088197,
      "loss": 4.8088,
      "step": 11831
    },
    {
      "epoch": 1.62037797863599,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007314982456601002,
      "loss": 4.7941,
      "step": 11832
    },
    {
      "epoch": 1.620514927417146,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007314662839785377,
      "loss": 4.6951,
      "step": 11833
    },
    {
      "epoch": 1.620651876198302,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000731434319964397,
      "loss": 4.7789,
      "step": 11834
    },
    {
      "epoch": 1.6207888249794578,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007314023536179429,
      "loss": 4.8098,
      "step": 11835
    },
    {
      "epoch": 1.6209257737606135,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007313703849394406,
      "loss": 4.7515,
      "step": 11836
    },
    {
      "epoch": 1.6210627225417693,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007313384139291547,
      "loss": 4.765,
      "step": 11837
    },
    {
      "epoch": 1.621199671322925,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007313064405873505,
      "loss": 4.756,
      "step": 11838
    },
    {
      "epoch": 1.621336620104081,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007312744649142928,
      "loss": 4.9134,
      "step": 11839
    },
    {
      "epoch": 1.621473568885237,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007312424869102465,
      "loss": 4.8468,
      "step": 11840
    },
    {
      "epoch": 1.6216105176663929,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007312105065754767,
      "loss": 4.8233,
      "step": 11841
    },
    {
      "epoch": 1.6217474664475486,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007311785239102484,
      "loss": 4.8653,
      "step": 11842
    },
    {
      "epoch": 1.6218844152287044,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007311465389148266,
      "loss": 4.8597,
      "step": 11843
    },
    {
      "epoch": 1.6220213640098602,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007311145515894766,
      "loss": 4.7639,
      "step": 11844
    },
    {
      "epoch": 1.6221583127910162,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007310825619344633,
      "loss": 4.8008,
      "step": 11845
    },
    {
      "epoch": 1.622295261572172,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007310505699500518,
      "loss": 4.8381,
      "step": 11846
    },
    {
      "epoch": 1.622432210353328,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007310185756365073,
      "loss": 4.8929,
      "step": 11847
    },
    {
      "epoch": 1.6225691591344837,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007309865789940948,
      "loss": 4.833,
      "step": 11848
    },
    {
      "epoch": 1.6227061079156395,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007309545800230796,
      "loss": 4.8507,
      "step": 11849
    },
    {
      "epoch": 1.6228430566967953,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000730922578723727,
      "loss": 4.8485,
      "step": 11850
    },
    {
      "epoch": 1.6229800054779513,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007308905750963018,
      "loss": 4.7709,
      "step": 11851
    },
    {
      "epoch": 1.623116954259107,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007308585691410697,
      "loss": 4.9096,
      "step": 11852
    },
    {
      "epoch": 1.623253903040263,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007308265608582955,
      "loss": 4.8714,
      "step": 11853
    },
    {
      "epoch": 1.6233908518214188,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007307945502482448,
      "loss": 4.7496,
      "step": 11854
    },
    {
      "epoch": 1.6235278006025746,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007307625373111828,
      "loss": 4.8021,
      "step": 11855
    },
    {
      "epoch": 1.6236647493837304,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007307305220473746,
      "loss": 4.8109,
      "step": 11856
    },
    {
      "epoch": 1.6238016981648862,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007306985044570858,
      "loss": 4.8711,
      "step": 11857
    },
    {
      "epoch": 1.6239386469460422,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007306664845405815,
      "loss": 4.8313,
      "step": 11858
    },
    {
      "epoch": 1.6240755957271982,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007306344622981273,
      "loss": 4.8503,
      "step": 11859
    },
    {
      "epoch": 1.624212544508354,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007306024377299883,
      "loss": 4.7928,
      "step": 11860
    },
    {
      "epoch": 1.6243494932895097,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007305704108364301,
      "loss": 4.838,
      "step": 11861
    },
    {
      "epoch": 1.6244864420706655,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007305383816177181,
      "loss": 4.8881,
      "step": 11862
    },
    {
      "epoch": 1.6246233908518213,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007305063500741176,
      "loss": 4.8211,
      "step": 11863
    },
    {
      "epoch": 1.6247603396329773,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007304743162058942,
      "loss": 4.8315,
      "step": 11864
    },
    {
      "epoch": 1.6248972884141333,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007304422800133132,
      "loss": 4.8665,
      "step": 11865
    },
    {
      "epoch": 1.625034237195289,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007304102414966403,
      "loss": 4.7496,
      "step": 11866
    },
    {
      "epoch": 1.6251711859764448,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007303782006561407,
      "loss": 4.85,
      "step": 11867
    },
    {
      "epoch": 1.6253081347576006,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007303461574920805,
      "loss": 4.8718,
      "step": 11868
    },
    {
      "epoch": 1.6254450835387564,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007303141120047246,
      "loss": 4.8716,
      "step": 11869
    },
    {
      "epoch": 1.6255820323199124,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000730282064194339,
      "loss": 4.8126,
      "step": 11870
    },
    {
      "epoch": 1.6257189811010682,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007302500140611892,
      "loss": 4.8836,
      "step": 11871
    },
    {
      "epoch": 1.6258559298822242,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007302179616055405,
      "loss": 4.7821,
      "step": 11872
    },
    {
      "epoch": 1.62599287866338,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000730185906827659,
      "loss": 4.7487,
      "step": 11873
    },
    {
      "epoch": 1.6261298274445357,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007301538497278101,
      "loss": 4.7716,
      "step": 11874
    },
    {
      "epoch": 1.6262667762256915,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007301217903062594,
      "loss": 4.8212,
      "step": 11875
    },
    {
      "epoch": 1.6264037250068475,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007300897285632729,
      "loss": 4.7865,
      "step": 11876
    },
    {
      "epoch": 1.6265406737880033,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007300576644991159,
      "loss": 4.8216,
      "step": 11877
    },
    {
      "epoch": 1.6266776225691593,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007300255981140542,
      "loss": 4.7666,
      "step": 11878
    },
    {
      "epoch": 1.626814571350315,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007299935294083539,
      "loss": 4.7472,
      "step": 11879
    },
    {
      "epoch": 1.6269515201314708,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007299614583822803,
      "loss": 4.8715,
      "step": 11880
    },
    {
      "epoch": 1.6270884689126266,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007299293850360994,
      "loss": 4.8125,
      "step": 11881
    },
    {
      "epoch": 1.6272254176937824,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000729897309370077,
      "loss": 4.8776,
      "step": 11882
    },
    {
      "epoch": 1.6273623664749384,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007298652313844791,
      "loss": 4.8161,
      "step": 11883
    },
    {
      "epoch": 1.6274993152560944,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000729833151079571,
      "loss": 4.818,
      "step": 11884
    },
    {
      "epoch": 1.6276362640372501,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007298010684556191,
      "loss": 4.7966,
      "step": 11885
    },
    {
      "epoch": 1.627773212818406,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000729768983512889,
      "loss": 4.8873,
      "step": 11886
    },
    {
      "epoch": 1.6279101615995617,
      "grad_norm": 0.375,
      "learning_rate": 0.0007297368962516467,
      "loss": 4.7607,
      "step": 11887
    },
    {
      "epoch": 1.6280471103807175,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007297048066721581,
      "loss": 4.7292,
      "step": 11888
    },
    {
      "epoch": 1.6281840591618735,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007296727147746891,
      "loss": 4.8397,
      "step": 11889
    },
    {
      "epoch": 1.6283210079430293,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007296406205595058,
      "loss": 4.8727,
      "step": 11890
    },
    {
      "epoch": 1.6284579567241853,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007296085240268738,
      "loss": 4.7794,
      "step": 11891
    },
    {
      "epoch": 1.628594905505341,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007295764251770596,
      "loss": 4.7623,
      "step": 11892
    },
    {
      "epoch": 1.6287318542864968,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007295443240103288,
      "loss": 4.8594,
      "step": 11893
    },
    {
      "epoch": 1.6288688030676526,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007295122205269475,
      "loss": 4.8629,
      "step": 11894
    },
    {
      "epoch": 1.6290057518488086,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000729480114727182,
      "loss": 4.9039,
      "step": 11895
    },
    {
      "epoch": 1.6291427006299644,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007294480066112978,
      "loss": 4.896,
      "step": 11896
    },
    {
      "epoch": 1.6292796494111204,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000729415896179562,
      "loss": 4.8926,
      "step": 11897
    },
    {
      "epoch": 1.6294165981922761,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007293837834322397,
      "loss": 4.76,
      "step": 11898
    },
    {
      "epoch": 1.629553546973432,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007293516683695975,
      "loss": 4.8486,
      "step": 11899
    },
    {
      "epoch": 1.6296904957545877,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007293195509919013,
      "loss": 4.8797,
      "step": 11900
    },
    {
      "epoch": 1.6298274445357437,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007292874312994176,
      "loss": 4.7733,
      "step": 11901
    },
    {
      "epoch": 1.6299643933168995,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007292553092924123,
      "loss": 4.8428,
      "step": 11902
    },
    {
      "epoch": 1.6301013420980555,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007292231849711517,
      "loss": 4.8558,
      "step": 11903
    },
    {
      "epoch": 1.6302382908792112,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007291910583359022,
      "loss": 4.85,
      "step": 11904
    },
    {
      "epoch": 1.630375239660367,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007291589293869297,
      "loss": 4.8138,
      "step": 11905
    },
    {
      "epoch": 1.6305121884415228,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007291267981245007,
      "loss": 4.7664,
      "step": 11906
    },
    {
      "epoch": 1.6306491372226786,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007290946645488813,
      "loss": 4.7675,
      "step": 11907
    },
    {
      "epoch": 1.6307860860038346,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007290625286603381,
      "loss": 4.7931,
      "step": 11908
    },
    {
      "epoch": 1.6309230347849906,
      "grad_norm": 0.375,
      "learning_rate": 0.0007290303904591372,
      "loss": 4.7944,
      "step": 11909
    },
    {
      "epoch": 1.6310599835661463,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000728998249945545,
      "loss": 4.8781,
      "step": 11910
    },
    {
      "epoch": 1.6311969323473021,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007289661071198276,
      "loss": 4.8438,
      "step": 11911
    },
    {
      "epoch": 1.631333881128458,
      "grad_norm": 0.375,
      "learning_rate": 0.0007289339619822519,
      "loss": 4.8094,
      "step": 11912
    },
    {
      "epoch": 1.6314708299096137,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007289018145330838,
      "loss": 4.7946,
      "step": 11913
    },
    {
      "epoch": 1.6316077786907697,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007288696647725899,
      "loss": 4.9409,
      "step": 11914
    },
    {
      "epoch": 1.6317447274719254,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007288375127010365,
      "loss": 4.7822,
      "step": 11915
    },
    {
      "epoch": 1.6318816762530814,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007288053583186905,
      "loss": 4.8329,
      "step": 11916
    },
    {
      "epoch": 1.6320186250342372,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000728773201625818,
      "loss": 4.8152,
      "step": 11917
    },
    {
      "epoch": 1.632155573815393,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007287410426226853,
      "loss": 4.8092,
      "step": 11918
    },
    {
      "epoch": 1.6322925225965488,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007287088813095594,
      "loss": 4.8707,
      "step": 11919
    },
    {
      "epoch": 1.6324294713777048,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0007286767176867065,
      "loss": 4.8647,
      "step": 11920
    },
    {
      "epoch": 1.6325664201588606,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007286445517543932,
      "loss": 4.8093,
      "step": 11921
    },
    {
      "epoch": 1.6327033689400166,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007286123835128863,
      "loss": 4.7433,
      "step": 11922
    },
    {
      "epoch": 1.6328403177211723,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000728580212962452,
      "loss": 4.8674,
      "step": 11923
    },
    {
      "epoch": 1.632977266502328,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007285480401033571,
      "loss": 4.8515,
      "step": 11924
    },
    {
      "epoch": 1.6331142152834839,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007285158649358682,
      "loss": 4.8383,
      "step": 11925
    },
    {
      "epoch": 1.6332511640646397,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007284836874602521,
      "loss": 4.8221,
      "step": 11926
    },
    {
      "epoch": 1.6333881128457957,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007284515076767752,
      "loss": 4.7672,
      "step": 11927
    },
    {
      "epoch": 1.6335250616269517,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007284193255857045,
      "loss": 4.7942,
      "step": 11928
    },
    {
      "epoch": 1.6336620104081074,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0007283871411873061,
      "loss": 4.8012,
      "step": 11929
    },
    {
      "epoch": 1.6337989591892632,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007283549544818475,
      "loss": 4.8261,
      "step": 11930
    },
    {
      "epoch": 1.633935907970419,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007283227654695949,
      "loss": 4.7791,
      "step": 11931
    },
    {
      "epoch": 1.6340728567515748,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007282905741508151,
      "loss": 4.8148,
      "step": 11932
    },
    {
      "epoch": 1.6342098055327308,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007282583805257752,
      "loss": 4.8202,
      "step": 11933
    },
    {
      "epoch": 1.6343467543138868,
      "grad_norm": 0.375,
      "learning_rate": 0.0007282261845947418,
      "loss": 4.8034,
      "step": 11934
    },
    {
      "epoch": 1.6344837030950425,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007281939863579816,
      "loss": 4.7995,
      "step": 11935
    },
    {
      "epoch": 1.6346206518761983,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007281617858157616,
      "loss": 4.884,
      "step": 11936
    },
    {
      "epoch": 1.634757600657354,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007281295829683487,
      "loss": 4.8499,
      "step": 11937
    },
    {
      "epoch": 1.6348945494385099,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007280973778160096,
      "loss": 4.8453,
      "step": 11938
    },
    {
      "epoch": 1.6350314982196659,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007280651703590111,
      "loss": 4.7975,
      "step": 11939
    },
    {
      "epoch": 1.6351684470008216,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007280329605976205,
      "loss": 4.7638,
      "step": 11940
    },
    {
      "epoch": 1.6353053957819776,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007280007485321044,
      "loss": 4.8172,
      "step": 11941
    },
    {
      "epoch": 1.6354423445631334,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00072796853416273,
      "loss": 4.7867,
      "step": 11942
    },
    {
      "epoch": 1.6355792933442892,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007279363174897639,
      "loss": 4.8301,
      "step": 11943
    },
    {
      "epoch": 1.635716242125445,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007279040985134734,
      "loss": 4.8153,
      "step": 11944
    },
    {
      "epoch": 1.635853190906601,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007278718772341253,
      "loss": 4.8716,
      "step": 11945
    },
    {
      "epoch": 1.6359901396877568,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007278396536519869,
      "loss": 4.7988,
      "step": 11946
    },
    {
      "epoch": 1.6361270884689127,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007278074277673249,
      "loss": 4.7742,
      "step": 11947
    },
    {
      "epoch": 1.6362640372500685,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007277751995804068,
      "loss": 4.8541,
      "step": 11948
    },
    {
      "epoch": 1.6364009860312243,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007277429690914993,
      "loss": 4.6953,
      "step": 11949
    },
    {
      "epoch": 1.63653793481238,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007277107363008697,
      "loss": 4.7861,
      "step": 11950
    },
    {
      "epoch": 1.6366748835935359,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000727678501208785,
      "loss": 4.8687,
      "step": 11951
    },
    {
      "epoch": 1.6368118323746919,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007276462638155124,
      "loss": 4.7926,
      "step": 11952
    },
    {
      "epoch": 1.6369487811558479,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007276140241213191,
      "loss": 4.8503,
      "step": 11953
    },
    {
      "epoch": 1.6370857299370036,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007275817821264721,
      "loss": 4.7666,
      "step": 11954
    },
    {
      "epoch": 1.6372226787181594,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007275495378312388,
      "loss": 4.8742,
      "step": 11955
    },
    {
      "epoch": 1.6373596274993152,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007275172912358863,
      "loss": 4.7793,
      "step": 11956
    },
    {
      "epoch": 1.637496576280471,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000727485042340682,
      "loss": 4.8539,
      "step": 11957
    },
    {
      "epoch": 1.637633525061627,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007274527911458929,
      "loss": 4.8122,
      "step": 11958
    },
    {
      "epoch": 1.637770473842783,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007274205376517865,
      "loss": 4.8908,
      "step": 11959
    },
    {
      "epoch": 1.6379074226239387,
      "grad_norm": 0.375,
      "learning_rate": 0.00072738828185863,
      "loss": 4.7559,
      "step": 11960
    },
    {
      "epoch": 1.6380443714050945,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007273560237666906,
      "loss": 4.8056,
      "step": 11961
    },
    {
      "epoch": 1.6381813201862503,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007273237633762359,
      "loss": 4.8472,
      "step": 11962
    },
    {
      "epoch": 1.638318268967406,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007272915006875329,
      "loss": 4.796,
      "step": 11963
    },
    {
      "epoch": 1.638455217748562,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007272592357008493,
      "loss": 4.8318,
      "step": 11964
    },
    {
      "epoch": 1.6385921665297178,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0007272269684164524,
      "loss": 4.7771,
      "step": 11965
    },
    {
      "epoch": 1.6387291153108738,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007271946988346093,
      "loss": 4.8647,
      "step": 11966
    },
    {
      "epoch": 1.6388660640920296,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007271624269555879,
      "loss": 4.8668,
      "step": 11967
    },
    {
      "epoch": 1.6390030128731854,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007271301527796553,
      "loss": 4.8858,
      "step": 11968
    },
    {
      "epoch": 1.6391399616543412,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007270978763070792,
      "loss": 4.7904,
      "step": 11969
    },
    {
      "epoch": 1.6392769104354972,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007270655975381269,
      "loss": 4.7937,
      "step": 11970
    },
    {
      "epoch": 1.639413859216653,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000727033316473066,
      "loss": 4.7078,
      "step": 11971
    },
    {
      "epoch": 1.639550807997809,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007270010331121641,
      "loss": 4.7637,
      "step": 11972
    },
    {
      "epoch": 1.6396877567789647,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0007269687474556884,
      "loss": 4.7757,
      "step": 11973
    },
    {
      "epoch": 1.6398247055601205,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007269364595039069,
      "loss": 4.8489,
      "step": 11974
    },
    {
      "epoch": 1.6399616543412763,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0007269041692570869,
      "loss": 4.8224,
      "step": 11975
    },
    {
      "epoch": 1.640098603122432,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000726871876715496,
      "loss": 4.7148,
      "step": 11976
    },
    {
      "epoch": 1.640235551903588,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007268395818794019,
      "loss": 4.8077,
      "step": 11977
    },
    {
      "epoch": 1.640372500684744,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007268072847490724,
      "loss": 4.8163,
      "step": 11978
    },
    {
      "epoch": 1.6405094494658998,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007267749853247747,
      "loss": 4.8252,
      "step": 11979
    },
    {
      "epoch": 1.6406463982470556,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007267426836067769,
      "loss": 4.8092,
      "step": 11980
    },
    {
      "epoch": 1.6407833470282114,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007267103795953466,
      "loss": 4.8601,
      "step": 11981
    },
    {
      "epoch": 1.6409202958093672,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007266780732907513,
      "loss": 4.8128,
      "step": 11982
    },
    {
      "epoch": 1.6410572445905232,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007266457646932588,
      "loss": 4.8441,
      "step": 11983
    },
    {
      "epoch": 1.6411941933716792,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007266134538031372,
      "loss": 4.8212,
      "step": 11984
    },
    {
      "epoch": 1.641331142152835,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007265811406206538,
      "loss": 4.7626,
      "step": 11985
    },
    {
      "epoch": 1.6414680909339907,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007265488251460766,
      "loss": 4.7884,
      "step": 11986
    },
    {
      "epoch": 1.6416050397151465,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0007265165073796734,
      "loss": 4.8174,
      "step": 11987
    },
    {
      "epoch": 1.6417419884963023,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007264841873217121,
      "loss": 4.8016,
      "step": 11988
    },
    {
      "epoch": 1.6418789372774583,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007264518649724603,
      "loss": 4.7825,
      "step": 11989
    },
    {
      "epoch": 1.642015886058614,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007264195403321861,
      "loss": 4.8409,
      "step": 11990
    },
    {
      "epoch": 1.64215283483977,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007263872134011574,
      "loss": 4.8325,
      "step": 11991
    },
    {
      "epoch": 1.6422897836209258,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000726354884179642,
      "loss": 4.7352,
      "step": 11992
    },
    {
      "epoch": 1.6424267324020816,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007263225526679077,
      "loss": 4.8059,
      "step": 11993
    },
    {
      "epoch": 1.6425636811832374,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007262902188662225,
      "loss": 4.7764,
      "step": 11994
    },
    {
      "epoch": 1.6427006299643934,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007262578827748545,
      "loss": 4.8615,
      "step": 11995
    },
    {
      "epoch": 1.6428375787455491,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007262255443940716,
      "loss": 4.8693,
      "step": 11996
    },
    {
      "epoch": 1.6429745275267051,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007261932037241419,
      "loss": 4.878,
      "step": 11997
    },
    {
      "epoch": 1.643111476307861,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007261608607653333,
      "loss": 4.899,
      "step": 11998
    },
    {
      "epoch": 1.6432484250890167,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007261285155179137,
      "loss": 4.8756,
      "step": 11999
    },
    {
      "epoch": 1.6433853738701725,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007260961679821514,
      "loss": 4.7856,
      "step": 12000
    },
    {
      "epoch": 1.6435223226513282,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007260638181583143,
      "loss": 4.7969,
      "step": 12001
    },
    {
      "epoch": 1.6436592714324842,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007260314660466705,
      "loss": 4.7995,
      "step": 12002
    },
    {
      "epoch": 1.6437962202136402,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007259991116474882,
      "loss": 4.7448,
      "step": 12003
    },
    {
      "epoch": 1.643933168994796,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007259667549610356,
      "loss": 4.8292,
      "step": 12004
    },
    {
      "epoch": 1.6440701177759518,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007259343959875806,
      "loss": 4.8208,
      "step": 12005
    },
    {
      "epoch": 1.6442070665571076,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007259020347273917,
      "loss": 4.8663,
      "step": 12006
    },
    {
      "epoch": 1.6443440153382634,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007258696711807367,
      "loss": 4.8885,
      "step": 12007
    },
    {
      "epoch": 1.6444809641194194,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000725837305347884,
      "loss": 4.7553,
      "step": 12008
    },
    {
      "epoch": 1.6446179129005751,
      "grad_norm": 0.390625,
      "learning_rate": 0.000725804937229102,
      "loss": 4.762,
      "step": 12009
    },
    {
      "epoch": 1.6447548616817311,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007257725668246587,
      "loss": 4.7903,
      "step": 12010
    },
    {
      "epoch": 1.644891810462887,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007257401941348224,
      "loss": 4.759,
      "step": 12011
    },
    {
      "epoch": 1.6450287592440427,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007257078191598614,
      "loss": 4.8229,
      "step": 12012
    },
    {
      "epoch": 1.6451657080251985,
      "grad_norm": 0.40625,
      "learning_rate": 0.000725675441900044,
      "loss": 4.8753,
      "step": 12013
    },
    {
      "epoch": 1.6453026568063545,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007256430623556384,
      "loss": 4.8258,
      "step": 12014
    },
    {
      "epoch": 1.6454396055875102,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007256106805269133,
      "loss": 4.7866,
      "step": 12015
    },
    {
      "epoch": 1.6455765543686662,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007255782964141367,
      "loss": 4.8663,
      "step": 12016
    },
    {
      "epoch": 1.645713503149822,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007255459100175771,
      "loss": 4.7288,
      "step": 12017
    },
    {
      "epoch": 1.6458504519309778,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007255135213375029,
      "loss": 4.7799,
      "step": 12018
    },
    {
      "epoch": 1.6459874007121336,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007254811303741824,
      "loss": 4.8663,
      "step": 12019
    },
    {
      "epoch": 1.6461243494932896,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007254487371278843,
      "loss": 4.8625,
      "step": 12020
    },
    {
      "epoch": 1.6462612982744453,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007254163415988767,
      "loss": 4.7705,
      "step": 12021
    },
    {
      "epoch": 1.6463982470556013,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007253839437874283,
      "loss": 4.873,
      "step": 12022
    },
    {
      "epoch": 1.6465351958367571,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007253515436938077,
      "loss": 4.8186,
      "step": 12023
    },
    {
      "epoch": 1.646672144617913,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000725319141318283,
      "loss": 4.8196,
      "step": 12024
    },
    {
      "epoch": 1.6468090933990687,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007252867366611232,
      "loss": 4.7817,
      "step": 12025
    },
    {
      "epoch": 1.6469460421802244,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007252543297225964,
      "loss": 4.8774,
      "step": 12026
    },
    {
      "epoch": 1.6470829909613804,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007252219205029715,
      "loss": 4.8448,
      "step": 12027
    },
    {
      "epoch": 1.6472199397425364,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000725189509002517,
      "loss": 4.8071,
      "step": 12028
    },
    {
      "epoch": 1.6473568885236922,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007251570952215014,
      "loss": 4.7848,
      "step": 12029
    },
    {
      "epoch": 1.647493837304848,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007251246791601933,
      "loss": 4.8677,
      "step": 12030
    },
    {
      "epoch": 1.6476307860860038,
      "grad_norm": 0.375,
      "learning_rate": 0.0007250922608188616,
      "loss": 4.7745,
      "step": 12031
    },
    {
      "epoch": 1.6477677348671596,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007250598401977748,
      "loss": 4.7559,
      "step": 12032
    },
    {
      "epoch": 1.6479046836483155,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007250274172972014,
      "loss": 4.8291,
      "step": 12033
    },
    {
      "epoch": 1.6480416324294713,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007249949921174103,
      "loss": 4.818,
      "step": 12034
    },
    {
      "epoch": 1.6481785812106273,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007249625646586702,
      "loss": 4.7952,
      "step": 12035
    },
    {
      "epoch": 1.648315529991783,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007249301349212498,
      "loss": 4.7456,
      "step": 12036
    },
    {
      "epoch": 1.6484524787729389,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007248977029054177,
      "loss": 4.7888,
      "step": 12037
    },
    {
      "epoch": 1.6485894275540947,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007248652686114432,
      "loss": 4.75,
      "step": 12038
    },
    {
      "epoch": 1.6487263763352507,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007248328320395946,
      "loss": 4.7876,
      "step": 12039
    },
    {
      "epoch": 1.6488633251164064,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007248003931901407,
      "loss": 4.727,
      "step": 12040
    },
    {
      "epoch": 1.6490002738975624,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007247679520633504,
      "loss": 4.7506,
      "step": 12041
    },
    {
      "epoch": 1.6491372226787182,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007247355086594928,
      "loss": 4.7957,
      "step": 12042
    },
    {
      "epoch": 1.649274171459874,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007247030629788364,
      "loss": 4.8351,
      "step": 12043
    },
    {
      "epoch": 1.6494111202410298,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007246706150216503,
      "loss": 4.8935,
      "step": 12044
    },
    {
      "epoch": 1.6495480690221855,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007246381647882035,
      "loss": 4.8362,
      "step": 12045
    },
    {
      "epoch": 1.6496850178033415,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007246057122787646,
      "loss": 4.8253,
      "step": 12046
    },
    {
      "epoch": 1.6498219665844975,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000724573257493603,
      "loss": 4.7763,
      "step": 12047
    },
    {
      "epoch": 1.6499589153656533,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007245408004329871,
      "loss": 4.8049,
      "step": 12048
    },
    {
      "epoch": 1.650095864146809,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007245083410971863,
      "loss": 4.7974,
      "step": 12049
    },
    {
      "epoch": 1.6502328129279649,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007244758794864694,
      "loss": 4.7915,
      "step": 12050
    },
    {
      "epoch": 1.6503697617091206,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007244434156011057,
      "loss": 4.7761,
      "step": 12051
    },
    {
      "epoch": 1.6505067104902766,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007244109494413636,
      "loss": 4.8055,
      "step": 12052
    },
    {
      "epoch": 1.6506436592714326,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007243784810075129,
      "loss": 4.8694,
      "step": 12053
    },
    {
      "epoch": 1.6507806080525884,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007243460102998224,
      "loss": 4.8038,
      "step": 12054
    },
    {
      "epoch": 1.6509175568337442,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007243135373185609,
      "loss": 4.8824,
      "step": 12055
    },
    {
      "epoch": 1.6510545056149,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007242810620639978,
      "loss": 4.86,
      "step": 12056
    },
    {
      "epoch": 1.6511914543960557,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007242485845364024,
      "loss": 4.7926,
      "step": 12057
    },
    {
      "epoch": 1.6513284031772117,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007242161047360434,
      "loss": 4.8108,
      "step": 12058
    },
    {
      "epoch": 1.6514653519583675,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007241836226631902,
      "loss": 4.7435,
      "step": 12059
    },
    {
      "epoch": 1.6516023007395235,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007241511383181122,
      "loss": 4.7983,
      "step": 12060
    },
    {
      "epoch": 1.6517392495206793,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007241186517010781,
      "loss": 4.8361,
      "step": 12061
    },
    {
      "epoch": 1.651876198301835,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007240861628123577,
      "loss": 4.7361,
      "step": 12062
    },
    {
      "epoch": 1.6520131470829909,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007240536716522197,
      "loss": 4.834,
      "step": 12063
    },
    {
      "epoch": 1.6521500958641469,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007240211782209337,
      "loss": 4.7823,
      "step": 12064
    },
    {
      "epoch": 1.6522870446453026,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000723988682518769,
      "loss": 4.7767,
      "step": 12065
    },
    {
      "epoch": 1.6524239934264586,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007239561845459947,
      "loss": 4.8782,
      "step": 12066
    },
    {
      "epoch": 1.6525609422076144,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007239236843028803,
      "loss": 4.7871,
      "step": 12067
    },
    {
      "epoch": 1.6526978909887702,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007238911817896949,
      "loss": 4.8256,
      "step": 12068
    },
    {
      "epoch": 1.652834839769926,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007238586770067083,
      "loss": 4.7746,
      "step": 12069
    },
    {
      "epoch": 1.6529717885510817,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007238261699541892,
      "loss": 4.8852,
      "step": 12070
    },
    {
      "epoch": 1.6531087373322377,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007237936606324076,
      "loss": 4.7998,
      "step": 12071
    },
    {
      "epoch": 1.6532456861133937,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007237611490416327,
      "loss": 4.81,
      "step": 12072
    },
    {
      "epoch": 1.6533826348945495,
      "grad_norm": 0.359375,
      "learning_rate": 0.000723728635182134,
      "loss": 4.8,
      "step": 12073
    },
    {
      "epoch": 1.6535195836757053,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007236961190541808,
      "loss": 4.8183,
      "step": 12074
    },
    {
      "epoch": 1.653656532456861,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007236636006580426,
      "loss": 4.8346,
      "step": 12075
    },
    {
      "epoch": 1.6537934812380168,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007236310799939888,
      "loss": 4.8959,
      "step": 12076
    },
    {
      "epoch": 1.6539304300191728,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007235985570622893,
      "loss": 4.7457,
      "step": 12077
    },
    {
      "epoch": 1.6540673788003288,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007235660318632133,
      "loss": 4.8552,
      "step": 12078
    },
    {
      "epoch": 1.6542043275814846,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007235335043970303,
      "loss": 4.8054,
      "step": 12079
    },
    {
      "epoch": 1.6543412763626404,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00072350097466401,
      "loss": 4.8305,
      "step": 12080
    },
    {
      "epoch": 1.6544782251437962,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000723468442664422,
      "loss": 4.7871,
      "step": 12081
    },
    {
      "epoch": 1.654615173924952,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007234359083985358,
      "loss": 4.7409,
      "step": 12082
    },
    {
      "epoch": 1.654752122706108,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007234033718666211,
      "loss": 4.8803,
      "step": 12083
    },
    {
      "epoch": 1.6548890714872637,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007233708330689476,
      "loss": 4.8178,
      "step": 12084
    },
    {
      "epoch": 1.6550260202684197,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007233382920057848,
      "loss": 4.7631,
      "step": 12085
    },
    {
      "epoch": 1.6551629690495755,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007233057486774023,
      "loss": 4.7446,
      "step": 12086
    },
    {
      "epoch": 1.6552999178307313,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00072327320308407,
      "loss": 4.7515,
      "step": 12087
    },
    {
      "epoch": 1.655436866611887,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007232406552260576,
      "loss": 4.8628,
      "step": 12088
    },
    {
      "epoch": 1.655573815393043,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007232081051036348,
      "loss": 4.8584,
      "step": 12089
    },
    {
      "epoch": 1.6557107641741988,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007231755527170714,
      "loss": 4.8383,
      "step": 12090
    },
    {
      "epoch": 1.6558477129553548,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000723142998066637,
      "loss": 4.7871,
      "step": 12091
    },
    {
      "epoch": 1.6559846617365106,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007231104411526014,
      "loss": 4.8479,
      "step": 12092
    },
    {
      "epoch": 1.6561216105176664,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007230778819752346,
      "loss": 4.7607,
      "step": 12093
    },
    {
      "epoch": 1.6562585592988222,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007230453205348063,
      "loss": 4.7391,
      "step": 12094
    },
    {
      "epoch": 1.656395508079978,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007230127568315864,
      "loss": 4.8697,
      "step": 12095
    },
    {
      "epoch": 1.656532456861134,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007229801908658447,
      "loss": 4.7349,
      "step": 12096
    },
    {
      "epoch": 1.65666940564229,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007229476226378513,
      "loss": 4.7434,
      "step": 12097
    },
    {
      "epoch": 1.6568063544234457,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007229150521478756,
      "loss": 4.9456,
      "step": 12098
    },
    {
      "epoch": 1.6569433032046015,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007228824793961881,
      "loss": 4.8017,
      "step": 12099
    },
    {
      "epoch": 1.6570802519857573,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007228499043830583,
      "loss": 4.8749,
      "step": 12100
    },
    {
      "epoch": 1.657217200766913,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007228173271087564,
      "loss": 4.7385,
      "step": 12101
    },
    {
      "epoch": 1.657354149548069,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007227847475735522,
      "loss": 4.7603,
      "step": 12102
    },
    {
      "epoch": 1.657491098329225,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007227521657777158,
      "loss": 4.8605,
      "step": 12103
    },
    {
      "epoch": 1.6576280471103808,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007227195817215173,
      "loss": 4.8542,
      "step": 12104
    },
    {
      "epoch": 1.6577649958915366,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007226869954052267,
      "loss": 4.7869,
      "step": 12105
    },
    {
      "epoch": 1.6579019446726924,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007226544068291138,
      "loss": 4.782,
      "step": 12106
    },
    {
      "epoch": 1.6580388934538481,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000722621815993449,
      "loss": 4.8025,
      "step": 12107
    },
    {
      "epoch": 1.6581758422350041,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007225892228985021,
      "loss": 4.7341,
      "step": 12108
    },
    {
      "epoch": 1.65831279101616,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007225566275445435,
      "loss": 4.8669,
      "step": 12109
    },
    {
      "epoch": 1.658449739797316,
      "grad_norm": 0.375,
      "learning_rate": 0.0007225240299318431,
      "loss": 4.74,
      "step": 12110
    },
    {
      "epoch": 1.6585866885784717,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007224914300606712,
      "loss": 4.8681,
      "step": 12111
    },
    {
      "epoch": 1.6587236373596275,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007224588279312979,
      "loss": 4.891,
      "step": 12112
    },
    {
      "epoch": 1.6588605861407832,
      "grad_norm": 0.375,
      "learning_rate": 0.0007224262235439933,
      "loss": 4.7227,
      "step": 12113
    },
    {
      "epoch": 1.6589975349219392,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007223936168990277,
      "loss": 4.8967,
      "step": 12114
    },
    {
      "epoch": 1.659134483703095,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007223610079966712,
      "loss": 4.8639,
      "step": 12115
    },
    {
      "epoch": 1.659271432484251,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007223283968371943,
      "loss": 4.8686,
      "step": 12116
    },
    {
      "epoch": 1.6594083812654068,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000722295783420867,
      "loss": 4.8745,
      "step": 12117
    },
    {
      "epoch": 1.6595453300465626,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007222631677479596,
      "loss": 4.8511,
      "step": 12118
    },
    {
      "epoch": 1.6596822788277183,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007222305498187425,
      "loss": 4.783,
      "step": 12119
    },
    {
      "epoch": 1.6598192276088741,
      "grad_norm": 0.40625,
      "learning_rate": 0.000722197929633486,
      "loss": 4.825,
      "step": 12120
    },
    {
      "epoch": 1.6599561763900301,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007221653071924604,
      "loss": 4.8652,
      "step": 12121
    },
    {
      "epoch": 1.6600931251711861,
      "grad_norm": 0.421875,
      "learning_rate": 0.000722132682495936,
      "loss": 4.6656,
      "step": 12122
    },
    {
      "epoch": 1.660230073952342,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007221000555441832,
      "loss": 4.8112,
      "step": 12123
    },
    {
      "epoch": 1.6603670227334977,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007220674263374726,
      "loss": 4.8197,
      "step": 12124
    },
    {
      "epoch": 1.6605039715146535,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007220347948760741,
      "loss": 4.8146,
      "step": 12125
    },
    {
      "epoch": 1.6606409202958092,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007220021611602584,
      "loss": 4.8829,
      "step": 12126
    },
    {
      "epoch": 1.6607778690769652,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007219695251902962,
      "loss": 4.6537,
      "step": 12127
    },
    {
      "epoch": 1.6609148178581212,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0007219368869664578,
      "loss": 4.8225,
      "step": 12128
    },
    {
      "epoch": 1.661051766639277,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0007219042464890135,
      "loss": 4.791,
      "step": 12129
    },
    {
      "epoch": 1.6611887154204328,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007218716037582339,
      "loss": 4.7793,
      "step": 12130
    },
    {
      "epoch": 1.6613256642015886,
      "grad_norm": 0.375,
      "learning_rate": 0.0007218389587743896,
      "loss": 4.7498,
      "step": 12131
    },
    {
      "epoch": 1.6614626129827443,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0007218063115377511,
      "loss": 4.8674,
      "step": 12132
    },
    {
      "epoch": 1.6615995617639003,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007217736620485888,
      "loss": 4.8135,
      "step": 12133
    },
    {
      "epoch": 1.6617365105450561,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0007217410103071734,
      "loss": 4.7847,
      "step": 12134
    },
    {
      "epoch": 1.661873459326212,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007217083563137755,
      "loss": 4.8191,
      "step": 12135
    },
    {
      "epoch": 1.6620104081073679,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0007216757000686659,
      "loss": 4.8618,
      "step": 12136
    },
    {
      "epoch": 1.6621473568885237,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007216430415721148,
      "loss": 4.795,
      "step": 12137
    },
    {
      "epoch": 1.6622843056696794,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0007216103808243932,
      "loss": 4.8577,
      "step": 12138
    },
    {
      "epoch": 1.6624212544508354,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0007215777178257716,
      "loss": 4.8159,
      "step": 12139
    },
    {
      "epoch": 1.6625582032319912,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007215450525765208,
      "loss": 4.7626,
      "step": 12140
    },
    {
      "epoch": 1.6626951520131472,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0007215123850769114,
      "loss": 4.8194,
      "step": 12141
    },
    {
      "epoch": 1.662832100794303,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007214797153272142,
      "loss": 4.882,
      "step": 12142
    },
    {
      "epoch": 1.6629690495754588,
      "grad_norm": 0.453125,
      "learning_rate": 0.0007214470433276999,
      "loss": 4.7221,
      "step": 12143
    },
    {
      "epoch": 1.6631059983566145,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007214143690786393,
      "loss": 4.8487,
      "step": 12144
    },
    {
      "epoch": 1.6632429471377703,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000721381692580303,
      "loss": 4.7961,
      "step": 12145
    },
    {
      "epoch": 1.6633798959189263,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007213490138329622,
      "loss": 4.8912,
      "step": 12146
    },
    {
      "epoch": 1.6635168447000823,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007213163328368872,
      "loss": 4.8157,
      "step": 12147
    },
    {
      "epoch": 1.663653793481238,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007212836495923494,
      "loss": 4.7785,
      "step": 12148
    },
    {
      "epoch": 1.6637907422623939,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007212509640996189,
      "loss": 4.8751,
      "step": 12149
    },
    {
      "epoch": 1.6639276910435497,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007212182763589675,
      "loss": 4.8424,
      "step": 12150
    },
    {
      "epoch": 1.6640646398247054,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007211855863706654,
      "loss": 4.8009,
      "step": 12151
    },
    {
      "epoch": 1.6642015886058614,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007211528941349837,
      "loss": 4.8735,
      "step": 12152
    },
    {
      "epoch": 1.6643385373870172,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007211201996521933,
      "loss": 4.8408,
      "step": 12153
    },
    {
      "epoch": 1.6644754861681732,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007210875029225655,
      "loss": 4.895,
      "step": 12154
    },
    {
      "epoch": 1.664612434949329,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007210548039463706,
      "loss": 4.808,
      "step": 12155
    },
    {
      "epoch": 1.6647493837304848,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007210221027238802,
      "loss": 4.8579,
      "step": 12156
    },
    {
      "epoch": 1.6648863325116405,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007209893992553649,
      "loss": 4.8422,
      "step": 12157
    },
    {
      "epoch": 1.6650232812927965,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007209566935410959,
      "loss": 4.7721,
      "step": 12158
    },
    {
      "epoch": 1.6651602300739523,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007209239855813445,
      "loss": 4.8446,
      "step": 12159
    },
    {
      "epoch": 1.6652971788551083,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007208912753763811,
      "loss": 4.9269,
      "step": 12160
    },
    {
      "epoch": 1.665434127636264,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007208585629264772,
      "loss": 4.808,
      "step": 12161
    },
    {
      "epoch": 1.6655710764174199,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007208258482319039,
      "loss": 4.8613,
      "step": 12162
    },
    {
      "epoch": 1.6657080251985756,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007207931312929323,
      "loss": 4.8122,
      "step": 12163
    },
    {
      "epoch": 1.6658449739797316,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007207604121098335,
      "loss": 4.7403,
      "step": 12164
    },
    {
      "epoch": 1.6659819227608874,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007207276906828787,
      "loss": 4.7768,
      "step": 12165
    },
    {
      "epoch": 1.6661188715420434,
      "grad_norm": 0.375,
      "learning_rate": 0.0007206949670123389,
      "loss": 4.8747,
      "step": 12166
    },
    {
      "epoch": 1.6662558203231992,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007206622410984854,
      "loss": 4.7664,
      "step": 12167
    },
    {
      "epoch": 1.666392769104355,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007206295129415896,
      "loss": 4.8131,
      "step": 12168
    },
    {
      "epoch": 1.6665297178855107,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007205967825419224,
      "loss": 4.7992,
      "step": 12169
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007205640498997552,
      "loss": 4.8547,
      "step": 12170
    },
    {
      "epoch": 1.6668036154478225,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007205313150153591,
      "loss": 4.7447,
      "step": 12171
    },
    {
      "epoch": 1.6669405642289785,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007204985778890057,
      "loss": 4.8955,
      "step": 12172
    },
    {
      "epoch": 1.6670775130101343,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007204658385209661,
      "loss": 4.7574,
      "step": 12173
    },
    {
      "epoch": 1.66721446179129,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007204330969115117,
      "loss": 4.8297,
      "step": 12174
    },
    {
      "epoch": 1.6673514105724458,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007204003530609138,
      "loss": 4.8433,
      "step": 12175
    },
    {
      "epoch": 1.6674883593536016,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007203676069694435,
      "loss": 4.794,
      "step": 12176
    },
    {
      "epoch": 1.6676253081347576,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007203348586373726,
      "loss": 4.8327,
      "step": 12177
    },
    {
      "epoch": 1.6677622569159134,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007203021080649724,
      "loss": 4.9146,
      "step": 12178
    },
    {
      "epoch": 1.6678992056970694,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007202693552525141,
      "loss": 4.7846,
      "step": 12179
    },
    {
      "epoch": 1.6680361544782252,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007202366002002692,
      "loss": 4.7956,
      "step": 12180
    },
    {
      "epoch": 1.668173103259381,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007202038429085094,
      "loss": 4.7642,
      "step": 12181
    },
    {
      "epoch": 1.6683100520405367,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007201710833775057,
      "loss": 4.8066,
      "step": 12182
    },
    {
      "epoch": 1.6684470008216927,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007201383216075299,
      "loss": 4.7735,
      "step": 12183
    },
    {
      "epoch": 1.6685839496028485,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007201055575988536,
      "loss": 4.8085,
      "step": 12184
    },
    {
      "epoch": 1.6687208983840045,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007200727913517481,
      "loss": 4.8774,
      "step": 12185
    },
    {
      "epoch": 1.6688578471651603,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007200400228664851,
      "loss": 4.8067,
      "step": 12186
    },
    {
      "epoch": 1.668994795946316,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007200072521433361,
      "loss": 4.7982,
      "step": 12187
    },
    {
      "epoch": 1.6691317447274718,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007199744791825725,
      "loss": 4.8365,
      "step": 12188
    },
    {
      "epoch": 1.6692686935086276,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007199417039844662,
      "loss": 4.7868,
      "step": 12189
    },
    {
      "epoch": 1.6694056422897836,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007199089265492887,
      "loss": 4.8025,
      "step": 12190
    },
    {
      "epoch": 1.6695425910709396,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007198761468773115,
      "loss": 4.8756,
      "step": 12191
    },
    {
      "epoch": 1.6696795398520954,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007198433649688063,
      "loss": 4.7682,
      "step": 12192
    },
    {
      "epoch": 1.6698164886332512,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000719810580824045,
      "loss": 4.8037,
      "step": 12193
    },
    {
      "epoch": 1.669953437414407,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000719777794443299,
      "loss": 4.7919,
      "step": 12194
    },
    {
      "epoch": 1.6700903861955627,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007197450058268402,
      "loss": 4.811,
      "step": 12195
    },
    {
      "epoch": 1.6702273349767187,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007197122149749403,
      "loss": 4.8935,
      "step": 12196
    },
    {
      "epoch": 1.6703642837578747,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000719679421887871,
      "loss": 4.8446,
      "step": 12197
    },
    {
      "epoch": 1.6705012325390305,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000719646626565904,
      "loss": 4.7923,
      "step": 12198
    },
    {
      "epoch": 1.6706381813201863,
      "grad_norm": 0.375,
      "learning_rate": 0.0007196138290093112,
      "loss": 4.7647,
      "step": 12199
    },
    {
      "epoch": 1.670775130101342,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007195810292183643,
      "loss": 4.7714,
      "step": 12200
    },
    {
      "epoch": 1.6709120788824978,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007195482271933352,
      "loss": 4.861,
      "step": 12201
    },
    {
      "epoch": 1.6710490276636538,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007195154229344957,
      "loss": 4.778,
      "step": 12202
    },
    {
      "epoch": 1.6711859764448096,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007194826164421176,
      "loss": 4.8811,
      "step": 12203
    },
    {
      "epoch": 1.6713229252259656,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007194498077164731,
      "loss": 4.7983,
      "step": 12204
    },
    {
      "epoch": 1.6714598740071214,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007194169967578336,
      "loss": 4.8237,
      "step": 12205
    },
    {
      "epoch": 1.6715968227882771,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007193841835664714,
      "loss": 4.7685,
      "step": 12206
    },
    {
      "epoch": 1.671733771569433,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007193513681426581,
      "loss": 4.8825,
      "step": 12207
    },
    {
      "epoch": 1.671870720350589,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000719318550486666,
      "loss": 4.7817,
      "step": 12208
    },
    {
      "epoch": 1.6720076691317447,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007192857305987668,
      "loss": 4.8111,
      "step": 12209
    },
    {
      "epoch": 1.6721446179129007,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0007192529084792326,
      "loss": 4.8146,
      "step": 12210
    },
    {
      "epoch": 1.6722815666940565,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007192200841283354,
      "loss": 4.7932,
      "step": 12211
    },
    {
      "epoch": 1.6724185154752123,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007191872575463471,
      "loss": 4.8435,
      "step": 12212
    },
    {
      "epoch": 1.672555464256368,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00071915442873354,
      "loss": 4.7922,
      "step": 12213
    },
    {
      "epoch": 1.6726924130375238,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000719121597690186,
      "loss": 4.8473,
      "step": 12214
    },
    {
      "epoch": 1.6728293618186798,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007190887644165571,
      "loss": 4.8207,
      "step": 12215
    },
    {
      "epoch": 1.6729663105998358,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007190559289129257,
      "loss": 4.8813,
      "step": 12216
    },
    {
      "epoch": 1.6731032593809916,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007190230911795635,
      "loss": 4.896,
      "step": 12217
    },
    {
      "epoch": 1.6732402081621474,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007189902512167428,
      "loss": 4.8752,
      "step": 12218
    },
    {
      "epoch": 1.6733771569433031,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000718957409024736,
      "loss": 4.8154,
      "step": 12219
    },
    {
      "epoch": 1.673514105724459,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0007189245646038149,
      "loss": 4.8441,
      "step": 12220
    },
    {
      "epoch": 1.673651054505615,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007188917179542518,
      "loss": 4.8131,
      "step": 12221
    },
    {
      "epoch": 1.673788003286771,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0007188588690763191,
      "loss": 4.7886,
      "step": 12222
    },
    {
      "epoch": 1.6739249520679267,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007188260179702889,
      "loss": 4.8318,
      "step": 12223
    },
    {
      "epoch": 1.6740619008490825,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007187931646364334,
      "loss": 4.7772,
      "step": 12224
    },
    {
      "epoch": 1.6741988496302382,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0007187603090750248,
      "loss": 4.8168,
      "step": 12225
    },
    {
      "epoch": 1.674335798411394,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007187274512863356,
      "loss": 4.8192,
      "step": 12226
    },
    {
      "epoch": 1.67447274719255,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007186945912706379,
      "loss": 4.8143,
      "step": 12227
    },
    {
      "epoch": 1.6746096959737058,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007186617290282041,
      "loss": 4.8686,
      "step": 12228
    },
    {
      "epoch": 1.6747466447548618,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007186288645593065,
      "loss": 4.8179,
      "step": 12229
    },
    {
      "epoch": 1.6748835935360176,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007185959978642176,
      "loss": 4.8704,
      "step": 12230
    },
    {
      "epoch": 1.6750205423171733,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007185631289432096,
      "loss": 4.8624,
      "step": 12231
    },
    {
      "epoch": 1.6751574910983291,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000718530257796555,
      "loss": 4.9298,
      "step": 12232
    },
    {
      "epoch": 1.6752944398794851,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007184973844245262,
      "loss": 4.8183,
      "step": 12233
    },
    {
      "epoch": 1.675431388660641,
      "grad_norm": 0.375,
      "learning_rate": 0.0007184645088273955,
      "loss": 4.848,
      "step": 12234
    },
    {
      "epoch": 1.675568337441797,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007184316310054355,
      "loss": 4.8312,
      "step": 12235
    },
    {
      "epoch": 1.6757052862229527,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007183987509589185,
      "loss": 4.85,
      "step": 12236
    },
    {
      "epoch": 1.6758422350041084,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007183658686881171,
      "loss": 4.7667,
      "step": 12237
    },
    {
      "epoch": 1.6759791837852642,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007183329841933039,
      "loss": 4.7836,
      "step": 12238
    },
    {
      "epoch": 1.67611613256642,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007183000974747513,
      "loss": 4.824,
      "step": 12239
    },
    {
      "epoch": 1.676253081347576,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007182672085327319,
      "loss": 4.9242,
      "step": 12240
    },
    {
      "epoch": 1.676390030128732,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007182343173675181,
      "loss": 4.8187,
      "step": 12241
    },
    {
      "epoch": 1.6765269789098878,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007182014239793828,
      "loss": 4.7805,
      "step": 12242
    },
    {
      "epoch": 1.6766639276910436,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000718168528368598,
      "loss": 4.863,
      "step": 12243
    },
    {
      "epoch": 1.6768008764721993,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007181356305354371,
      "loss": 4.7736,
      "step": 12244
    },
    {
      "epoch": 1.676937825253355,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007181027304801721,
      "loss": 4.8717,
      "step": 12245
    },
    {
      "epoch": 1.677074774034511,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007180698282030761,
      "loss": 4.7868,
      "step": 12246
    },
    {
      "epoch": 1.677211722815667,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007180369237044213,
      "loss": 4.8297,
      "step": 12247
    },
    {
      "epoch": 1.6773486715968229,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007180040169844809,
      "loss": 4.8343,
      "step": 12248
    },
    {
      "epoch": 1.6774856203779787,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007179711080435272,
      "loss": 4.8567,
      "step": 12249
    },
    {
      "epoch": 1.6776225691591344,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007179381968818331,
      "loss": 4.8619,
      "step": 12250
    },
    {
      "epoch": 1.6777595179402902,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007179052834996715,
      "loss": 4.7972,
      "step": 12251
    },
    {
      "epoch": 1.6778964667214462,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007178723678973147,
      "loss": 4.7893,
      "step": 12252
    },
    {
      "epoch": 1.678033415502602,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007178394500750359,
      "loss": 4.7847,
      "step": 12253
    },
    {
      "epoch": 1.678170364283758,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007178065300331077,
      "loss": 4.8017,
      "step": 12254
    },
    {
      "epoch": 1.6783073130649138,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007177736077718029,
      "loss": 4.8564,
      "step": 12255
    },
    {
      "epoch": 1.6784442618460695,
      "grad_norm": 0.328125,
      "learning_rate": 0.0007177406832913945,
      "loss": 4.8212,
      "step": 12256
    },
    {
      "epoch": 1.6785812106272253,
      "grad_norm": 0.375,
      "learning_rate": 0.0007177077565921552,
      "loss": 4.8949,
      "step": 12257
    },
    {
      "epoch": 1.6787181594083813,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000717674827674358,
      "loss": 4.8562,
      "step": 12258
    },
    {
      "epoch": 1.678855108189537,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007176418965382757,
      "loss": 4.7929,
      "step": 12259
    },
    {
      "epoch": 1.678992056970693,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007176089631841813,
      "loss": 4.8339,
      "step": 12260
    },
    {
      "epoch": 1.6791290057518489,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007175760276123476,
      "loss": 4.7831,
      "step": 12261
    },
    {
      "epoch": 1.6792659545330046,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007175430898230475,
      "loss": 4.8937,
      "step": 12262
    },
    {
      "epoch": 1.6794029033141604,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007175101498165541,
      "loss": 4.8599,
      "step": 12263
    },
    {
      "epoch": 1.6795398520953162,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007174772075931404,
      "loss": 4.8092,
      "step": 12264
    },
    {
      "epoch": 1.6796768008764722,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007174442631530793,
      "loss": 4.7754,
      "step": 12265
    },
    {
      "epoch": 1.6798137496576282,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007174113164966439,
      "loss": 4.8173,
      "step": 12266
    },
    {
      "epoch": 1.679950698438784,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007173783676241072,
      "loss": 4.7664,
      "step": 12267
    },
    {
      "epoch": 1.6800876472199398,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007173454165357422,
      "loss": 4.8149,
      "step": 12268
    },
    {
      "epoch": 1.6802245960010955,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000717312463231822,
      "loss": 4.794,
      "step": 12269
    },
    {
      "epoch": 1.6803615447822513,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007172795077126199,
      "loss": 4.8019,
      "step": 12270
    },
    {
      "epoch": 1.6804984935634073,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007172465499784088,
      "loss": 4.7895,
      "step": 12271
    },
    {
      "epoch": 1.680635442344563,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007172135900294617,
      "loss": 4.731,
      "step": 12272
    },
    {
      "epoch": 1.680772391125719,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007171806278660521,
      "loss": 4.7363,
      "step": 12273
    },
    {
      "epoch": 1.6809093399068749,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007171476634884527,
      "loss": 4.7947,
      "step": 12274
    },
    {
      "epoch": 1.6810462886880306,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007171146968969372,
      "loss": 4.8175,
      "step": 12275
    },
    {
      "epoch": 1.6811832374691864,
      "grad_norm": 0.375,
      "learning_rate": 0.0007170817280917784,
      "loss": 4.8418,
      "step": 12276
    },
    {
      "epoch": 1.6813201862503424,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007170487570732498,
      "loss": 4.9285,
      "step": 12277
    },
    {
      "epoch": 1.6814571350314982,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007170157838416244,
      "loss": 4.8646,
      "step": 12278
    },
    {
      "epoch": 1.6815940838126542,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007169828083971756,
      "loss": 4.8012,
      "step": 12279
    },
    {
      "epoch": 1.68173103259381,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007169498307401768,
      "loss": 4.764,
      "step": 12280
    },
    {
      "epoch": 1.6818679813749657,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007169168508709009,
      "loss": 4.7425,
      "step": 12281
    },
    {
      "epoch": 1.6820049301561215,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007168838687896216,
      "loss": 4.8003,
      "step": 12282
    },
    {
      "epoch": 1.6821418789372775,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007168508844966121,
      "loss": 4.8808,
      "step": 12283
    },
    {
      "epoch": 1.6822788277184333,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007168178979921458,
      "loss": 4.8129,
      "step": 12284
    },
    {
      "epoch": 1.6824157764995893,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007167849092764958,
      "loss": 4.8513,
      "step": 12285
    },
    {
      "epoch": 1.682552725280745,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007167519183499358,
      "loss": 4.8207,
      "step": 12286
    },
    {
      "epoch": 1.6826896740619008,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007167189252127391,
      "loss": 4.8647,
      "step": 12287
    },
    {
      "epoch": 1.6828266228430566,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000716685929865179,
      "loss": 4.8227,
      "step": 12288
    },
    {
      "epoch": 1.6829635716242124,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007166529323075293,
      "loss": 4.836,
      "step": 12289
    },
    {
      "epoch": 1.6831005204053684,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007166199325400629,
      "loss": 4.8516,
      "step": 12290
    },
    {
      "epoch": 1.6832374691865244,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007165869305630539,
      "loss": 4.8179,
      "step": 12291
    },
    {
      "epoch": 1.6833744179676802,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007165539263767751,
      "loss": 4.8095,
      "step": 12292
    },
    {
      "epoch": 1.683511366748836,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007165209199815008,
      "loss": 4.7494,
      "step": 12293
    },
    {
      "epoch": 1.6836483155299917,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007164879113775039,
      "loss": 4.8486,
      "step": 12294
    },
    {
      "epoch": 1.6837852643111475,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007164549005650581,
      "loss": 4.7835,
      "step": 12295
    },
    {
      "epoch": 1.6839222130923035,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007164218875444372,
      "loss": 4.8673,
      "step": 12296
    },
    {
      "epoch": 1.6840591618734593,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007163888723159147,
      "loss": 4.8271,
      "step": 12297
    },
    {
      "epoch": 1.6841961106546153,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007163558548797639,
      "loss": 4.7714,
      "step": 12298
    },
    {
      "epoch": 1.684333059435771,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007163228352362589,
      "loss": 4.7947,
      "step": 12299
    },
    {
      "epoch": 1.6844700082169268,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007162898133856729,
      "loss": 4.8336,
      "step": 12300
    },
    {
      "epoch": 1.6846069569980826,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007162567893282798,
      "loss": 4.7431,
      "step": 12301
    },
    {
      "epoch": 1.6847439057792386,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007162237630643533,
      "loss": 4.8767,
      "step": 12302
    },
    {
      "epoch": 1.6848808545603944,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007161907345941671,
      "loss": 4.8409,
      "step": 12303
    },
    {
      "epoch": 1.6850178033415504,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007161577039179947,
      "loss": 4.7644,
      "step": 12304
    },
    {
      "epoch": 1.6851547521227062,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007161246710361102,
      "loss": 4.7479,
      "step": 12305
    },
    {
      "epoch": 1.685291700903862,
      "grad_norm": 0.40625,
      "learning_rate": 0.000716091635948787,
      "loss": 4.8623,
      "step": 12306
    },
    {
      "epoch": 1.6854286496850177,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007160585986562991,
      "loss": 4.852,
      "step": 12307
    },
    {
      "epoch": 1.6855655984661735,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00071602555915892,
      "loss": 4.7808,
      "step": 12308
    },
    {
      "epoch": 1.6857025472473295,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007159925174569238,
      "loss": 4.854,
      "step": 12309
    },
    {
      "epoch": 1.6858394960284855,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007159594735505842,
      "loss": 4.7501,
      "step": 12310
    },
    {
      "epoch": 1.6859764448096413,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007159264274401752,
      "loss": 4.8147,
      "step": 12311
    },
    {
      "epoch": 1.686113393590797,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007158933791259705,
      "loss": 4.8155,
      "step": 12312
    },
    {
      "epoch": 1.6862503423719528,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007158603286082439,
      "loss": 4.8168,
      "step": 12313
    },
    {
      "epoch": 1.6863872911531086,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007158272758872695,
      "loss": 4.8822,
      "step": 12314
    },
    {
      "epoch": 1.6865242399342646,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000715794220963321,
      "loss": 4.7628,
      "step": 12315
    },
    {
      "epoch": 1.6866611887154206,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007157611638366725,
      "loss": 4.7967,
      "step": 12316
    },
    {
      "epoch": 1.6867981374965764,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000715728104507598,
      "loss": 4.8551,
      "step": 12317
    },
    {
      "epoch": 1.6869350862777321,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0007156950429763714,
      "loss": 4.8934,
      "step": 12318
    },
    {
      "epoch": 1.687072035058888,
      "grad_norm": 0.375,
      "learning_rate": 0.0007156619792432665,
      "loss": 4.845,
      "step": 12319
    },
    {
      "epoch": 1.6872089838400437,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007156289133085575,
      "loss": 4.8196,
      "step": 12320
    },
    {
      "epoch": 1.6873459326211997,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007155958451725185,
      "loss": 4.8344,
      "step": 12321
    },
    {
      "epoch": 1.6874828814023555,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007155627748354234,
      "loss": 4.8061,
      "step": 12322
    },
    {
      "epoch": 1.6876198301835115,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007155297022975463,
      "loss": 4.7102,
      "step": 12323
    },
    {
      "epoch": 1.6877567789646672,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007154966275591614,
      "loss": 4.7884,
      "step": 12324
    },
    {
      "epoch": 1.687893727745823,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007154635506205424,
      "loss": 4.7961,
      "step": 12325
    },
    {
      "epoch": 1.6880306765269788,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007154304714819639,
      "loss": 4.7744,
      "step": 12326
    },
    {
      "epoch": 1.6881676253081348,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007153973901436999,
      "loss": 4.8946,
      "step": 12327
    },
    {
      "epoch": 1.6883045740892906,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007153643066060244,
      "loss": 4.7838,
      "step": 12328
    },
    {
      "epoch": 1.6884415228704466,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007153312208692115,
      "loss": 4.7906,
      "step": 12329
    },
    {
      "epoch": 1.6885784716516024,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007152981329335358,
      "loss": 4.8342,
      "step": 12330
    },
    {
      "epoch": 1.6887154204327581,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007152650427992712,
      "loss": 4.8696,
      "step": 12331
    },
    {
      "epoch": 1.688852369213914,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007152319504666919,
      "loss": 4.8822,
      "step": 12332
    },
    {
      "epoch": 1.6889893179950697,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0007151988559360721,
      "loss": 4.8597,
      "step": 12333
    },
    {
      "epoch": 1.6891262667762257,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007151657592076865,
      "loss": 4.7767,
      "step": 12334
    },
    {
      "epoch": 1.6892632155573817,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007151326602818089,
      "loss": 4.7957,
      "step": 12335
    },
    {
      "epoch": 1.6894001643385375,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007150995591587139,
      "loss": 4.8081,
      "step": 12336
    },
    {
      "epoch": 1.6895371131196932,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007150664558386755,
      "loss": 4.7352,
      "step": 12337
    },
    {
      "epoch": 1.689674061900849,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007150333503219683,
      "loss": 4.7747,
      "step": 12338
    },
    {
      "epoch": 1.6898110106820048,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007150002426088665,
      "loss": 4.8233,
      "step": 12339
    },
    {
      "epoch": 1.6899479594631608,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007149671326996446,
      "loss": 4.76,
      "step": 12340
    },
    {
      "epoch": 1.6900849082443168,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007149340205945769,
      "loss": 4.8179,
      "step": 12341
    },
    {
      "epoch": 1.6902218570254726,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0007149009062939379,
      "loss": 4.8187,
      "step": 12342
    },
    {
      "epoch": 1.6903588058066283,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000714867789798002,
      "loss": 4.8224,
      "step": 12343
    },
    {
      "epoch": 1.6904957545877841,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007148346711070434,
      "loss": 4.8653,
      "step": 12344
    },
    {
      "epoch": 1.69063270336894,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007148015502213371,
      "loss": 4.8749,
      "step": 12345
    },
    {
      "epoch": 1.690769652150096,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007147684271411569,
      "loss": 4.846,
      "step": 12346
    },
    {
      "epoch": 1.6909066009312517,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000714735301866778,
      "loss": 4.82,
      "step": 12347
    },
    {
      "epoch": 1.6910435497124077,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007147021743984744,
      "loss": 4.849,
      "step": 12348
    },
    {
      "epoch": 1.6911804984935634,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007146690447365207,
      "loss": 4.8238,
      "step": 12349
    },
    {
      "epoch": 1.6913174472747192,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007146359128811917,
      "loss": 4.8043,
      "step": 12350
    },
    {
      "epoch": 1.691454396055875,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007146027788327618,
      "loss": 4.8216,
      "step": 12351
    },
    {
      "epoch": 1.691591344837031,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007145696425915056,
      "loss": 4.8585,
      "step": 12352
    },
    {
      "epoch": 1.6917282936181868,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007145365041576978,
      "loss": 4.8799,
      "step": 12353
    },
    {
      "epoch": 1.6918652423993428,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007145033635316129,
      "loss": 4.8105,
      "step": 12354
    },
    {
      "epoch": 1.6920021911804985,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007144702207135256,
      "loss": 4.7976,
      "step": 12355
    },
    {
      "epoch": 1.6921391399616543,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007144370757037106,
      "loss": 4.8214,
      "step": 12356
    },
    {
      "epoch": 1.69227608874281,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007144039285024424,
      "loss": 4.7781,
      "step": 12357
    },
    {
      "epoch": 1.6924130375239659,
      "grad_norm": 0.359375,
      "learning_rate": 0.000714370779109996,
      "loss": 4.8482,
      "step": 12358
    },
    {
      "epoch": 1.6925499863051219,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000714337627526646,
      "loss": 4.8019,
      "step": 12359
    },
    {
      "epoch": 1.6926869350862779,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000714304473752667,
      "loss": 4.8583,
      "step": 12360
    },
    {
      "epoch": 1.6928238838674337,
      "grad_norm": 0.375,
      "learning_rate": 0.0007142713177883337,
      "loss": 4.8353,
      "step": 12361
    },
    {
      "epoch": 1.6929608326485894,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007142381596339212,
      "loss": 4.8494,
      "step": 12362
    },
    {
      "epoch": 1.6930977814297452,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007142049992897041,
      "loss": 4.8063,
      "step": 12363
    },
    {
      "epoch": 1.693234730210901,
      "grad_norm": 0.375,
      "learning_rate": 0.0007141718367559572,
      "loss": 4.8645,
      "step": 12364
    },
    {
      "epoch": 1.693371678992057,
      "grad_norm": 0.375,
      "learning_rate": 0.0007141386720329552,
      "loss": 4.8372,
      "step": 12365
    },
    {
      "epoch": 1.693508627773213,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007141055051209732,
      "loss": 4.7917,
      "step": 12366
    },
    {
      "epoch": 1.6936455765543688,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007140723360202861,
      "loss": 4.7827,
      "step": 12367
    },
    {
      "epoch": 1.6937825253355245,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007140391647311685,
      "loss": 4.7525,
      "step": 12368
    },
    {
      "epoch": 1.6939194741166803,
      "grad_norm": 0.375,
      "learning_rate": 0.0007140059912538952,
      "loss": 4.8416,
      "step": 12369
    },
    {
      "epoch": 1.694056422897836,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007139728155887416,
      "loss": 4.7127,
      "step": 12370
    },
    {
      "epoch": 1.694193371678992,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007139396377359823,
      "loss": 4.6897,
      "step": 12371
    },
    {
      "epoch": 1.6943303204601479,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007139064576958924,
      "loss": 4.8099,
      "step": 12372
    },
    {
      "epoch": 1.6944672692413039,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007138732754687466,
      "loss": 4.8281,
      "step": 12373
    },
    {
      "epoch": 1.6946042180224596,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007138400910548205,
      "loss": 4.7963,
      "step": 12374
    },
    {
      "epoch": 1.6947411668036154,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007138069044543883,
      "loss": 4.8222,
      "step": 12375
    },
    {
      "epoch": 1.6948781155847712,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007137737156677256,
      "loss": 4.887,
      "step": 12376
    },
    {
      "epoch": 1.6950150643659272,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007137405246951072,
      "loss": 4.7527,
      "step": 12377
    },
    {
      "epoch": 1.695152013147083,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007137073315368083,
      "loss": 4.7614,
      "step": 12378
    },
    {
      "epoch": 1.695288961928239,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007136741361931037,
      "loss": 4.7644,
      "step": 12379
    },
    {
      "epoch": 1.6954259107093947,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007136409386642688,
      "loss": 4.866,
      "step": 12380
    },
    {
      "epoch": 1.6955628594905505,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007136077389505788,
      "loss": 4.8721,
      "step": 12381
    },
    {
      "epoch": 1.6956998082717063,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007135745370523085,
      "loss": 4.7472,
      "step": 12382
    },
    {
      "epoch": 1.695836757052862,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007135413329697332,
      "loss": 4.878,
      "step": 12383
    },
    {
      "epoch": 1.695973705834018,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007135081267031281,
      "loss": 4.7388,
      "step": 12384
    },
    {
      "epoch": 1.696110654615174,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007134749182527685,
      "loss": 4.8323,
      "step": 12385
    },
    {
      "epoch": 1.6962476033963299,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007134417076189293,
      "loss": 4.8316,
      "step": 12386
    },
    {
      "epoch": 1.6963845521774856,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007134084948018859,
      "loss": 4.7033,
      "step": 12387
    },
    {
      "epoch": 1.6965215009586414,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007133752798019136,
      "loss": 4.852,
      "step": 12388
    },
    {
      "epoch": 1.6966584497397972,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007133420626192875,
      "loss": 4.7567,
      "step": 12389
    },
    {
      "epoch": 1.6967953985209532,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007133088432542831,
      "loss": 4.7349,
      "step": 12390
    },
    {
      "epoch": 1.696932347302109,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007132756217071755,
      "loss": 4.7504,
      "step": 12391
    },
    {
      "epoch": 1.697069296083265,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0007132423979782401,
      "loss": 4.8138,
      "step": 12392
    },
    {
      "epoch": 1.6972062448644207,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007132091720677521,
      "loss": 4.9123,
      "step": 12393
    },
    {
      "epoch": 1.6973431936455765,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0007131759439759872,
      "loss": 4.836,
      "step": 12394
    },
    {
      "epoch": 1.6974801424267323,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007131427137032204,
      "loss": 4.8118,
      "step": 12395
    },
    {
      "epoch": 1.6976170912078883,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007131094812497273,
      "loss": 4.8851,
      "step": 12396
    },
    {
      "epoch": 1.697754039989044,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007130762466157832,
      "loss": 4.8119,
      "step": 12397
    },
    {
      "epoch": 1.6978909887702,
      "grad_norm": 0.375,
      "learning_rate": 0.0007130430098016635,
      "loss": 4.7486,
      "step": 12398
    },
    {
      "epoch": 1.6980279375513558,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007130097708076437,
      "loss": 4.7467,
      "step": 12399
    },
    {
      "epoch": 1.6981648863325116,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007129765296339993,
      "loss": 4.8015,
      "step": 12400
    },
    {
      "epoch": 1.6983018351136674,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007129432862810058,
      "loss": 4.8638,
      "step": 12401
    },
    {
      "epoch": 1.6984387838948234,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007129100407489386,
      "loss": 4.8354,
      "step": 12402
    },
    {
      "epoch": 1.6985757326759792,
      "grad_norm": 0.375,
      "learning_rate": 0.0007128767930380732,
      "loss": 4.7784,
      "step": 12403
    },
    {
      "epoch": 1.6987126814571352,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007128435431486851,
      "loss": 4.7577,
      "step": 12404
    },
    {
      "epoch": 1.698849630238291,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007128102910810501,
      "loss": 4.8288,
      "step": 12405
    },
    {
      "epoch": 1.6989865790194467,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007127770368354436,
      "loss": 4.8066,
      "step": 12406
    },
    {
      "epoch": 1.6991235278006025,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000712743780412141,
      "loss": 4.7239,
      "step": 12407
    },
    {
      "epoch": 1.6992604765817583,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007127105218114182,
      "loss": 4.7913,
      "step": 12408
    },
    {
      "epoch": 1.6993974253629143,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007126772610335507,
      "loss": 4.7944,
      "step": 12409
    },
    {
      "epoch": 1.6995343741440703,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007126439980788141,
      "loss": 4.8699,
      "step": 12410
    },
    {
      "epoch": 1.699671322925226,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0007126107329474841,
      "loss": 4.7795,
      "step": 12411
    },
    {
      "epoch": 1.6998082717063818,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007125774656398364,
      "loss": 4.8222,
      "step": 12412
    },
    {
      "epoch": 1.6999452204875376,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007125441961561466,
      "loss": 4.7838,
      "step": 12413
    },
    {
      "epoch": 1.7000821692686934,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007125109244966908,
      "loss": 4.8184,
      "step": 12414
    },
    {
      "epoch": 1.7002191180498494,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007124776506617441,
      "loss": 4.8883,
      "step": 12415
    },
    {
      "epoch": 1.7003560668310052,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007124443746515827,
      "loss": 4.8253,
      "step": 12416
    },
    {
      "epoch": 1.7004930156121612,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007124110964664821,
      "loss": 4.7516,
      "step": 12417
    },
    {
      "epoch": 1.700629964393317,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007123778161067183,
      "loss": 4.8252,
      "step": 12418
    },
    {
      "epoch": 1.7007669131744727,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007123445335725669,
      "loss": 4.8131,
      "step": 12419
    },
    {
      "epoch": 1.7009038619556285,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007123112488643039,
      "loss": 4.7911,
      "step": 12420
    },
    {
      "epoch": 1.7010408107367845,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007122779619822052,
      "loss": 4.7714,
      "step": 12421
    },
    {
      "epoch": 1.7011777595179403,
      "grad_norm": 0.375,
      "learning_rate": 0.0007122446729265464,
      "loss": 4.8144,
      "step": 12422
    },
    {
      "epoch": 1.7013147082990963,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007122113816976037,
      "loss": 4.779,
      "step": 12423
    },
    {
      "epoch": 1.701451657080252,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007121780882956526,
      "loss": 4.8464,
      "step": 12424
    },
    {
      "epoch": 1.7015886058614078,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007121447927209693,
      "loss": 4.8078,
      "step": 12425
    },
    {
      "epoch": 1.7017255546425636,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007121114949738296,
      "loss": 4.8456,
      "step": 12426
    },
    {
      "epoch": 1.7018625034237194,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007120781950545094,
      "loss": 4.8505,
      "step": 12427
    },
    {
      "epoch": 1.7019994522048754,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000712044892963285,
      "loss": 4.7852,
      "step": 12428
    },
    {
      "epoch": 1.7021364009860314,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007120115887004319,
      "loss": 4.8216,
      "step": 12429
    },
    {
      "epoch": 1.7022733497671871,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007119782822662264,
      "loss": 4.8266,
      "step": 12430
    },
    {
      "epoch": 1.702410298548343,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007119449736609445,
      "loss": 4.8027,
      "step": 12431
    },
    {
      "epoch": 1.7025472473294987,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007119116628848621,
      "loss": 4.8774,
      "step": 12432
    },
    {
      "epoch": 1.7026841961106545,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007118783499382553,
      "loss": 4.8872,
      "step": 12433
    },
    {
      "epoch": 1.7028211448918105,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007118450348214003,
      "loss": 4.7135,
      "step": 12434
    },
    {
      "epoch": 1.7029580936729665,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007118117175345731,
      "loss": 4.8551,
      "step": 12435
    },
    {
      "epoch": 1.7030950424541222,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007117783980780497,
      "loss": 4.805,
      "step": 12436
    },
    {
      "epoch": 1.703231991235278,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007117450764521065,
      "loss": 4.6778,
      "step": 12437
    },
    {
      "epoch": 1.7033689400164338,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0007117117526570194,
      "loss": 4.8178,
      "step": 12438
    },
    {
      "epoch": 1.7035058887975896,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007116784266930648,
      "loss": 4.7663,
      "step": 12439
    },
    {
      "epoch": 1.7036428375787456,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007116450985605184,
      "loss": 4.8707,
      "step": 12440
    },
    {
      "epoch": 1.7037797863599013,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007116117682596569,
      "loss": 4.811,
      "step": 12441
    },
    {
      "epoch": 1.7039167351410573,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007115784357907564,
      "loss": 4.8484,
      "step": 12442
    },
    {
      "epoch": 1.7040536839222131,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000711545101154093,
      "loss": 4.8801,
      "step": 12443
    },
    {
      "epoch": 1.704190632703369,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007115117643499431,
      "loss": 4.8381,
      "step": 12444
    },
    {
      "epoch": 1.7043275814845247,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007114784253785827,
      "loss": 4.7812,
      "step": 12445
    },
    {
      "epoch": 1.7044645302656807,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007114450842402885,
      "loss": 4.8418,
      "step": 12446
    },
    {
      "epoch": 1.7046014790468365,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007114117409353364,
      "loss": 4.7986,
      "step": 12447
    },
    {
      "epoch": 1.7047384278279925,
      "grad_norm": 0.359375,
      "learning_rate": 0.000711378395464003,
      "loss": 4.6991,
      "step": 12448
    },
    {
      "epoch": 1.7048753766091482,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007113450478265646,
      "loss": 4.8549,
      "step": 12449
    },
    {
      "epoch": 1.705012325390304,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007113116980232975,
      "loss": 4.8332,
      "step": 12450
    },
    {
      "epoch": 1.7051492741714598,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007112783460544781,
      "loss": 4.8476,
      "step": 12451
    },
    {
      "epoch": 1.7052862229526156,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007112449919203827,
      "loss": 4.8207,
      "step": 12452
    },
    {
      "epoch": 1.7054231717337716,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007112116356212879,
      "loss": 4.8479,
      "step": 12453
    },
    {
      "epoch": 1.7055601205149276,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00071117827715747,
      "loss": 4.8703,
      "step": 12454
    },
    {
      "epoch": 1.7056970692960833,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007111449165292054,
      "loss": 4.797,
      "step": 12455
    },
    {
      "epoch": 1.7058340180772391,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007111115537367708,
      "loss": 4.9104,
      "step": 12456
    },
    {
      "epoch": 1.7059709668583949,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007110781887804424,
      "loss": 4.8169,
      "step": 12457
    },
    {
      "epoch": 1.7061079156395507,
      "grad_norm": 0.375,
      "learning_rate": 0.0007110448216604969,
      "loss": 4.7124,
      "step": 12458
    },
    {
      "epoch": 1.7062448644207067,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007110114523772108,
      "loss": 4.7995,
      "step": 12459
    },
    {
      "epoch": 1.7063818132018627,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007109780809308605,
      "loss": 4.785,
      "step": 12460
    },
    {
      "epoch": 1.7065187619830184,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007109447073217226,
      "loss": 4.8291,
      "step": 12461
    },
    {
      "epoch": 1.7066557107641742,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007109113315500738,
      "loss": 4.7406,
      "step": 12462
    },
    {
      "epoch": 1.70679265954533,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007108779536161907,
      "loss": 4.7704,
      "step": 12463
    },
    {
      "epoch": 1.7069296083264858,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007108445735203498,
      "loss": 4.7876,
      "step": 12464
    },
    {
      "epoch": 1.7070665571076418,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007108111912628277,
      "loss": 4.8487,
      "step": 12465
    },
    {
      "epoch": 1.7072035058887975,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007107778068439013,
      "loss": 4.7056,
      "step": 12466
    },
    {
      "epoch": 1.7073404546699535,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007107444202638469,
      "loss": 4.8514,
      "step": 12467
    },
    {
      "epoch": 1.7074774034511093,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007107110315229414,
      "loss": 4.7794,
      "step": 12468
    },
    {
      "epoch": 1.707614352232265,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007106776406214614,
      "loss": 4.8592,
      "step": 12469
    },
    {
      "epoch": 1.7077513010134209,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007106442475596839,
      "loss": 4.7815,
      "step": 12470
    },
    {
      "epoch": 1.7078882497945769,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007106108523378851,
      "loss": 4.7787,
      "step": 12471
    },
    {
      "epoch": 1.7080251985757327,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007105774549563422,
      "loss": 4.7984,
      "step": 12472
    },
    {
      "epoch": 1.7081621473568886,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000710544055415332,
      "loss": 4.8015,
      "step": 12473
    },
    {
      "epoch": 1.7082990961380444,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007105106537151308,
      "loss": 4.769,
      "step": 12474
    },
    {
      "epoch": 1.7084360449192002,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000710477249856016,
      "loss": 4.8515,
      "step": 12475
    },
    {
      "epoch": 1.708572993700356,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007104438438382641,
      "loss": 4.841,
      "step": 12476
    },
    {
      "epoch": 1.7087099424815118,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007104104356621519,
      "loss": 4.8449,
      "step": 12477
    },
    {
      "epoch": 1.7088468912626678,
      "grad_norm": 0.375,
      "learning_rate": 0.0007103770253279564,
      "loss": 4.8524,
      "step": 12478
    },
    {
      "epoch": 1.7089838400438238,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007103436128359546,
      "loss": 4.8036,
      "step": 12479
    },
    {
      "epoch": 1.7091207888249795,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0007103101981864231,
      "loss": 4.8314,
      "step": 12480
    },
    {
      "epoch": 1.7092577376061353,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007102767813796391,
      "loss": 4.7608,
      "step": 12481
    },
    {
      "epoch": 1.709394686387291,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007102433624158792,
      "loss": 4.805,
      "step": 12482
    },
    {
      "epoch": 1.7095316351684469,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007102099412954207,
      "loss": 4.7723,
      "step": 12483
    },
    {
      "epoch": 1.7096685839496029,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007101765180185403,
      "loss": 4.8359,
      "step": 12484
    },
    {
      "epoch": 1.7098055327307589,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007101430925855151,
      "loss": 4.8567,
      "step": 12485
    },
    {
      "epoch": 1.7099424815119146,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007101096649966221,
      "loss": 4.8152,
      "step": 12486
    },
    {
      "epoch": 1.7100794302930704,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007100762352521384,
      "loss": 4.8303,
      "step": 12487
    },
    {
      "epoch": 1.7102163790742262,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000710042803352341,
      "loss": 4.824,
      "step": 12488
    },
    {
      "epoch": 1.710353327855382,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007100093692975069,
      "loss": 4.8092,
      "step": 12489
    },
    {
      "epoch": 1.710490276636538,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007099759330879131,
      "loss": 4.8408,
      "step": 12490
    },
    {
      "epoch": 1.7106272254176937,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007099424947238368,
      "loss": 4.864,
      "step": 12491
    },
    {
      "epoch": 1.7107641741988497,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007099090542055552,
      "loss": 4.7583,
      "step": 12492
    },
    {
      "epoch": 1.7109011229800055,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007098756115333454,
      "loss": 4.8585,
      "step": 12493
    },
    {
      "epoch": 1.7110380717611613,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007098421667074842,
      "loss": 4.9048,
      "step": 12494
    },
    {
      "epoch": 1.711175020542317,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007098087197282494,
      "loss": 4.8141,
      "step": 12495
    },
    {
      "epoch": 1.711311969323473,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007097752705959176,
      "loss": 4.7902,
      "step": 12496
    },
    {
      "epoch": 1.7114489181046288,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007097418193107663,
      "loss": 4.8342,
      "step": 12497
    },
    {
      "epoch": 1.7115858668857848,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007097083658730726,
      "loss": 4.8674,
      "step": 12498
    },
    {
      "epoch": 1.7117228156669406,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007096749102831138,
      "loss": 4.8668,
      "step": 12499
    },
    {
      "epoch": 1.7118597644480964,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007096414525411672,
      "loss": 4.8017,
      "step": 12500
    },
    {
      "epoch": 1.7119967132292522,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007096079926475099,
      "loss": 4.7638,
      "step": 12501
    },
    {
      "epoch": 1.712133662010408,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007095745306024195,
      "loss": 4.8996,
      "step": 12502
    },
    {
      "epoch": 1.712270610791564,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007095410664061729,
      "loss": 4.8226,
      "step": 12503
    },
    {
      "epoch": 1.71240755957272,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0007095076000590478,
      "loss": 4.7806,
      "step": 12504
    },
    {
      "epoch": 1.7125445083538757,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007094741315613213,
      "loss": 4.8643,
      "step": 12505
    },
    {
      "epoch": 1.7126814571350315,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007094406609132709,
      "loss": 4.7462,
      "step": 12506
    },
    {
      "epoch": 1.7128184059161873,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007094071881151738,
      "loss": 4.8637,
      "step": 12507
    },
    {
      "epoch": 1.712955354697343,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007093737131673077,
      "loss": 4.7825,
      "step": 12508
    },
    {
      "epoch": 1.713092303478499,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0007093402360699498,
      "loss": 4.8509,
      "step": 12509
    },
    {
      "epoch": 1.713229252259655,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007093067568233772,
      "loss": 4.7841,
      "step": 12510
    },
    {
      "epoch": 1.7133662010408108,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007092732754278682,
      "loss": 4.8458,
      "step": 12511
    },
    {
      "epoch": 1.7135031498219666,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007092397918836995,
      "loss": 4.7695,
      "step": 12512
    },
    {
      "epoch": 1.7136400986031224,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007092063061911491,
      "loss": 4.8252,
      "step": 12513
    },
    {
      "epoch": 1.7137770473842782,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000709172818350494,
      "loss": 4.8137,
      "step": 12514
    },
    {
      "epoch": 1.7139139961654342,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0007091393283620122,
      "loss": 4.9046,
      "step": 12515
    },
    {
      "epoch": 1.71405094494659,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000709105836225981,
      "loss": 4.8719,
      "step": 12516
    },
    {
      "epoch": 1.714187893727746,
      "grad_norm": 0.375,
      "learning_rate": 0.000709072341942678,
      "loss": 4.7549,
      "step": 12517
    },
    {
      "epoch": 1.7143248425089017,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007090388455123806,
      "loss": 4.8013,
      "step": 12518
    },
    {
      "epoch": 1.7144617912900575,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007090053469353668,
      "loss": 4.8142,
      "step": 12519
    },
    {
      "epoch": 1.7145987400712133,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007089718462119139,
      "loss": 4.8434,
      "step": 12520
    },
    {
      "epoch": 1.7147356888523693,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007089383433422995,
      "loss": 4.8632,
      "step": 12521
    },
    {
      "epoch": 1.714872637633525,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007089048383268014,
      "loss": 4.7772,
      "step": 12522
    },
    {
      "epoch": 1.715009586414681,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007088713311656973,
      "loss": 4.889,
      "step": 12523
    },
    {
      "epoch": 1.7151465351958368,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007088378218592647,
      "loss": 4.8158,
      "step": 12524
    },
    {
      "epoch": 1.7152834839769926,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007088043104077815,
      "loss": 4.7721,
      "step": 12525
    },
    {
      "epoch": 1.7154204327581484,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007087707968115251,
      "loss": 4.8661,
      "step": 12526
    },
    {
      "epoch": 1.7155573815393041,
      "grad_norm": 0.40625,
      "learning_rate": 0.0007087372810707736,
      "loss": 4.7965,
      "step": 12527
    },
    {
      "epoch": 1.7156943303204601,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0007087037631858045,
      "loss": 4.7815,
      "step": 12528
    },
    {
      "epoch": 1.7158312791016161,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007086702431568959,
      "loss": 4.8322,
      "step": 12529
    },
    {
      "epoch": 1.715968227882772,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007086367209843252,
      "loss": 4.8322,
      "step": 12530
    },
    {
      "epoch": 1.7161051766639277,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007086031966683703,
      "loss": 4.7585,
      "step": 12531
    },
    {
      "epoch": 1.7162421254450835,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007085696702093091,
      "loss": 4.791,
      "step": 12532
    },
    {
      "epoch": 1.7163790742262393,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0007085361416074195,
      "loss": 4.8686,
      "step": 12533
    },
    {
      "epoch": 1.7165160230073953,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007085026108629792,
      "loss": 4.8556,
      "step": 12534
    },
    {
      "epoch": 1.716652971788551,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0007084690779762664,
      "loss": 4.7843,
      "step": 12535
    },
    {
      "epoch": 1.716789920569707,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007084355429475585,
      "loss": 4.7893,
      "step": 12536
    },
    {
      "epoch": 1.7169268693508628,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007084020057771338,
      "loss": 4.8443,
      "step": 12537
    },
    {
      "epoch": 1.7170638181320186,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007083684664652701,
      "loss": 4.792,
      "step": 12538
    },
    {
      "epoch": 1.7172007669131744,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007083349250122452,
      "loss": 4.7752,
      "step": 12539
    },
    {
      "epoch": 1.7173377156943304,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007083013814183374,
      "loss": 4.7938,
      "step": 12540
    },
    {
      "epoch": 1.7174746644754861,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007082678356838244,
      "loss": 4.7615,
      "step": 12541
    },
    {
      "epoch": 1.7176116132566421,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007082342878089844,
      "loss": 4.6672,
      "step": 12542
    },
    {
      "epoch": 1.717748562037798,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007082007377940953,
      "loss": 4.8049,
      "step": 12543
    },
    {
      "epoch": 1.7178855108189537,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007081671856394351,
      "loss": 4.7411,
      "step": 12544
    },
    {
      "epoch": 1.7180224596001095,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007081336313452819,
      "loss": 4.81,
      "step": 12545
    },
    {
      "epoch": 1.7181594083812652,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007081000749119138,
      "loss": 4.8036,
      "step": 12546
    },
    {
      "epoch": 1.7182963571624212,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000708066516339609,
      "loss": 4.8144,
      "step": 12547
    },
    {
      "epoch": 1.7184333059435772,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007080329556286454,
      "loss": 4.8161,
      "step": 12548
    },
    {
      "epoch": 1.718570254724733,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007079993927793012,
      "loss": 4.7614,
      "step": 12549
    },
    {
      "epoch": 1.7187072035058888,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007079658277918546,
      "loss": 4.8107,
      "step": 12550
    },
    {
      "epoch": 1.7188441522870446,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007079322606665836,
      "loss": 4.8138,
      "step": 12551
    },
    {
      "epoch": 1.7189811010682003,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007078986914037665,
      "loss": 4.8717,
      "step": 12552
    },
    {
      "epoch": 1.7191180498493563,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007078651200036817,
      "loss": 4.7922,
      "step": 12553
    },
    {
      "epoch": 1.7192549986305123,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000707831546466607,
      "loss": 4.8109,
      "step": 12554
    },
    {
      "epoch": 1.7193919474116681,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007077979707928209,
      "loss": 4.8296,
      "step": 12555
    },
    {
      "epoch": 1.719528896192824,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007077643929826017,
      "loss": 4.7898,
      "step": 12556
    },
    {
      "epoch": 1.7196658449739797,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007077308130362273,
      "loss": 4.797,
      "step": 12557
    },
    {
      "epoch": 1.7198027937551355,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007076972309539765,
      "loss": 4.7539,
      "step": 12558
    },
    {
      "epoch": 1.7199397425362914,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007076636467361272,
      "loss": 4.8802,
      "step": 12559
    },
    {
      "epoch": 1.7200766913174472,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007076300603829579,
      "loss": 4.8336,
      "step": 12560
    },
    {
      "epoch": 1.7202136400986032,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007075964718947468,
      "loss": 4.8749,
      "step": 12561
    },
    {
      "epoch": 1.720350588879759,
      "grad_norm": 0.375,
      "learning_rate": 0.0007075628812717725,
      "loss": 4.8449,
      "step": 12562
    },
    {
      "epoch": 1.7204875376609148,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007075292885143131,
      "loss": 4.7724,
      "step": 12563
    },
    {
      "epoch": 1.7206244864420706,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0007074956936226471,
      "loss": 4.8143,
      "step": 12564
    },
    {
      "epoch": 1.7207614352232266,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007074620965970531,
      "loss": 4.8105,
      "step": 12565
    },
    {
      "epoch": 1.7208983840043823,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007074284974378092,
      "loss": 4.6976,
      "step": 12566
    },
    {
      "epoch": 1.7210353327855383,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007073948961451943,
      "loss": 4.7307,
      "step": 12567
    },
    {
      "epoch": 1.721172281566694,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007073612927194862,
      "loss": 4.7738,
      "step": 12568
    },
    {
      "epoch": 1.7213092303478499,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007073276871609639,
      "loss": 4.8362,
      "step": 12569
    },
    {
      "epoch": 1.7214461791290057,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007072940794699057,
      "loss": 4.8212,
      "step": 12570
    },
    {
      "epoch": 1.7215831279101614,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0007072604696465901,
      "loss": 4.7283,
      "step": 12571
    },
    {
      "epoch": 1.7217200766913174,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007072268576912958,
      "loss": 4.8249,
      "step": 12572
    },
    {
      "epoch": 1.7218570254724734,
      "grad_norm": 0.328125,
      "learning_rate": 0.0007071932436043012,
      "loss": 4.7525,
      "step": 12573
    },
    {
      "epoch": 1.7219939742536292,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007071596273858848,
      "loss": 4.7846,
      "step": 12574
    },
    {
      "epoch": 1.722130923034785,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007071260090363254,
      "loss": 4.8193,
      "step": 12575
    },
    {
      "epoch": 1.7222678718159408,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007070923885559014,
      "loss": 4.7313,
      "step": 12576
    },
    {
      "epoch": 1.7224048205970965,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0007070587659448917,
      "loss": 4.8732,
      "step": 12577
    },
    {
      "epoch": 1.7225417693782525,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007070251412035746,
      "loss": 4.8304,
      "step": 12578
    },
    {
      "epoch": 1.7226787181594085,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000706991514332229,
      "loss": 4.7545,
      "step": 12579
    },
    {
      "epoch": 1.7228156669405643,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007069578853311335,
      "loss": 4.8568,
      "step": 12580
    },
    {
      "epoch": 1.72295261572172,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007069242542005667,
      "loss": 4.9081,
      "step": 12581
    },
    {
      "epoch": 1.7230895645028759,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007068906209408073,
      "loss": 4.8715,
      "step": 12582
    },
    {
      "epoch": 1.7232265132840316,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007068569855521342,
      "loss": 4.7765,
      "step": 12583
    },
    {
      "epoch": 1.7233634620651876,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007068233480348261,
      "loss": 4.8432,
      "step": 12584
    },
    {
      "epoch": 1.7235004108463434,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007067897083891618,
      "loss": 4.7541,
      "step": 12585
    },
    {
      "epoch": 1.7236373596274994,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007067560666154198,
      "loss": 4.7785,
      "step": 12586
    },
    {
      "epoch": 1.7237743084086552,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007067224227138791,
      "loss": 4.7492,
      "step": 12587
    },
    {
      "epoch": 1.723911257189811,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0007066887766848186,
      "loss": 4.7661,
      "step": 12588
    },
    {
      "epoch": 1.7240482059709668,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000706655128528517,
      "loss": 4.8443,
      "step": 12589
    },
    {
      "epoch": 1.7241851547521228,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007066214782452532,
      "loss": 4.773,
      "step": 12590
    },
    {
      "epoch": 1.7243221035332785,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007065878258353061,
      "loss": 4.8435,
      "step": 12591
    },
    {
      "epoch": 1.7244590523144345,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007065541712989545,
      "loss": 4.7194,
      "step": 12592
    },
    {
      "epoch": 1.7245960010955903,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007065205146364774,
      "loss": 4.7913,
      "step": 12593
    },
    {
      "epoch": 1.724732949876746,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007064868558481536,
      "loss": 4.8362,
      "step": 12594
    },
    {
      "epoch": 1.7248698986579019,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007064531949342621,
      "loss": 4.8219,
      "step": 12595
    },
    {
      "epoch": 1.7250068474390576,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007064195318950819,
      "loss": 4.7971,
      "step": 12596
    },
    {
      "epoch": 1.7251437962202136,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000706385866730892,
      "loss": 4.8447,
      "step": 12597
    },
    {
      "epoch": 1.7252807450013696,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007063521994419712,
      "loss": 4.8211,
      "step": 12598
    },
    {
      "epoch": 1.7254176937825254,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007063185300285988,
      "loss": 4.8617,
      "step": 12599
    },
    {
      "epoch": 1.7255546425636812,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007062848584910536,
      "loss": 4.6985,
      "step": 12600
    },
    {
      "epoch": 1.725691591344837,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007062511848296147,
      "loss": 4.7753,
      "step": 12601
    },
    {
      "epoch": 1.7258285401259927,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007062175090445611,
      "loss": 4.7651,
      "step": 12602
    },
    {
      "epoch": 1.7259654889071487,
      "grad_norm": 0.359375,
      "learning_rate": 0.000706183831136172,
      "loss": 4.8485,
      "step": 12603
    },
    {
      "epoch": 1.7261024376883047,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007061501511047264,
      "loss": 4.762,
      "step": 12604
    },
    {
      "epoch": 1.7262393864694605,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007061164689505037,
      "loss": 4.7945,
      "step": 12605
    },
    {
      "epoch": 1.7263763352506163,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007060827846737826,
      "loss": 4.8283,
      "step": 12606
    },
    {
      "epoch": 1.726513284031772,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007060490982748424,
      "loss": 4.7596,
      "step": 12607
    },
    {
      "epoch": 1.7266502328129278,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007060154097539623,
      "loss": 4.9213,
      "step": 12608
    },
    {
      "epoch": 1.7267871815940838,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0007059817191114215,
      "loss": 4.7349,
      "step": 12609
    },
    {
      "epoch": 1.7269241303752396,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007059480263474992,
      "loss": 4.858,
      "step": 12610
    },
    {
      "epoch": 1.7270610791563956,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007059143314624747,
      "loss": 4.7675,
      "step": 12611
    },
    {
      "epoch": 1.7271980279375514,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007058806344566272,
      "loss": 4.8195,
      "step": 12612
    },
    {
      "epoch": 1.7273349767187072,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007058469353302357,
      "loss": 4.7874,
      "step": 12613
    },
    {
      "epoch": 1.727471925499863,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007058132340835799,
      "loss": 4.7534,
      "step": 12614
    },
    {
      "epoch": 1.727608874281019,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007057795307169387,
      "loss": 4.737,
      "step": 12615
    },
    {
      "epoch": 1.7277458230621747,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007057458252305918,
      "loss": 4.7494,
      "step": 12616
    },
    {
      "epoch": 1.7278827718433307,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000705712117624818,
      "loss": 4.7763,
      "step": 12617
    },
    {
      "epoch": 1.7280197206244865,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007056784078998972,
      "loss": 4.814,
      "step": 12618
    },
    {
      "epoch": 1.7281566694056423,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007056446960561084,
      "loss": 4.8191,
      "step": 12619
    },
    {
      "epoch": 1.728293618186798,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007056109820937312,
      "loss": 4.828,
      "step": 12620
    },
    {
      "epoch": 1.7284305669679538,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007055772660130448,
      "loss": 4.7649,
      "step": 12621
    },
    {
      "epoch": 1.7285675157491098,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007055435478143285,
      "loss": 4.8536,
      "step": 12622
    },
    {
      "epoch": 1.7287044645302658,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007055098274978622,
      "loss": 4.7883,
      "step": 12623
    },
    {
      "epoch": 1.7288414133114216,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007054761050639249,
      "loss": 4.8266,
      "step": 12624
    },
    {
      "epoch": 1.7289783620925774,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007054423805127964,
      "loss": 4.8353,
      "step": 12625
    },
    {
      "epoch": 1.7291153108737332,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000705408653844756,
      "loss": 4.7353,
      "step": 12626
    },
    {
      "epoch": 1.729252259654889,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007053749250600833,
      "loss": 4.8725,
      "step": 12627
    },
    {
      "epoch": 1.729389208436045,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007053411941590575,
      "loss": 4.8861,
      "step": 12628
    },
    {
      "epoch": 1.729526157217201,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0007053074611419585,
      "loss": 4.853,
      "step": 12629
    },
    {
      "epoch": 1.7296631059983567,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007052737260090658,
      "loss": 4.758,
      "step": 12630
    },
    {
      "epoch": 1.7298000547795125,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007052399887606588,
      "loss": 4.7824,
      "step": 12631
    },
    {
      "epoch": 1.7299370035606683,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007052062493970172,
      "loss": 4.8323,
      "step": 12632
    },
    {
      "epoch": 1.730073952341824,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007051725079184207,
      "loss": 4.7726,
      "step": 12633
    },
    {
      "epoch": 1.73021090112298,
      "grad_norm": 0.34375,
      "learning_rate": 0.0007051387643251488,
      "loss": 4.8735,
      "step": 12634
    },
    {
      "epoch": 1.7303478499041358,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007051050186174811,
      "loss": 4.7543,
      "step": 12635
    },
    {
      "epoch": 1.7304847986852918,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007050712707956973,
      "loss": 4.7785,
      "step": 12636
    },
    {
      "epoch": 1.7306217474664476,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007050375208600771,
      "loss": 4.763,
      "step": 12637
    },
    {
      "epoch": 1.7307586962476034,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007050037688109002,
      "loss": 4.7813,
      "step": 12638
    },
    {
      "epoch": 1.7308956450287591,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007049700146484464,
      "loss": 4.8735,
      "step": 12639
    },
    {
      "epoch": 1.7310325938099151,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007049362583729953,
      "loss": 4.77,
      "step": 12640
    },
    {
      "epoch": 1.731169542591071,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007049024999848265,
      "loss": 4.7733,
      "step": 12641
    },
    {
      "epoch": 1.731306491372227,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007048687394842202,
      "loss": 4.8624,
      "step": 12642
    },
    {
      "epoch": 1.7314434401533827,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007048349768714558,
      "loss": 4.7612,
      "step": 12643
    },
    {
      "epoch": 1.7315803889345385,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0007048012121468132,
      "loss": 4.7361,
      "step": 12644
    },
    {
      "epoch": 1.7317173377156942,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007047674453105724,
      "loss": 4.8826,
      "step": 12645
    },
    {
      "epoch": 1.73185428649685,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0007047336763630128,
      "loss": 4.699,
      "step": 12646
    },
    {
      "epoch": 1.731991235278006,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007046999053044147,
      "loss": 4.8116,
      "step": 12647
    },
    {
      "epoch": 1.732128184059162,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007046661321350577,
      "loss": 4.8059,
      "step": 12648
    },
    {
      "epoch": 1.7322651328403178,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007046323568552219,
      "loss": 4.7835,
      "step": 12649
    },
    {
      "epoch": 1.7324020816214736,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007045985794651871,
      "loss": 4.8348,
      "step": 12650
    },
    {
      "epoch": 1.7325390304026294,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000704564799965233,
      "loss": 4.769,
      "step": 12651
    },
    {
      "epoch": 1.7326759791837851,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00070453101835564,
      "loss": 4.7669,
      "step": 12652
    },
    {
      "epoch": 1.7328129279649411,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007044972346366877,
      "loss": 4.7863,
      "step": 12653
    },
    {
      "epoch": 1.732949876746097,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000704463448808656,
      "loss": 4.8582,
      "step": 12654
    },
    {
      "epoch": 1.733086825527253,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007044296608718253,
      "loss": 4.8092,
      "step": 12655
    },
    {
      "epoch": 1.7332237743084087,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007043958708264753,
      "loss": 4.7493,
      "step": 12656
    },
    {
      "epoch": 1.7333607230895645,
      "grad_norm": 0.375,
      "learning_rate": 0.0007043620786728861,
      "loss": 4.7455,
      "step": 12657
    },
    {
      "epoch": 1.7334976718707202,
      "grad_norm": 0.4375,
      "learning_rate": 0.0007043282844113377,
      "loss": 4.8508,
      "step": 12658
    },
    {
      "epoch": 1.7336346206518762,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007042944880421101,
      "loss": 4.8139,
      "step": 12659
    },
    {
      "epoch": 1.733771569433032,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007042606895654838,
      "loss": 4.7559,
      "step": 12660
    },
    {
      "epoch": 1.733908518214188,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007042268889817384,
      "loss": 4.756,
      "step": 12661
    },
    {
      "epoch": 1.7340454669953438,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007041930862911541,
      "loss": 4.8003,
      "step": 12662
    },
    {
      "epoch": 1.7341824157764996,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007041592814940113,
      "loss": 4.8069,
      "step": 12663
    },
    {
      "epoch": 1.7343193645576553,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0007041254745905898,
      "loss": 4.7233,
      "step": 12664
    },
    {
      "epoch": 1.7344563133388113,
      "grad_norm": 0.390625,
      "learning_rate": 0.00070409166558117,
      "loss": 4.8078,
      "step": 12665
    },
    {
      "epoch": 1.7345932621199671,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000704057854466032,
      "loss": 4.8421,
      "step": 12666
    },
    {
      "epoch": 1.7347302109011231,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007040240412454561,
      "loss": 4.8103,
      "step": 12667
    },
    {
      "epoch": 1.734867159682279,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007039902259197224,
      "loss": 4.7793,
      "step": 12668
    },
    {
      "epoch": 1.7350041084634347,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007039564084891112,
      "loss": 4.8107,
      "step": 12669
    },
    {
      "epoch": 1.7351410572445904,
      "grad_norm": 0.375,
      "learning_rate": 0.0007039225889539026,
      "loss": 4.7768,
      "step": 12670
    },
    {
      "epoch": 1.7352780060257462,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007038887673143772,
      "loss": 4.7916,
      "step": 12671
    },
    {
      "epoch": 1.7354149548069022,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007038549435708151,
      "loss": 4.7719,
      "step": 12672
    },
    {
      "epoch": 1.7355519035880582,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007038211177234963,
      "loss": 4.7198,
      "step": 12673
    },
    {
      "epoch": 1.735688852369214,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007037872897727018,
      "loss": 4.822,
      "step": 12674
    },
    {
      "epoch": 1.7358258011503698,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007037534597187112,
      "loss": 4.8455,
      "step": 12675
    },
    {
      "epoch": 1.7359627499315256,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007037196275618054,
      "loss": 4.8127,
      "step": 12676
    },
    {
      "epoch": 1.7360996987126813,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007036857933022645,
      "loss": 4.7947,
      "step": 12677
    },
    {
      "epoch": 1.7362366474938373,
      "grad_norm": 0.40625,
      "learning_rate": 0.000703651956940369,
      "loss": 4.8235,
      "step": 12678
    },
    {
      "epoch": 1.736373596274993,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007036181184763994,
      "loss": 4.7729,
      "step": 12679
    },
    {
      "epoch": 1.736510545056149,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007035842779106358,
      "loss": 4.7412,
      "step": 12680
    },
    {
      "epoch": 1.7366474938373049,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0007035504352433591,
      "loss": 4.7888,
      "step": 12681
    },
    {
      "epoch": 1.7367844426184607,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007035165904748494,
      "loss": 4.8396,
      "step": 12682
    },
    {
      "epoch": 1.7369213913996164,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007034827436053874,
      "loss": 4.8308,
      "step": 12683
    },
    {
      "epoch": 1.7370583401807724,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007034488946352534,
      "loss": 4.7783,
      "step": 12684
    },
    {
      "epoch": 1.7371952889619282,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007034150435647281,
      "loss": 4.7311,
      "step": 12685
    },
    {
      "epoch": 1.7373322377430842,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000703381190394092,
      "loss": 4.7447,
      "step": 12686
    },
    {
      "epoch": 1.73746918652424,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0007033473351236252,
      "loss": 4.7912,
      "step": 12687
    },
    {
      "epoch": 1.7376061353053958,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000703313477753609,
      "loss": 4.8264,
      "step": 12688
    },
    {
      "epoch": 1.7377430840865515,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007032796182843237,
      "loss": 4.686,
      "step": 12689
    },
    {
      "epoch": 1.7378800328677073,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007032457567160498,
      "loss": 4.8706,
      "step": 12690
    },
    {
      "epoch": 1.7380169816488633,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000703211893049068,
      "loss": 4.7969,
      "step": 12691
    },
    {
      "epoch": 1.7381539304300193,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007031780272836588,
      "loss": 4.7742,
      "step": 12692
    },
    {
      "epoch": 1.738290879211175,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000703144159420103,
      "loss": 4.8307,
      "step": 12693
    },
    {
      "epoch": 1.7384278279923309,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007031102894586812,
      "loss": 4.8386,
      "step": 12694
    },
    {
      "epoch": 1.7385647767734866,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007030764173996741,
      "loss": 4.809,
      "step": 12695
    },
    {
      "epoch": 1.7387017255546424,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007030425432433625,
      "loss": 4.7745,
      "step": 12696
    },
    {
      "epoch": 1.7388386743357984,
      "grad_norm": 0.359375,
      "learning_rate": 0.000703008666990027,
      "loss": 4.8511,
      "step": 12697
    },
    {
      "epoch": 1.7389756231169544,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007029747886399485,
      "loss": 4.8299,
      "step": 12698
    },
    {
      "epoch": 1.7391125718981102,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007029409081934074,
      "loss": 4.8674,
      "step": 12699
    },
    {
      "epoch": 1.739249520679266,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007029070256506849,
      "loss": 4.8272,
      "step": 12700
    },
    {
      "epoch": 1.7393864694604217,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007028731410120616,
      "loss": 4.6738,
      "step": 12701
    },
    {
      "epoch": 1.7395234182415775,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007028392542778183,
      "loss": 4.7627,
      "step": 12702
    },
    {
      "epoch": 1.7396603670227335,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007028053654482359,
      "loss": 4.7844,
      "step": 12703
    },
    {
      "epoch": 1.7397973158038893,
      "grad_norm": 0.390625,
      "learning_rate": 0.000702771474523595,
      "loss": 4.7882,
      "step": 12704
    },
    {
      "epoch": 1.7399342645850453,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000702737581504177,
      "loss": 4.7269,
      "step": 12705
    },
    {
      "epoch": 1.740071213366201,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007027036863902623,
      "loss": 4.9202,
      "step": 12706
    },
    {
      "epoch": 1.7402081621473569,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0007026697891821318,
      "loss": 4.8443,
      "step": 12707
    },
    {
      "epoch": 1.7403451109285126,
      "grad_norm": 0.421875,
      "learning_rate": 0.0007026358898800666,
      "loss": 4.8592,
      "step": 12708
    },
    {
      "epoch": 1.7404820597096686,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007026019884843477,
      "loss": 4.7932,
      "step": 12709
    },
    {
      "epoch": 1.7406190084908244,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007025680849952557,
      "loss": 4.807,
      "step": 12710
    },
    {
      "epoch": 1.7407559572719804,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007025341794130721,
      "loss": 4.8658,
      "step": 12711
    },
    {
      "epoch": 1.7408929060531362,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007025002717380775,
      "loss": 4.816,
      "step": 12712
    },
    {
      "epoch": 1.741029854834292,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007024663619705528,
      "loss": 4.8148,
      "step": 12713
    },
    {
      "epoch": 1.7411668036154477,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007024324501107793,
      "loss": 4.7818,
      "step": 12714
    },
    {
      "epoch": 1.7413037523966035,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000702398536159038,
      "loss": 4.777,
      "step": 12715
    },
    {
      "epoch": 1.7414407011777595,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007023646201156097,
      "loss": 4.7821,
      "step": 12716
    },
    {
      "epoch": 1.7415776499589155,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007023307019807758,
      "loss": 4.7731,
      "step": 12717
    },
    {
      "epoch": 1.7417145987400713,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0007022967817548173,
      "loss": 4.7756,
      "step": 12718
    },
    {
      "epoch": 1.741851547521227,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000702262859438015,
      "loss": 4.7704,
      "step": 12719
    },
    {
      "epoch": 1.7419884963023828,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007022289350306506,
      "loss": 4.7915,
      "step": 12720
    },
    {
      "epoch": 1.7421254450835386,
      "grad_norm": 0.375,
      "learning_rate": 0.0007021950085330048,
      "loss": 4.8551,
      "step": 12721
    },
    {
      "epoch": 1.7422623938646946,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007021610799453588,
      "loss": 4.7545,
      "step": 12722
    },
    {
      "epoch": 1.7423993426458506,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007021271492679939,
      "loss": 4.801,
      "step": 12723
    },
    {
      "epoch": 1.7425362914270064,
      "grad_norm": 0.359375,
      "learning_rate": 0.0007020932165011912,
      "loss": 4.908,
      "step": 12724
    },
    {
      "epoch": 1.7426732402081622,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000702059281645232,
      "loss": 4.8089,
      "step": 12725
    },
    {
      "epoch": 1.742810188989318,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007020253447003974,
      "loss": 4.783,
      "step": 12726
    },
    {
      "epoch": 1.7429471377704737,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0007019914056669687,
      "loss": 4.7054,
      "step": 12727
    },
    {
      "epoch": 1.7430840865516297,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007019574645452272,
      "loss": 4.8517,
      "step": 12728
    },
    {
      "epoch": 1.7432210353327855,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007019235213354541,
      "loss": 4.7809,
      "step": 12729
    },
    {
      "epoch": 1.7433579841139415,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007018895760379309,
      "loss": 4.8875,
      "step": 12730
    },
    {
      "epoch": 1.7434949328950973,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007018556286529387,
      "loss": 4.7734,
      "step": 12731
    },
    {
      "epoch": 1.743631881676253,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0007018216791807589,
      "loss": 4.8029,
      "step": 12732
    },
    {
      "epoch": 1.7437688304574088,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0007017877276216727,
      "loss": 4.8933,
      "step": 12733
    },
    {
      "epoch": 1.7439057792385648,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007017537739759617,
      "loss": 4.7312,
      "step": 12734
    },
    {
      "epoch": 1.7440427280197206,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0007017198182439073,
      "loss": 4.7789,
      "step": 12735
    },
    {
      "epoch": 1.7441796768008766,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007016858604257906,
      "loss": 4.7863,
      "step": 12736
    },
    {
      "epoch": 1.7443166255820324,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007016519005218933,
      "loss": 4.8107,
      "step": 12737
    },
    {
      "epoch": 1.7444535743631882,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007016179385324968,
      "loss": 4.8457,
      "step": 12738
    },
    {
      "epoch": 1.744590523144344,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007015839744578824,
      "loss": 4.7877,
      "step": 12739
    },
    {
      "epoch": 1.7447274719254997,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0007015500082983316,
      "loss": 4.7787,
      "step": 12740
    },
    {
      "epoch": 1.7448644207066557,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000701516040054126,
      "loss": 4.8629,
      "step": 12741
    },
    {
      "epoch": 1.7450013694878117,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0007014820697255471,
      "loss": 4.8309,
      "step": 12742
    },
    {
      "epoch": 1.7451383182689675,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0007014480973128762,
      "loss": 4.7877,
      "step": 12743
    },
    {
      "epoch": 1.7452752670501233,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007014141228163952,
      "loss": 4.7324,
      "step": 12744
    },
    {
      "epoch": 1.745412215831279,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007013801462363853,
      "loss": 4.809,
      "step": 12745
    },
    {
      "epoch": 1.7455491646124348,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007013461675731283,
      "loss": 4.8929,
      "step": 12746
    },
    {
      "epoch": 1.7456861133935908,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007013121868269058,
      "loss": 4.8261,
      "step": 12747
    },
    {
      "epoch": 1.7458230621747468,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007012782039979992,
      "loss": 4.7777,
      "step": 12748
    },
    {
      "epoch": 1.7459600109559026,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0007012442190866903,
      "loss": 4.8616,
      "step": 12749
    },
    {
      "epoch": 1.7460969597370584,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0007012102320932606,
      "loss": 4.9432,
      "step": 12750
    },
    {
      "epoch": 1.7462339085182141,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000701176243017992,
      "loss": 4.7773,
      "step": 12751
    },
    {
      "epoch": 1.74637085729937,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007011422518611658,
      "loss": 4.7999,
      "step": 12752
    },
    {
      "epoch": 1.746507806080526,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007011082586230641,
      "loss": 4.8757,
      "step": 12753
    },
    {
      "epoch": 1.7466447548616817,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0007010742633039684,
      "loss": 4.8946,
      "step": 12754
    },
    {
      "epoch": 1.7467817036428377,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007010402659041604,
      "loss": 4.7849,
      "step": 12755
    },
    {
      "epoch": 1.7469186524239935,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0007010062664239219,
      "loss": 4.7882,
      "step": 12756
    },
    {
      "epoch": 1.7470556012051492,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007009722648635348,
      "loss": 4.8041,
      "step": 12757
    },
    {
      "epoch": 1.747192549986305,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007009382612232805,
      "loss": 4.8301,
      "step": 12758
    },
    {
      "epoch": 1.747329498767461,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007009042555034412,
      "loss": 4.8966,
      "step": 12759
    },
    {
      "epoch": 1.7474664475486168,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0007008702477042985,
      "loss": 4.7607,
      "step": 12760
    },
    {
      "epoch": 1.7476033963297728,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007008362378261341,
      "loss": 4.7996,
      "step": 12761
    },
    {
      "epoch": 1.7477403451109286,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0007008022258692304,
      "loss": 4.7691,
      "step": 12762
    },
    {
      "epoch": 1.7478772938920843,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0007007682118338685,
      "loss": 4.8137,
      "step": 12763
    },
    {
      "epoch": 1.7480142426732401,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007007341957203311,
      "loss": 4.8143,
      "step": 12764
    },
    {
      "epoch": 1.748151191454396,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007007001775288994,
      "loss": 4.8193,
      "step": 12765
    },
    {
      "epoch": 1.748288140235552,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0007006661572598556,
      "loss": 4.7606,
      "step": 12766
    },
    {
      "epoch": 1.748425089016708,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0007006321349134816,
      "loss": 4.8914,
      "step": 12767
    },
    {
      "epoch": 1.7485620377978637,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0007005981104900595,
      "loss": 4.8185,
      "step": 12768
    },
    {
      "epoch": 1.7486989865790195,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000700564083989871,
      "loss": 4.8684,
      "step": 12769
    },
    {
      "epoch": 1.7488359353601752,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0007005300554131984,
      "loss": 4.7927,
      "step": 12770
    },
    {
      "epoch": 1.748972884141331,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0007004960247603235,
      "loss": 4.7874,
      "step": 12771
    },
    {
      "epoch": 1.749109832922487,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007004619920315282,
      "loss": 4.7467,
      "step": 12772
    },
    {
      "epoch": 1.7492467817036428,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0007004279572270949,
      "loss": 4.8044,
      "step": 12773
    },
    {
      "epoch": 1.7493837304847988,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007003939203473052,
      "loss": 4.7705,
      "step": 12774
    },
    {
      "epoch": 1.7495206792659546,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0007003598813924417,
      "loss": 4.769,
      "step": 12775
    },
    {
      "epoch": 1.7496576280471103,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000700325840362786,
      "loss": 4.7852,
      "step": 12776
    },
    {
      "epoch": 1.7497945768282661,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0007002917972586205,
      "loss": 4.8503,
      "step": 12777
    },
    {
      "epoch": 1.7499315256094221,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0007002577520802271,
      "loss": 4.7282,
      "step": 12778
    },
    {
      "epoch": 1.7500684743905779,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0007002237048278883,
      "loss": 4.8115,
      "step": 12779
    },
    {
      "epoch": 1.7502054231717339,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0007001896555018857,
      "loss": 4.7548,
      "step": 12780
    },
    {
      "epoch": 1.7503423719528897,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000700155604102502,
      "loss": 4.8304,
      "step": 12781
    },
    {
      "epoch": 1.7504793207340454,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0007001215506300193,
      "loss": 4.8677,
      "step": 12782
    },
    {
      "epoch": 1.7506162695152012,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0007000874950847196,
      "loss": 4.8032,
      "step": 12783
    },
    {
      "epoch": 1.7507532182963572,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0007000534374668854,
      "loss": 4.8374,
      "step": 12784
    },
    {
      "epoch": 1.750890167077513,
      "grad_norm": 0.390625,
      "learning_rate": 0.0007000193777767984,
      "loss": 4.8289,
      "step": 12785
    },
    {
      "epoch": 1.751027115858669,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006999853160147416,
      "loss": 4.7007,
      "step": 12786
    },
    {
      "epoch": 1.7511640646398248,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006999512521809968,
      "loss": 4.8779,
      "step": 12787
    },
    {
      "epoch": 1.7513010134209805,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006999171862758465,
      "loss": 4.8054,
      "step": 12788
    },
    {
      "epoch": 1.7514379622021363,
      "grad_norm": 0.375,
      "learning_rate": 0.0006998831182995727,
      "loss": 4.7812,
      "step": 12789
    },
    {
      "epoch": 1.751574910983292,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006998490482524581,
      "loss": 4.7679,
      "step": 12790
    },
    {
      "epoch": 1.751711859764448,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006998149761347849,
      "loss": 4.7937,
      "step": 12791
    },
    {
      "epoch": 1.751848808545604,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006997809019468355,
      "loss": 4.7801,
      "step": 12792
    },
    {
      "epoch": 1.7519857573267599,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006997468256888922,
      "loss": 4.8609,
      "step": 12793
    },
    {
      "epoch": 1.7521227061079157,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006997127473612374,
      "loss": 4.8009,
      "step": 12794
    },
    {
      "epoch": 1.7522596548890714,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006996786669641537,
      "loss": 4.7907,
      "step": 12795
    },
    {
      "epoch": 1.7523966036702272,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006996445844979232,
      "loss": 4.7027,
      "step": 12796
    },
    {
      "epoch": 1.7525335524513832,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006996104999628287,
      "loss": 4.7583,
      "step": 12797
    },
    {
      "epoch": 1.752670501232539,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006995764133591524,
      "loss": 4.8402,
      "step": 12798
    },
    {
      "epoch": 1.752807450013695,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006995423246871768,
      "loss": 4.8486,
      "step": 12799
    },
    {
      "epoch": 1.7529443987948508,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006995082339471847,
      "loss": 4.8384,
      "step": 12800
    },
    {
      "epoch": 1.7530813475760065,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006994741411394584,
      "loss": 4.7601,
      "step": 12801
    },
    {
      "epoch": 1.7532182963571623,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006994400462642805,
      "loss": 4.8603,
      "step": 12802
    },
    {
      "epoch": 1.7533552451383183,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006994059493219333,
      "loss": 4.8143,
      "step": 12803
    },
    {
      "epoch": 1.753492193919474,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006993718503126997,
      "loss": 4.8857,
      "step": 12804
    },
    {
      "epoch": 1.75362914270063,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000699337749236862,
      "loss": 4.7607,
      "step": 12805
    },
    {
      "epoch": 1.7537660914817859,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006993036460947031,
      "loss": 4.6744,
      "step": 12806
    },
    {
      "epoch": 1.7539030402629416,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006992695408865054,
      "loss": 4.8686,
      "step": 12807
    },
    {
      "epoch": 1.7540399890440974,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006992354336125517,
      "loss": 4.8942,
      "step": 12808
    },
    {
      "epoch": 1.7541769378252532,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006992013242731245,
      "loss": 4.733,
      "step": 12809
    },
    {
      "epoch": 1.7543138866064092,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006991672128685066,
      "loss": 4.763,
      "step": 12810
    },
    {
      "epoch": 1.7544508353875652,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006991330993989806,
      "loss": 4.8223,
      "step": 12811
    },
    {
      "epoch": 1.754587784168721,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0006990989838648292,
      "loss": 4.8191,
      "step": 12812
    },
    {
      "epoch": 1.7547247329498767,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006990648662663352,
      "loss": 4.8525,
      "step": 12813
    },
    {
      "epoch": 1.7548616817310325,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006990307466037813,
      "loss": 4.8466,
      "step": 12814
    },
    {
      "epoch": 1.7549986305121883,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006989966248774503,
      "loss": 4.8185,
      "step": 12815
    },
    {
      "epoch": 1.7551355792933443,
      "grad_norm": 0.375,
      "learning_rate": 0.000698962501087625,
      "loss": 4.8382,
      "step": 12816
    },
    {
      "epoch": 1.7552725280745003,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000698928375234588,
      "loss": 4.792,
      "step": 12817
    },
    {
      "epoch": 1.755409476855656,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006988942473186222,
      "loss": 4.7456,
      "step": 12818
    },
    {
      "epoch": 1.7555464256368118,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006988601173400106,
      "loss": 4.8844,
      "step": 12819
    },
    {
      "epoch": 1.7556833744179676,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006988259852990359,
      "loss": 4.8072,
      "step": 12820
    },
    {
      "epoch": 1.7558203231991234,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006987918511959808,
      "loss": 4.7989,
      "step": 12821
    },
    {
      "epoch": 1.7559572719802794,
      "grad_norm": 0.375,
      "learning_rate": 0.0006987577150311285,
      "loss": 4.7843,
      "step": 12822
    },
    {
      "epoch": 1.7560942207614352,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006987235768047618,
      "loss": 4.8113,
      "step": 12823
    },
    {
      "epoch": 1.7562311695425912,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006986894365171635,
      "loss": 4.8243,
      "step": 12824
    },
    {
      "epoch": 1.756368118323747,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006986552941686164,
      "loss": 4.8287,
      "step": 12825
    },
    {
      "epoch": 1.7565050671049027,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000698621149759404,
      "loss": 4.7896,
      "step": 12826
    },
    {
      "epoch": 1.7566420158860585,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006985870032898085,
      "loss": 4.7041,
      "step": 12827
    },
    {
      "epoch": 1.7567789646672145,
      "grad_norm": 0.375,
      "learning_rate": 0.0006985528547601133,
      "loss": 4.7689,
      "step": 12828
    },
    {
      "epoch": 1.7569159134483703,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006985187041706016,
      "loss": 4.7883,
      "step": 12829
    },
    {
      "epoch": 1.7570528622295263,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006984845515215559,
      "loss": 4.7387,
      "step": 12830
    },
    {
      "epoch": 1.757189811010682,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006984503968132599,
      "loss": 4.806,
      "step": 12831
    },
    {
      "epoch": 1.7573267597918378,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000698416240045996,
      "loss": 4.855,
      "step": 12832
    },
    {
      "epoch": 1.7574637085729936,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006983820812200476,
      "loss": 4.8703,
      "step": 12833
    },
    {
      "epoch": 1.7576006573541494,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006983479203356976,
      "loss": 4.7769,
      "step": 12834
    },
    {
      "epoch": 1.7577376061353054,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006983137573932294,
      "loss": 4.7354,
      "step": 12835
    },
    {
      "epoch": 1.7578745549164614,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006982795923929257,
      "loss": 4.8172,
      "step": 12836
    },
    {
      "epoch": 1.7580115036976172,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006982454253350699,
      "loss": 4.7894,
      "step": 12837
    },
    {
      "epoch": 1.758148452478773,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006982112562199452,
      "loss": 4.7392,
      "step": 12838
    },
    {
      "epoch": 1.7582854012599287,
      "grad_norm": 0.375,
      "learning_rate": 0.0006981770850478346,
      "loss": 4.7832,
      "step": 12839
    },
    {
      "epoch": 1.7584223500410845,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006981429118190214,
      "loss": 4.7782,
      "step": 12840
    },
    {
      "epoch": 1.7585592988222405,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006981087365337887,
      "loss": 4.8219,
      "step": 12841
    },
    {
      "epoch": 1.7586962476033965,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006980745591924199,
      "loss": 4.845,
      "step": 12842
    },
    {
      "epoch": 1.7588331963845523,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000698040379795198,
      "loss": 4.8327,
      "step": 12843
    },
    {
      "epoch": 1.758970145165708,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006980061983424066,
      "loss": 4.7995,
      "step": 12844
    },
    {
      "epoch": 1.7591070939468638,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006979720148343284,
      "loss": 4.7723,
      "step": 12845
    },
    {
      "epoch": 1.7592440427280196,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006979378292712472,
      "loss": 4.841,
      "step": 12846
    },
    {
      "epoch": 1.7593809915091756,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006979036416534461,
      "loss": 4.8259,
      "step": 12847
    },
    {
      "epoch": 1.7595179402903314,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006978694519812084,
      "loss": 4.8321,
      "step": 12848
    },
    {
      "epoch": 1.7596548890714874,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006978352602548176,
      "loss": 4.745,
      "step": 12849
    },
    {
      "epoch": 1.7597918378526431,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006978010664745569,
      "loss": 4.8905,
      "step": 12850
    },
    {
      "epoch": 1.759928786633799,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006977668706407098,
      "loss": 4.7812,
      "step": 12851
    },
    {
      "epoch": 1.7600657354149547,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006977326727535594,
      "loss": 4.8054,
      "step": 12852
    },
    {
      "epoch": 1.7602026841961107,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006976984728133894,
      "loss": 4.8013,
      "step": 12853
    },
    {
      "epoch": 1.7603396329772665,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006976642708204832,
      "loss": 4.7608,
      "step": 12854
    },
    {
      "epoch": 1.7604765817584225,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000697630066775124,
      "loss": 4.7991,
      "step": 12855
    },
    {
      "epoch": 1.7606135305395783,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006975958606775957,
      "loss": 4.8036,
      "step": 12856
    },
    {
      "epoch": 1.760750479320734,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006975616525281812,
      "loss": 4.6773,
      "step": 12857
    },
    {
      "epoch": 1.7608874281018898,
      "grad_norm": 0.375,
      "learning_rate": 0.0006975274423271646,
      "loss": 4.8391,
      "step": 12858
    },
    {
      "epoch": 1.7610243768830456,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006974932300748289,
      "loss": 4.8356,
      "step": 12859
    },
    {
      "epoch": 1.7611613256642016,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006974590157714579,
      "loss": 4.7935,
      "step": 12860
    },
    {
      "epoch": 1.7612982744453576,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006974247994173349,
      "loss": 4.7956,
      "step": 12861
    },
    {
      "epoch": 1.7614352232265134,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006973905810127439,
      "loss": 4.7363,
      "step": 12862
    },
    {
      "epoch": 1.7615721720076691,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000697356360557968,
      "loss": 4.7819,
      "step": 12863
    },
    {
      "epoch": 1.761709120788825,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006973221380532911,
      "loss": 4.7813,
      "step": 12864
    },
    {
      "epoch": 1.7618460695699807,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006972879134989967,
      "loss": 4.8062,
      "step": 12865
    },
    {
      "epoch": 1.7619830183511367,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006972536868953685,
      "loss": 4.9214,
      "step": 12866
    },
    {
      "epoch": 1.7621199671322927,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00069721945824269,
      "loss": 4.7887,
      "step": 12867
    },
    {
      "epoch": 1.7622569159134485,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006971852275412449,
      "loss": 4.7572,
      "step": 12868
    },
    {
      "epoch": 1.7623938646946042,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006971509947913169,
      "loss": 4.7139,
      "step": 12869
    },
    {
      "epoch": 1.76253081347576,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006971167599931898,
      "loss": 4.7325,
      "step": 12870
    },
    {
      "epoch": 1.7626677622569158,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006970825231471473,
      "loss": 4.7994,
      "step": 12871
    },
    {
      "epoch": 1.7628047110380718,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006970482842534729,
      "loss": 4.7311,
      "step": 12872
    },
    {
      "epoch": 1.7629416598192276,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006970140433124506,
      "loss": 4.7391,
      "step": 12873
    },
    {
      "epoch": 1.7630786086003836,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000696979800324364,
      "loss": 4.7931,
      "step": 12874
    },
    {
      "epoch": 1.7632155573815393,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006969455552894968,
      "loss": 4.8271,
      "step": 12875
    },
    {
      "epoch": 1.7633525061626951,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006969113082081332,
      "loss": 4.7079,
      "step": 12876
    },
    {
      "epoch": 1.763489454943851,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006968770590805567,
      "loss": 4.8166,
      "step": 12877
    },
    {
      "epoch": 1.763626403725007,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006968428079070512,
      "loss": 4.8037,
      "step": 12878
    },
    {
      "epoch": 1.7637633525061627,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006968085546879004,
      "loss": 4.726,
      "step": 12879
    },
    {
      "epoch": 1.7639003012873187,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006967742994233885,
      "loss": 4.7988,
      "step": 12880
    },
    {
      "epoch": 1.7640372500684744,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006967400421137988,
      "loss": 4.8526,
      "step": 12881
    },
    {
      "epoch": 1.7641741988496302,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006967057827594159,
      "loss": 4.7803,
      "step": 12882
    },
    {
      "epoch": 1.764311147630786,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006966715213605233,
      "loss": 4.8671,
      "step": 12883
    },
    {
      "epoch": 1.7644480964119418,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000696637257917405,
      "loss": 4.8571,
      "step": 12884
    },
    {
      "epoch": 1.7645850451930978,
      "grad_norm": 0.375,
      "learning_rate": 0.0006966029924303451,
      "loss": 4.8414,
      "step": 12885
    },
    {
      "epoch": 1.7647219939742538,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006965687248996274,
      "loss": 4.7805,
      "step": 12886
    },
    {
      "epoch": 1.7648589427554096,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006965344553255358,
      "loss": 4.8043,
      "step": 12887
    },
    {
      "epoch": 1.7649958915365653,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006965001837083545,
      "loss": 4.803,
      "step": 12888
    },
    {
      "epoch": 1.765132840317721,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006964659100483673,
      "loss": 4.7619,
      "step": 12889
    },
    {
      "epoch": 1.7652697890988769,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006964316343458586,
      "loss": 4.8211,
      "step": 12890
    },
    {
      "epoch": 1.7654067378800329,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006963973566011121,
      "loss": 4.7537,
      "step": 12891
    },
    {
      "epoch": 1.7655436866611887,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006963630768144121,
      "loss": 4.775,
      "step": 12892
    },
    {
      "epoch": 1.7656806354423447,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006963287949860425,
      "loss": 4.8001,
      "step": 12893
    },
    {
      "epoch": 1.7658175842235004,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006962945111162875,
      "loss": 4.7724,
      "step": 12894
    },
    {
      "epoch": 1.7659545330046562,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006962602252054311,
      "loss": 4.8038,
      "step": 12895
    },
    {
      "epoch": 1.766091481785812,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006962259372537575,
      "loss": 4.8205,
      "step": 12896
    },
    {
      "epoch": 1.766228430566968,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006961916472615511,
      "loss": 4.7901,
      "step": 12897
    },
    {
      "epoch": 1.7663653793481238,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006961573552290958,
      "loss": 4.8373,
      "step": 12898
    },
    {
      "epoch": 1.7665023281292798,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006961230611566758,
      "loss": 4.7382,
      "step": 12899
    },
    {
      "epoch": 1.7666392769104355,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006960887650445753,
      "loss": 4.8374,
      "step": 12900
    },
    {
      "epoch": 1.7667762256915913,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006960544668930787,
      "loss": 4.7593,
      "step": 12901
    },
    {
      "epoch": 1.766913174472747,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006960201667024699,
      "loss": 4.6955,
      "step": 12902
    },
    {
      "epoch": 1.767050123253903,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006959858644730334,
      "loss": 4.8245,
      "step": 12903
    },
    {
      "epoch": 1.7671870720350589,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006959515602050535,
      "loss": 4.7781,
      "step": 12904
    },
    {
      "epoch": 1.7673240208162149,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006959172538988144,
      "loss": 4.7643,
      "step": 12905
    },
    {
      "epoch": 1.7674609695973706,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0006958829455546004,
      "loss": 4.8145,
      "step": 12906
    },
    {
      "epoch": 1.7675979183785264,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006958486351726959,
      "loss": 4.7851,
      "step": 12907
    },
    {
      "epoch": 1.7677348671596822,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006958143227533851,
      "loss": 4.7865,
      "step": 12908
    },
    {
      "epoch": 1.767871815940838,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006957800082969525,
      "loss": 4.7164,
      "step": 12909
    },
    {
      "epoch": 1.768008764721994,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0006957456918036824,
      "loss": 4.7558,
      "step": 12910
    },
    {
      "epoch": 1.76814571350315,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000695711373273859,
      "loss": 4.8173,
      "step": 12911
    },
    {
      "epoch": 1.7682826622843058,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0006956770527077671,
      "loss": 4.7935,
      "step": 12912
    },
    {
      "epoch": 1.7684196110654615,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006956427301056909,
      "loss": 4.767,
      "step": 12913
    },
    {
      "epoch": 1.7685565598466173,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006956084054679149,
      "loss": 4.8598,
      "step": 12914
    },
    {
      "epoch": 1.768693508627773,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006955740787947234,
      "loss": 4.7892,
      "step": 12915
    },
    {
      "epoch": 1.768830457408929,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006955397500864011,
      "loss": 4.9263,
      "step": 12916
    },
    {
      "epoch": 1.7689674061900849,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006955054193432324,
      "loss": 4.7828,
      "step": 12917
    },
    {
      "epoch": 1.7691043549712409,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006954710865655016,
      "loss": 4.8157,
      "step": 12918
    },
    {
      "epoch": 1.7692413037523966,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006954367517534937,
      "loss": 4.8056,
      "step": 12919
    },
    {
      "epoch": 1.7693782525335524,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006954024149074927,
      "loss": 4.85,
      "step": 12920
    },
    {
      "epoch": 1.7695152013147082,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006953680760277835,
      "loss": 4.7689,
      "step": 12921
    },
    {
      "epoch": 1.7696521500958642,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006953337351146506,
      "loss": 4.8541,
      "step": 12922
    },
    {
      "epoch": 1.76978909887702,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006952993921683786,
      "loss": 4.7624,
      "step": 12923
    },
    {
      "epoch": 1.769926047658176,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000695265047189252,
      "loss": 4.7847,
      "step": 12924
    },
    {
      "epoch": 1.7700629964393317,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006952307001775556,
      "loss": 4.7969,
      "step": 12925
    },
    {
      "epoch": 1.7701999452204875,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006951963511335739,
      "loss": 4.91,
      "step": 12926
    },
    {
      "epoch": 1.7703368940016433,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006951620000575916,
      "loss": 4.8653,
      "step": 12927
    },
    {
      "epoch": 1.770473842782799,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006951276469498934,
      "loss": 4.7074,
      "step": 12928
    },
    {
      "epoch": 1.770610791563955,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000695093291810764,
      "loss": 4.7312,
      "step": 12929
    },
    {
      "epoch": 1.770747740345111,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006950589346404881,
      "loss": 4.8114,
      "step": 12930
    },
    {
      "epoch": 1.7708846891262668,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006950245754393504,
      "loss": 4.8228,
      "step": 12931
    },
    {
      "epoch": 1.7710216379074226,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006949902142076357,
      "loss": 4.8536,
      "step": 12932
    },
    {
      "epoch": 1.7711585866885784,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006949558509456286,
      "loss": 4.8859,
      "step": 12933
    },
    {
      "epoch": 1.7712955354697342,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006949214856536141,
      "loss": 4.7938,
      "step": 12934
    },
    {
      "epoch": 1.7714324842508902,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006948871183318769,
      "loss": 4.9177,
      "step": 12935
    },
    {
      "epoch": 1.7715694330320462,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006948527489807017,
      "loss": 4.821,
      "step": 12936
    },
    {
      "epoch": 1.771706381813202,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006948183776003735,
      "loss": 4.8333,
      "step": 12937
    },
    {
      "epoch": 1.7718433305943577,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000694784004191177,
      "loss": 4.8354,
      "step": 12938
    },
    {
      "epoch": 1.7719802793755135,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006947496287533973,
      "loss": 4.8495,
      "step": 12939
    },
    {
      "epoch": 1.7721172281566693,
      "grad_norm": 0.375,
      "learning_rate": 0.0006947152512873188,
      "loss": 4.785,
      "step": 12940
    },
    {
      "epoch": 1.7722541769378253,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000694680871793227,
      "loss": 4.7564,
      "step": 12941
    },
    {
      "epoch": 1.772391125718981,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006946464902714063,
      "loss": 4.7747,
      "step": 12942
    },
    {
      "epoch": 1.772528074500137,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006946121067221419,
      "loss": 4.8675,
      "step": 12943
    },
    {
      "epoch": 1.7726650232812928,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006945777211457187,
      "loss": 4.6977,
      "step": 12944
    },
    {
      "epoch": 1.7728019720624486,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006945433335424217,
      "loss": 4.747,
      "step": 12945
    },
    {
      "epoch": 1.7729389208436044,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006945089439125357,
      "loss": 4.7902,
      "step": 12946
    },
    {
      "epoch": 1.7730758696247604,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000694474552256346,
      "loss": 4.7922,
      "step": 12947
    },
    {
      "epoch": 1.7732128184059162,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006944401585741373,
      "loss": 4.7114,
      "step": 12948
    },
    {
      "epoch": 1.7733497671870722,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006944057628661947,
      "loss": 4.7977,
      "step": 12949
    },
    {
      "epoch": 1.773486715968228,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006943713651328034,
      "loss": 4.8663,
      "step": 12950
    },
    {
      "epoch": 1.7736236647493837,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006943369653742482,
      "loss": 4.8086,
      "step": 12951
    },
    {
      "epoch": 1.7737606135305395,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006943025635908148,
      "loss": 4.8698,
      "step": 12952
    },
    {
      "epoch": 1.7738975623116953,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006942681597827874,
      "loss": 4.7718,
      "step": 12953
    },
    {
      "epoch": 1.7740345110928513,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006942337539504517,
      "loss": 4.7806,
      "step": 12954
    },
    {
      "epoch": 1.7741714598740073,
      "grad_norm": 0.375,
      "learning_rate": 0.0006941993460940927,
      "loss": 4.8542,
      "step": 12955
    },
    {
      "epoch": 1.774308408655163,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006941649362139956,
      "loss": 4.7327,
      "step": 12956
    },
    {
      "epoch": 1.7744453574363188,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006941305243104453,
      "loss": 4.7823,
      "step": 12957
    },
    {
      "epoch": 1.7745823062174746,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006940961103837273,
      "loss": 4.7995,
      "step": 12958
    },
    {
      "epoch": 1.7747192549986304,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006940616944341267,
      "loss": 4.7927,
      "step": 12959
    },
    {
      "epoch": 1.7748562037797864,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006940272764619285,
      "loss": 4.7852,
      "step": 12960
    },
    {
      "epoch": 1.7749931525609424,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006939928564674183,
      "loss": 4.8296,
      "step": 12961
    },
    {
      "epoch": 1.7751301013420981,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000693958434450881,
      "loss": 4.7529,
      "step": 12962
    },
    {
      "epoch": 1.775267050123254,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006939240104126023,
      "loss": 4.8944,
      "step": 12963
    },
    {
      "epoch": 1.7754039989044097,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006938895843528669,
      "loss": 4.7402,
      "step": 12964
    },
    {
      "epoch": 1.7755409476855655,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006938551562719606,
      "loss": 4.8486,
      "step": 12965
    },
    {
      "epoch": 1.7756778964667215,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006938207261701683,
      "loss": 4.8156,
      "step": 12966
    },
    {
      "epoch": 1.7758148452478772,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006937862940477757,
      "loss": 4.7193,
      "step": 12967
    },
    {
      "epoch": 1.7759517940290332,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000693751859905068,
      "loss": 4.8522,
      "step": 12968
    },
    {
      "epoch": 1.776088742810189,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006937174237423306,
      "loss": 4.7581,
      "step": 12969
    },
    {
      "epoch": 1.7762256915913448,
      "grad_norm": 0.375,
      "learning_rate": 0.0006936829855598486,
      "loss": 4.837,
      "step": 12970
    },
    {
      "epoch": 1.7763626403725006,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006936485453579078,
      "loss": 4.8197,
      "step": 12971
    },
    {
      "epoch": 1.7764995891536566,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006936141031367935,
      "loss": 4.8004,
      "step": 12972
    },
    {
      "epoch": 1.7766365379348124,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006935796588967909,
      "loss": 4.7621,
      "step": 12973
    },
    {
      "epoch": 1.7767734867159684,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006935452126381859,
      "loss": 4.8645,
      "step": 12974
    },
    {
      "epoch": 1.7769104354971241,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006935107643612635,
      "loss": 4.8686,
      "step": 12975
    },
    {
      "epoch": 1.77704738427828,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006934763140663093,
      "loss": 4.894,
      "step": 12976
    },
    {
      "epoch": 1.7771843330594357,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006934418617536089,
      "loss": 4.8042,
      "step": 12977
    },
    {
      "epoch": 1.7773212818405915,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006934074074234479,
      "loss": 4.7719,
      "step": 12978
    },
    {
      "epoch": 1.7774582306217475,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006933729510761117,
      "loss": 4.8138,
      "step": 12979
    },
    {
      "epoch": 1.7775951794029035,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006933384927118857,
      "loss": 4.8485,
      "step": 12980
    },
    {
      "epoch": 1.7777321281840592,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006933040323310559,
      "loss": 4.8466,
      "step": 12981
    },
    {
      "epoch": 1.777869076965215,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006932695699339073,
      "loss": 4.7788,
      "step": 12982
    },
    {
      "epoch": 1.7780060257463708,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006932351055207261,
      "loss": 4.8997,
      "step": 12983
    },
    {
      "epoch": 1.7781429745275266,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006932006390917974,
      "loss": 4.7142,
      "step": 12984
    },
    {
      "epoch": 1.7782799233086826,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006931661706474071,
      "loss": 4.7619,
      "step": 12985
    },
    {
      "epoch": 1.7784168720898386,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006931317001878409,
      "loss": 4.8819,
      "step": 12986
    },
    {
      "epoch": 1.7785538208709943,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006930972277133842,
      "loss": 4.7699,
      "step": 12987
    },
    {
      "epoch": 1.7786907696521501,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006930627532243229,
      "loss": 4.8289,
      "step": 12988
    },
    {
      "epoch": 1.778827718433306,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006930282767209428,
      "loss": 4.7675,
      "step": 12989
    },
    {
      "epoch": 1.7789646672144617,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006929937982035292,
      "loss": 4.7676,
      "step": 12990
    },
    {
      "epoch": 1.7791016159956177,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006929593176723684,
      "loss": 4.7505,
      "step": 12991
    },
    {
      "epoch": 1.7792385647767734,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006929248351277456,
      "loss": 4.844,
      "step": 12992
    },
    {
      "epoch": 1.7793755135579294,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006928903505699469,
      "loss": 4.8303,
      "step": 12993
    },
    {
      "epoch": 1.7795124623390852,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000692855863999258,
      "loss": 4.791,
      "step": 12994
    },
    {
      "epoch": 1.779649411120241,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006928213754159645,
      "loss": 4.8261,
      "step": 12995
    },
    {
      "epoch": 1.7797863599013968,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006927868848203528,
      "loss": 4.8257,
      "step": 12996
    },
    {
      "epoch": 1.7799233086825528,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000692752392212708,
      "loss": 4.8078,
      "step": 12997
    },
    {
      "epoch": 1.7800602574637086,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006927178975933164,
      "loss": 4.8136,
      "step": 12998
    },
    {
      "epoch": 1.7801972062448645,
      "grad_norm": 0.375,
      "learning_rate": 0.0006926834009624637,
      "loss": 4.7709,
      "step": 12999
    },
    {
      "epoch": 1.7803341550260203,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006926489023204358,
      "loss": 4.8152,
      "step": 13000
    },
    {
      "epoch": 1.780471103807176,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006926144016675187,
      "loss": 4.8354,
      "step": 13001
    },
    {
      "epoch": 1.7806080525883319,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006925798990039982,
      "loss": 4.8228,
      "step": 13002
    },
    {
      "epoch": 1.7807450013694877,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006925453943301604,
      "loss": 4.7662,
      "step": 13003
    },
    {
      "epoch": 1.7808819501506437,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006925108876462908,
      "loss": 4.766,
      "step": 13004
    },
    {
      "epoch": 1.7810188989317997,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000692476378952676,
      "loss": 4.8386,
      "step": 13005
    },
    {
      "epoch": 1.7811558477129554,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006924418682496016,
      "loss": 4.7583,
      "step": 13006
    },
    {
      "epoch": 1.7812927964941112,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006924073555373537,
      "loss": 4.8164,
      "step": 13007
    },
    {
      "epoch": 1.781429745275267,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006923728408162181,
      "loss": 4.8592,
      "step": 13008
    },
    {
      "epoch": 1.7815666940564228,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006923383240864811,
      "loss": 4.795,
      "step": 13009
    },
    {
      "epoch": 1.7817036428375788,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006923038053484288,
      "loss": 4.8129,
      "step": 13010
    },
    {
      "epoch": 1.7818405916187348,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0006922692846023469,
      "loss": 4.8502,
      "step": 13011
    },
    {
      "epoch": 1.7819775403998905,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006922347618485219,
      "loss": 4.8545,
      "step": 13012
    },
    {
      "epoch": 1.7821144891810463,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006922002370872396,
      "loss": 4.7455,
      "step": 13013
    },
    {
      "epoch": 1.782251437962202,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0006921657103187863,
      "loss": 4.8101,
      "step": 13014
    },
    {
      "epoch": 1.7823883867433579,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006921311815434479,
      "loss": 4.8295,
      "step": 13015
    },
    {
      "epoch": 1.7825253355245139,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0006920966507615108,
      "loss": 4.8303,
      "step": 13016
    },
    {
      "epoch": 1.7826622843056696,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006920621179732612,
      "loss": 4.8316,
      "step": 13017
    },
    {
      "epoch": 1.7827992330868256,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0006920275831789849,
      "loss": 4.7572,
      "step": 13018
    },
    {
      "epoch": 1.7829361818679814,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006919930463789686,
      "loss": 4.8407,
      "step": 13019
    },
    {
      "epoch": 1.7830731306491372,
      "grad_norm": 0.51171875,
      "learning_rate": 0.000691958507573498,
      "loss": 4.8354,
      "step": 13020
    },
    {
      "epoch": 1.783210079430293,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006919239667628598,
      "loss": 4.7739,
      "step": 13021
    },
    {
      "epoch": 1.783347028211449,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0006918894239473399,
      "loss": 4.8691,
      "step": 13022
    },
    {
      "epoch": 1.7834839769926047,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006918548791272248,
      "loss": 4.7605,
      "step": 13023
    },
    {
      "epoch": 1.7836209257737607,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0006918203323028007,
      "loss": 4.8074,
      "step": 13024
    },
    {
      "epoch": 1.7837578745549165,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006917857834743539,
      "loss": 4.8098,
      "step": 13025
    },
    {
      "epoch": 1.7838948233360723,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0006917512326421705,
      "loss": 4.8021,
      "step": 13026
    },
    {
      "epoch": 1.784031772117228,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006917166798065372,
      "loss": 4.819,
      "step": 13027
    },
    {
      "epoch": 1.7841687208983839,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006916821249677401,
      "loss": 4.766,
      "step": 13028
    },
    {
      "epoch": 1.7843056696795399,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006916475681260657,
      "loss": 4.7998,
      "step": 13029
    },
    {
      "epoch": 1.7844426184606959,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006916130092818005,
      "loss": 4.7395,
      "step": 13030
    },
    {
      "epoch": 1.7845795672418516,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006915784484352304,
      "loss": 4.7484,
      "step": 13031
    },
    {
      "epoch": 1.7847165160230074,
      "grad_norm": 0.375,
      "learning_rate": 0.0006915438855866422,
      "loss": 4.7693,
      "step": 13032
    },
    {
      "epoch": 1.7848534648041632,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006915093207363224,
      "loss": 4.779,
      "step": 13033
    },
    {
      "epoch": 1.784990413585319,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006914747538845572,
      "loss": 4.7928,
      "step": 13034
    },
    {
      "epoch": 1.785127362366475,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006914401850316332,
      "loss": 4.7452,
      "step": 13035
    },
    {
      "epoch": 1.7852643111476307,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006914056141778368,
      "loss": 4.7737,
      "step": 13036
    },
    {
      "epoch": 1.7854012599287867,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006913710413234547,
      "loss": 4.7164,
      "step": 13037
    },
    {
      "epoch": 1.7855382087099425,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006913364664687731,
      "loss": 4.8369,
      "step": 13038
    },
    {
      "epoch": 1.7856751574910983,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006913018896140789,
      "loss": 4.8221,
      "step": 13039
    },
    {
      "epoch": 1.785812106272254,
      "grad_norm": 0.375,
      "learning_rate": 0.0006912673107596582,
      "loss": 4.755,
      "step": 13040
    },
    {
      "epoch": 1.78594905505341,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006912327299057978,
      "loss": 4.6848,
      "step": 13041
    },
    {
      "epoch": 1.7860860038345658,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006911981470527845,
      "loss": 4.7921,
      "step": 13042
    },
    {
      "epoch": 1.7862229526157218,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006911635622009046,
      "loss": 4.7997,
      "step": 13043
    },
    {
      "epoch": 1.7863599013968776,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006911289753504448,
      "loss": 4.7379,
      "step": 13044
    },
    {
      "epoch": 1.7864968501780334,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006910943865016916,
      "loss": 4.7758,
      "step": 13045
    },
    {
      "epoch": 1.7866337989591892,
      "grad_norm": 0.359375,
      "learning_rate": 0.000691059795654932,
      "loss": 4.848,
      "step": 13046
    },
    {
      "epoch": 1.786770747740345,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006910252028104523,
      "loss": 4.8639,
      "step": 13047
    },
    {
      "epoch": 1.786907696521501,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006909906079685393,
      "loss": 4.8074,
      "step": 13048
    },
    {
      "epoch": 1.787044645302657,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006909560111294798,
      "loss": 4.6973,
      "step": 13049
    },
    {
      "epoch": 1.7871815940838127,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006909214122935602,
      "loss": 4.7573,
      "step": 13050
    },
    {
      "epoch": 1.7873185428649685,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006908868114610677,
      "loss": 4.8431,
      "step": 13051
    },
    {
      "epoch": 1.7874554916461243,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006908522086322886,
      "loss": 4.7571,
      "step": 13052
    },
    {
      "epoch": 1.78759244042728,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006908176038075099,
      "loss": 4.7419,
      "step": 13053
    },
    {
      "epoch": 1.787729389208436,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006907829969870184,
      "loss": 4.8189,
      "step": 13054
    },
    {
      "epoch": 1.787866337989592,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006907483881711008,
      "loss": 4.7566,
      "step": 13055
    },
    {
      "epoch": 1.7880032867707478,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000690713777360044,
      "loss": 4.8094,
      "step": 13056
    },
    {
      "epoch": 1.7881402355519036,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006906791645541347,
      "loss": 4.853,
      "step": 13057
    },
    {
      "epoch": 1.7882771843330594,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006906445497536597,
      "loss": 4.7687,
      "step": 13058
    },
    {
      "epoch": 1.7884141331142152,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006906099329589061,
      "loss": 4.8307,
      "step": 13059
    },
    {
      "epoch": 1.7885510818953712,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006905753141701607,
      "loss": 4.8703,
      "step": 13060
    },
    {
      "epoch": 1.788688030676527,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006905406933877102,
      "loss": 4.7049,
      "step": 13061
    },
    {
      "epoch": 1.788824979457683,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006905060706118418,
      "loss": 4.9115,
      "step": 13062
    },
    {
      "epoch": 1.7889619282388387,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006904714458428423,
      "loss": 4.8708,
      "step": 13063
    },
    {
      "epoch": 1.7890988770199945,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006904368190809986,
      "loss": 4.8034,
      "step": 13064
    },
    {
      "epoch": 1.7892358258011503,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006904021903265975,
      "loss": 4.8683,
      "step": 13065
    },
    {
      "epoch": 1.7893727745823063,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006903675595799263,
      "loss": 4.7901,
      "step": 13066
    },
    {
      "epoch": 1.789509723363462,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006903329268412718,
      "loss": 4.8768,
      "step": 13067
    },
    {
      "epoch": 1.789646672144618,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006902982921109214,
      "loss": 4.8186,
      "step": 13068
    },
    {
      "epoch": 1.7897836209257738,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006902636553891614,
      "loss": 4.7708,
      "step": 13069
    },
    {
      "epoch": 1.7899205697069296,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006902290166762793,
      "loss": 4.8736,
      "step": 13070
    },
    {
      "epoch": 1.7900575184880854,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006901943759725622,
      "loss": 4.8703,
      "step": 13071
    },
    {
      "epoch": 1.7901944672692411,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000690159733278297,
      "loss": 4.8134,
      "step": 13072
    },
    {
      "epoch": 1.7903314160503971,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006901250885937709,
      "loss": 4.7733,
      "step": 13073
    },
    {
      "epoch": 1.7904683648315531,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006900904419192709,
      "loss": 4.7872,
      "step": 13074
    },
    {
      "epoch": 1.790605313612709,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006900557932550843,
      "loss": 4.8085,
      "step": 13075
    },
    {
      "epoch": 1.7907422623938647,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006900211426014981,
      "loss": 4.8386,
      "step": 13076
    },
    {
      "epoch": 1.7908792111750205,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006899864899587995,
      "loss": 4.8534,
      "step": 13077
    },
    {
      "epoch": 1.7910161599561762,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006899518353272756,
      "loss": 4.8132,
      "step": 13078
    },
    {
      "epoch": 1.7911531087373322,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006899171787072138,
      "loss": 4.8524,
      "step": 13079
    },
    {
      "epoch": 1.7912900575184882,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006898825200989011,
      "loss": 4.7656,
      "step": 13080
    },
    {
      "epoch": 1.791427006299644,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006898478595026248,
      "loss": 4.8058,
      "step": 13081
    },
    {
      "epoch": 1.7915639550807998,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006898131969186721,
      "loss": 4.8554,
      "step": 13082
    },
    {
      "epoch": 1.7917009038619556,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006897785323473303,
      "loss": 4.7599,
      "step": 13083
    },
    {
      "epoch": 1.7918378526431114,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0006897438657888868,
      "loss": 4.8372,
      "step": 13084
    },
    {
      "epoch": 1.7919748014242673,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006897091972436286,
      "loss": 4.7384,
      "step": 13085
    },
    {
      "epoch": 1.7921117502054231,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006896745267118432,
      "loss": 4.8323,
      "step": 13086
    },
    {
      "epoch": 1.7922486989865791,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006896398541938178,
      "loss": 4.7386,
      "step": 13087
    },
    {
      "epoch": 1.792385647767735,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00068960517968984,
      "loss": 4.8956,
      "step": 13088
    },
    {
      "epoch": 1.7925225965488907,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006895705032001969,
      "loss": 4.7284,
      "step": 13089
    },
    {
      "epoch": 1.7926595453300465,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000689535824725176,
      "loss": 4.8315,
      "step": 13090
    },
    {
      "epoch": 1.7927964941112025,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006895011442650645,
      "loss": 4.7354,
      "step": 13091
    },
    {
      "epoch": 1.7929334428923582,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006894664618201502,
      "loss": 4.8412,
      "step": 13092
    },
    {
      "epoch": 1.7930703916735142,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00068943177739072,
      "loss": 4.7835,
      "step": 13093
    },
    {
      "epoch": 1.79320734045467,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006893970909770616,
      "loss": 4.8353,
      "step": 13094
    },
    {
      "epoch": 1.7933442892358258,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006893624025794627,
      "loss": 4.8179,
      "step": 13095
    },
    {
      "epoch": 1.7934812380169816,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0006893277121982104,
      "loss": 4.8006,
      "step": 13096
    },
    {
      "epoch": 1.7936181867981373,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006892930198335923,
      "loss": 4.7766,
      "step": 13097
    },
    {
      "epoch": 1.7937551355792933,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006892583254858961,
      "loss": 4.8299,
      "step": 13098
    },
    {
      "epoch": 1.7938920843604493,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006892236291554089,
      "loss": 4.7729,
      "step": 13099
    },
    {
      "epoch": 1.7940290331416051,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006891889308424186,
      "loss": 4.7984,
      "step": 13100
    },
    {
      "epoch": 1.7941659819227609,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006891542305472127,
      "loss": 4.7705,
      "step": 13101
    },
    {
      "epoch": 1.7943029307039167,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006891195282700786,
      "loss": 4.8016,
      "step": 13102
    },
    {
      "epoch": 1.7944398794850724,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006890848240113039,
      "loss": 4.8283,
      "step": 13103
    },
    {
      "epoch": 1.7945768282662284,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006890501177711765,
      "loss": 4.8016,
      "step": 13104
    },
    {
      "epoch": 1.7947137770473844,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006890154095499838,
      "loss": 4.8176,
      "step": 13105
    },
    {
      "epoch": 1.7948507258285402,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006889806993480134,
      "loss": 4.8536,
      "step": 13106
    },
    {
      "epoch": 1.794987674609696,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006889459871655529,
      "loss": 4.7499,
      "step": 13107
    },
    {
      "epoch": 1.7951246233908518,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006889112730028901,
      "loss": 4.7732,
      "step": 13108
    },
    {
      "epoch": 1.7952615721720075,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006888765568603126,
      "loss": 4.8508,
      "step": 13109
    },
    {
      "epoch": 1.7953985209531635,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006888418387381083,
      "loss": 4.8556,
      "step": 13110
    },
    {
      "epoch": 1.7955354697343193,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006888071186365647,
      "loss": 4.8185,
      "step": 13111
    },
    {
      "epoch": 1.7956724185154753,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006887723965559695,
      "loss": 4.806,
      "step": 13112
    },
    {
      "epoch": 1.795809367296631,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006887376724966108,
      "loss": 4.7591,
      "step": 13113
    },
    {
      "epoch": 1.7959463160777869,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000688702946458776,
      "loss": 4.7669,
      "step": 13114
    },
    {
      "epoch": 1.7960832648589427,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000688668218442753,
      "loss": 4.826,
      "step": 13115
    },
    {
      "epoch": 1.7962202136400987,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006886334884488296,
      "loss": 4.7767,
      "step": 13116
    },
    {
      "epoch": 1.7963571624212544,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006885987564772936,
      "loss": 4.806,
      "step": 13117
    },
    {
      "epoch": 1.7964941112024104,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006885640225284328,
      "loss": 4.7854,
      "step": 13118
    },
    {
      "epoch": 1.7966310599835662,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006885292866025352,
      "loss": 4.7441,
      "step": 13119
    },
    {
      "epoch": 1.796768008764722,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006884945486998885,
      "loss": 4.773,
      "step": 13120
    },
    {
      "epoch": 1.7969049575458778,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006884598088207805,
      "loss": 4.86,
      "step": 13121
    },
    {
      "epoch": 1.7970419063270335,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006884250669654995,
      "loss": 4.8157,
      "step": 13122
    },
    {
      "epoch": 1.7971788551081895,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006883903231343331,
      "loss": 4.8481,
      "step": 13123
    },
    {
      "epoch": 1.7973158038893455,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006883555773275693,
      "loss": 4.7698,
      "step": 13124
    },
    {
      "epoch": 1.7974527526705013,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000688320829545496,
      "loss": 4.8164,
      "step": 13125
    },
    {
      "epoch": 1.797589701451657,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006882860797884011,
      "loss": 4.7558,
      "step": 13126
    },
    {
      "epoch": 1.7977266502328129,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006882513280565728,
      "loss": 4.7633,
      "step": 13127
    },
    {
      "epoch": 1.7978635990139686,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006882165743502988,
      "loss": 4.7565,
      "step": 13128
    },
    {
      "epoch": 1.7980005477951246,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006881818186698675,
      "loss": 4.814,
      "step": 13129
    },
    {
      "epoch": 1.7981374965762806,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006881470610155666,
      "loss": 4.8636,
      "step": 13130
    },
    {
      "epoch": 1.7982744453574364,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006881123013876844,
      "loss": 4.8059,
      "step": 13131
    },
    {
      "epoch": 1.7984113941385922,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006880775397865086,
      "loss": 4.7512,
      "step": 13132
    },
    {
      "epoch": 1.798548342919748,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006880427762123276,
      "loss": 4.7084,
      "step": 13133
    },
    {
      "epoch": 1.7986852917009037,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006880080106654294,
      "loss": 4.7682,
      "step": 13134
    },
    {
      "epoch": 1.7988222404820597,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006879732431461021,
      "loss": 4.7476,
      "step": 13135
    },
    {
      "epoch": 1.7989591892632155,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006879384736546338,
      "loss": 4.7716,
      "step": 13136
    },
    {
      "epoch": 1.7990961380443715,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006879037021913127,
      "loss": 4.8046,
      "step": 13137
    },
    {
      "epoch": 1.7992330868255273,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006878689287564269,
      "loss": 4.7609,
      "step": 13138
    },
    {
      "epoch": 1.799370035606683,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006878341533502646,
      "loss": 4.7872,
      "step": 13139
    },
    {
      "epoch": 1.7995069843878388,
      "grad_norm": 1.0,
      "learning_rate": 0.0006877993759731142,
      "loss": 4.8025,
      "step": 13140
    },
    {
      "epoch": 1.7996439331689948,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006877645966252634,
      "loss": 4.8165,
      "step": 13141
    },
    {
      "epoch": 1.7997808819501506,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006877298153070008,
      "loss": 4.8196,
      "step": 13142
    },
    {
      "epoch": 1.7999178307313066,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006876950320186146,
      "loss": 4.7834,
      "step": 13143
    },
    {
      "epoch": 1.8000547795124624,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006876602467603931,
      "loss": 4.7666,
      "step": 13144
    },
    {
      "epoch": 1.8001917282936182,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0006876254595326245,
      "loss": 4.7688,
      "step": 13145
    },
    {
      "epoch": 1.800328677074774,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006875906703355971,
      "loss": 4.7519,
      "step": 13146
    },
    {
      "epoch": 1.8004656258559297,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006875558791695992,
      "loss": 4.8785,
      "step": 13147
    },
    {
      "epoch": 1.8006025746370857,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006875210860349191,
      "loss": 4.7644,
      "step": 13148
    },
    {
      "epoch": 1.8007395234182417,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006874862909318451,
      "loss": 4.8252,
      "step": 13149
    },
    {
      "epoch": 1.8008764721993975,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006874514938606655,
      "loss": 4.8155,
      "step": 13150
    },
    {
      "epoch": 1.8010134209805533,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006874166948216692,
      "loss": 4.8413,
      "step": 13151
    },
    {
      "epoch": 1.801150369761709,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006873818938151439,
      "loss": 4.8204,
      "step": 13152
    },
    {
      "epoch": 1.8012873185428648,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006873470908413782,
      "loss": 4.8567,
      "step": 13153
    },
    {
      "epoch": 1.8014242673240208,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0006873122859006607,
      "loss": 4.7444,
      "step": 13154
    },
    {
      "epoch": 1.8015612161051766,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006872774789932797,
      "loss": 4.8869,
      "step": 13155
    },
    {
      "epoch": 1.8016981648863326,
      "grad_norm": 0.4375,
      "learning_rate": 0.0006872426701195239,
      "loss": 4.8884,
      "step": 13156
    },
    {
      "epoch": 1.8018351136674884,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006872078592796813,
      "loss": 4.8395,
      "step": 13157
    },
    {
      "epoch": 1.8019720624486442,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006871730464740408,
      "loss": 4.7043,
      "step": 13158
    },
    {
      "epoch": 1.8021090112298,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006871382317028906,
      "loss": 4.7769,
      "step": 13159
    },
    {
      "epoch": 1.802245960010956,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006871034149665196,
      "loss": 4.8588,
      "step": 13160
    },
    {
      "epoch": 1.8023829087921117,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006870685962652159,
      "loss": 4.7558,
      "step": 13161
    },
    {
      "epoch": 1.8025198575732677,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006870337755992684,
      "loss": 4.8105,
      "step": 13162
    },
    {
      "epoch": 1.8026568063544235,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006869989529689654,
      "loss": 4.7791,
      "step": 13163
    },
    {
      "epoch": 1.8027937551355793,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006869641283745955,
      "loss": 4.8443,
      "step": 13164
    },
    {
      "epoch": 1.802930703916735,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006869293018164477,
      "loss": 4.8509,
      "step": 13165
    },
    {
      "epoch": 1.803067652697891,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006868944732948101,
      "loss": 4.81,
      "step": 13166
    },
    {
      "epoch": 1.8032046014790468,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006868596428099717,
      "loss": 4.8624,
      "step": 13167
    },
    {
      "epoch": 1.8033415502602028,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006868248103622208,
      "loss": 4.8496,
      "step": 13168
    },
    {
      "epoch": 1.8034784990413586,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006867899759518464,
      "loss": 4.785,
      "step": 13169
    },
    {
      "epoch": 1.8036154478225144,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006867551395791371,
      "loss": 4.7387,
      "step": 13170
    },
    {
      "epoch": 1.8037523966036701,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006867203012443815,
      "loss": 4.7019,
      "step": 13171
    },
    {
      "epoch": 1.803889345384826,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006866854609478682,
      "loss": 4.8414,
      "step": 13172
    },
    {
      "epoch": 1.804026294165982,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006866506186898862,
      "loss": 4.8597,
      "step": 13173
    },
    {
      "epoch": 1.804163242947138,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006866157744707242,
      "loss": 4.812,
      "step": 13174
    },
    {
      "epoch": 1.8043001917282937,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006865809282906708,
      "loss": 4.7651,
      "step": 13175
    },
    {
      "epoch": 1.8044371405094495,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006865460801500149,
      "loss": 4.704,
      "step": 13176
    },
    {
      "epoch": 1.8045740892906053,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006865112300490452,
      "loss": 4.8835,
      "step": 13177
    },
    {
      "epoch": 1.804711038071761,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006864763779880508,
      "loss": 4.7967,
      "step": 13178
    },
    {
      "epoch": 1.804847986852917,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006864415239673201,
      "loss": 4.7587,
      "step": 13179
    },
    {
      "epoch": 1.8049849356340728,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006864066679871422,
      "loss": 4.7896,
      "step": 13180
    },
    {
      "epoch": 1.8051218844152288,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006863718100478059,
      "loss": 4.8635,
      "step": 13181
    },
    {
      "epoch": 1.8052588331963846,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006863369501496001,
      "loss": 4.7948,
      "step": 13182
    },
    {
      "epoch": 1.8053957819775404,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006863020882928138,
      "loss": 4.7586,
      "step": 13183
    },
    {
      "epoch": 1.8055327307586961,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006862672244777356,
      "loss": 4.8893,
      "step": 13184
    },
    {
      "epoch": 1.8056696795398521,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006862323587046547,
      "loss": 4.7919,
      "step": 13185
    },
    {
      "epoch": 1.805806628321008,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006861974909738599,
      "loss": 4.8055,
      "step": 13186
    },
    {
      "epoch": 1.805943577102164,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006861626212856403,
      "loss": 4.7707,
      "step": 13187
    },
    {
      "epoch": 1.8060805258833197,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006861277496402847,
      "loss": 4.7625,
      "step": 13188
    },
    {
      "epoch": 1.8062174746644755,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006860928760380822,
      "loss": 4.8435,
      "step": 13189
    },
    {
      "epoch": 1.8063544234456312,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006860580004793219,
      "loss": 4.8074,
      "step": 13190
    },
    {
      "epoch": 1.806491372226787,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006860231229642924,
      "loss": 4.7562,
      "step": 13191
    },
    {
      "epoch": 1.806628321007943,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006859882434932834,
      "loss": 4.8265,
      "step": 13192
    },
    {
      "epoch": 1.806765269789099,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006859533620665833,
      "loss": 4.8745,
      "step": 13193
    },
    {
      "epoch": 1.8069022185702548,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006859184786844816,
      "loss": 4.8226,
      "step": 13194
    },
    {
      "epoch": 1.8070391673514106,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006858835933472671,
      "loss": 4.8053,
      "step": 13195
    },
    {
      "epoch": 1.8071761161325663,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006858487060552293,
      "loss": 4.7611,
      "step": 13196
    },
    {
      "epoch": 1.8073130649137221,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000685813816808657,
      "loss": 4.8038,
      "step": 13197
    },
    {
      "epoch": 1.8074500136948781,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006857789256078394,
      "loss": 4.726,
      "step": 13198
    },
    {
      "epoch": 1.8075869624760341,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006857440324530655,
      "loss": 4.9075,
      "step": 13199
    },
    {
      "epoch": 1.80772391125719,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006857091373446247,
      "loss": 4.7677,
      "step": 13200
    },
    {
      "epoch": 1.8078608600383457,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006856742402828062,
      "loss": 4.8418,
      "step": 13201
    },
    {
      "epoch": 1.8079978088195015,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006856393412678991,
      "loss": 4.8915,
      "step": 13202
    },
    {
      "epoch": 1.8081347576006572,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006856044403001926,
      "loss": 4.7398,
      "step": 13203
    },
    {
      "epoch": 1.8082717063818132,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000685569537379976,
      "loss": 4.8569,
      "step": 13204
    },
    {
      "epoch": 1.808408655162969,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006855346325075385,
      "loss": 4.784,
      "step": 13205
    },
    {
      "epoch": 1.808545603944125,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006854997256831693,
      "loss": 4.8171,
      "step": 13206
    },
    {
      "epoch": 1.8086825527252808,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006854648169071579,
      "loss": 4.8452,
      "step": 13207
    },
    {
      "epoch": 1.8088195015064366,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006854299061797933,
      "loss": 4.848,
      "step": 13208
    },
    {
      "epoch": 1.8089564502875923,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006853949935013649,
      "loss": 4.8134,
      "step": 13209
    },
    {
      "epoch": 1.8090933990687483,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006853600788721623,
      "loss": 4.7732,
      "step": 13210
    },
    {
      "epoch": 1.809230347849904,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006853251622924746,
      "loss": 4.8164,
      "step": 13211
    },
    {
      "epoch": 1.80936729663106,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006852902437625911,
      "loss": 4.7823,
      "step": 13212
    },
    {
      "epoch": 1.8095042454122159,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006852553232828013,
      "loss": 4.8008,
      "step": 13213
    },
    {
      "epoch": 1.8096411941933717,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006852204008533949,
      "loss": 4.8559,
      "step": 13214
    },
    {
      "epoch": 1.8097781429745274,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006851854764746606,
      "loss": 4.8086,
      "step": 13215
    },
    {
      "epoch": 1.8099150917556832,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006851505501468883,
      "loss": 4.7725,
      "step": 13216
    },
    {
      "epoch": 1.8100520405368392,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006851156218703675,
      "loss": 4.8139,
      "step": 13217
    },
    {
      "epoch": 1.8101889893179952,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006850806916453874,
      "loss": 4.8648,
      "step": 13218
    },
    {
      "epoch": 1.810325938099151,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006850457594722375,
      "loss": 4.8792,
      "step": 13219
    },
    {
      "epoch": 1.8104628868803068,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006850108253512076,
      "loss": 4.7991,
      "step": 13220
    },
    {
      "epoch": 1.8105998356614625,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006849758892825868,
      "loss": 4.8283,
      "step": 13221
    },
    {
      "epoch": 1.8107367844426183,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006849409512666649,
      "loss": 4.7877,
      "step": 13222
    },
    {
      "epoch": 1.8108737332237743,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006849060113037313,
      "loss": 4.7976,
      "step": 13223
    },
    {
      "epoch": 1.8110106820049303,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006848710693940758,
      "loss": 4.7546,
      "step": 13224
    },
    {
      "epoch": 1.811147630786086,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006848361255379874,
      "loss": 4.8093,
      "step": 13225
    },
    {
      "epoch": 1.8112845795672419,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006848011797357565,
      "loss": 4.7796,
      "step": 13226
    },
    {
      "epoch": 1.8114215283483976,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000684766231987672,
      "loss": 4.8343,
      "step": 13227
    },
    {
      "epoch": 1.8115584771295534,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006847312822940238,
      "loss": 4.7929,
      "step": 13228
    },
    {
      "epoch": 1.8116954259107094,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006846963306551017,
      "loss": 4.738,
      "step": 13229
    },
    {
      "epoch": 1.8118323746918652,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006846613770711949,
      "loss": 4.8791,
      "step": 13230
    },
    {
      "epoch": 1.8119693234730212,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006846264215425937,
      "loss": 4.8385,
      "step": 13231
    },
    {
      "epoch": 1.812106272254177,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006845914640695872,
      "loss": 4.8056,
      "step": 13232
    },
    {
      "epoch": 1.8122432210353328,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006845565046524654,
      "loss": 4.8538,
      "step": 13233
    },
    {
      "epoch": 1.8123801698164885,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000684521543291518,
      "loss": 4.7382,
      "step": 13234
    },
    {
      "epoch": 1.8125171185976445,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006844865799870346,
      "loss": 4.8508,
      "step": 13235
    },
    {
      "epoch": 1.8126540673788003,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006844516147393052,
      "loss": 4.829,
      "step": 13236
    },
    {
      "epoch": 1.8127910161599563,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006844166475486192,
      "loss": 4.8189,
      "step": 13237
    },
    {
      "epoch": 1.812927964941112,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006843816784152667,
      "loss": 4.8526,
      "step": 13238
    },
    {
      "epoch": 1.8130649137222679,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006843467073395374,
      "loss": 4.8779,
      "step": 13239
    },
    {
      "epoch": 1.8132018625034236,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006843117343217211,
      "loss": 4.8754,
      "step": 13240
    },
    {
      "epoch": 1.8133388112845794,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006842767593621076,
      "loss": 4.8368,
      "step": 13241
    },
    {
      "epoch": 1.8134757600657354,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006842417824609866,
      "loss": 4.7502,
      "step": 13242
    },
    {
      "epoch": 1.8136127088468914,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006842068036186483,
      "loss": 4.8066,
      "step": 13243
    },
    {
      "epoch": 1.8137496576280472,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006841718228353823,
      "loss": 4.8259,
      "step": 13244
    },
    {
      "epoch": 1.813886606409203,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006841368401114789,
      "loss": 4.8586,
      "step": 13245
    },
    {
      "epoch": 1.8140235551903587,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006841018554472274,
      "loss": 4.7866,
      "step": 13246
    },
    {
      "epoch": 1.8141605039715145,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006840668688429183,
      "loss": 4.7987,
      "step": 13247
    },
    {
      "epoch": 1.8142974527526705,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000684031880298841,
      "loss": 4.8226,
      "step": 13248
    },
    {
      "epoch": 1.8144344015338265,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006839968898152858,
      "loss": 4.7376,
      "step": 13249
    },
    {
      "epoch": 1.8145713503149823,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006839618973925428,
      "loss": 4.7784,
      "step": 13250
    },
    {
      "epoch": 1.814708299096138,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006839269030309015,
      "loss": 4.8169,
      "step": 13251
    },
    {
      "epoch": 1.8148452478772938,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006838919067306525,
      "loss": 4.793,
      "step": 13252
    },
    {
      "epoch": 1.8149821966584496,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006838569084920854,
      "loss": 4.7484,
      "step": 13253
    },
    {
      "epoch": 1.8151191454396056,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006838219083154906,
      "loss": 4.7713,
      "step": 13254
    },
    {
      "epoch": 1.8152560942207614,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006837869062011575,
      "loss": 4.7982,
      "step": 13255
    },
    {
      "epoch": 1.8153930430019174,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006837519021493769,
      "loss": 4.845,
      "step": 13256
    },
    {
      "epoch": 1.8155299917830732,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006837168961604386,
      "loss": 4.6605,
      "step": 13257
    },
    {
      "epoch": 1.815666940564229,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006836818882346325,
      "loss": 4.8604,
      "step": 13258
    },
    {
      "epoch": 1.8158038893453847,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006836468783722492,
      "loss": 4.8344,
      "step": 13259
    },
    {
      "epoch": 1.8159408381265407,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006836118665735783,
      "loss": 4.774,
      "step": 13260
    },
    {
      "epoch": 1.8160777869076965,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006835768528389103,
      "loss": 4.8599,
      "step": 13261
    },
    {
      "epoch": 1.8162147356888525,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006835418371685352,
      "loss": 4.8088,
      "step": 13262
    },
    {
      "epoch": 1.8163516844700083,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006835068195627434,
      "loss": 4.8111,
      "step": 13263
    },
    {
      "epoch": 1.816488633251164,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006834718000218247,
      "loss": 4.8666,
      "step": 13264
    },
    {
      "epoch": 1.8166255820323198,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006834367785460697,
      "loss": 4.7577,
      "step": 13265
    },
    {
      "epoch": 1.8167625308134756,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006834017551357685,
      "loss": 4.8269,
      "step": 13266
    },
    {
      "epoch": 1.8168994795946316,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006833667297912113,
      "loss": 4.8178,
      "step": 13267
    },
    {
      "epoch": 1.8170364283757876,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006833317025126885,
      "loss": 4.7142,
      "step": 13268
    },
    {
      "epoch": 1.8171733771569434,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00068329667330049,
      "loss": 4.7974,
      "step": 13269
    },
    {
      "epoch": 1.8173103259380992,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006832616421549066,
      "loss": 4.7818,
      "step": 13270
    },
    {
      "epoch": 1.817447274719255,
      "grad_norm": 0.375,
      "learning_rate": 0.0006832266090762284,
      "loss": 4.7769,
      "step": 13271
    },
    {
      "epoch": 1.8175842235004107,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006831915740647456,
      "loss": 4.71,
      "step": 13272
    },
    {
      "epoch": 1.8177211722815667,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006831565371207488,
      "loss": 4.8154,
      "step": 13273
    },
    {
      "epoch": 1.8178581210627225,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006831214982445283,
      "loss": 4.8184,
      "step": 13274
    },
    {
      "epoch": 1.8179950698438785,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006830864574363741,
      "loss": 4.7947,
      "step": 13275
    },
    {
      "epoch": 1.8181320186250343,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000683051414696577,
      "loss": 4.8012,
      "step": 13276
    },
    {
      "epoch": 1.81826896740619,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0006830163700254273,
      "loss": 4.8179,
      "step": 13277
    },
    {
      "epoch": 1.8184059161873458,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006829813234232155,
      "loss": 4.8004,
      "step": 13278
    },
    {
      "epoch": 1.8185428649685018,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006829462748902318,
      "loss": 4.8161,
      "step": 13279
    },
    {
      "epoch": 1.8186798137496576,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000682911224426767,
      "loss": 4.7223,
      "step": 13280
    },
    {
      "epoch": 1.8188167625308136,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006828761720331114,
      "loss": 4.7015,
      "step": 13281
    },
    {
      "epoch": 1.8189537113119694,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006828411177095553,
      "loss": 4.7843,
      "step": 13282
    },
    {
      "epoch": 1.8190906600931251,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006828060614563895,
      "loss": 4.7272,
      "step": 13283
    },
    {
      "epoch": 1.819227608874281,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006827710032739042,
      "loss": 4.7769,
      "step": 13284
    },
    {
      "epoch": 1.819364557655437,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006827359431623903,
      "loss": 4.8049,
      "step": 13285
    },
    {
      "epoch": 1.8195015064365927,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006827008811221381,
      "loss": 4.7502,
      "step": 13286
    },
    {
      "epoch": 1.8196384552177487,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006826658171534384,
      "loss": 4.8361,
      "step": 13287
    },
    {
      "epoch": 1.8197754039989045,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006826307512565814,
      "loss": 4.8531,
      "step": 13288
    },
    {
      "epoch": 1.8199123527800602,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006825956834318581,
      "loss": 4.7593,
      "step": 13289
    },
    {
      "epoch": 1.820049301561216,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006825606136795589,
      "loss": 4.8158,
      "step": 13290
    },
    {
      "epoch": 1.8201862503423718,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006825255419999744,
      "loss": 4.7309,
      "step": 13291
    },
    {
      "epoch": 1.8203231991235278,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006824904683933956,
      "loss": 4.8648,
      "step": 13292
    },
    {
      "epoch": 1.8204601479046838,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006824553928601125,
      "loss": 4.8645,
      "step": 13293
    },
    {
      "epoch": 1.8205970966858396,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006824203154004164,
      "loss": 4.786,
      "step": 13294
    },
    {
      "epoch": 1.8207340454669954,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006823852360145978,
      "loss": 4.825,
      "step": 13295
    },
    {
      "epoch": 1.8208709942481511,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0006823501547029473,
      "loss": 4.8312,
      "step": 13296
    },
    {
      "epoch": 1.821007943029307,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006823150714657557,
      "loss": 4.8232,
      "step": 13297
    },
    {
      "epoch": 1.821144891810463,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0006822799863033137,
      "loss": 4.7996,
      "step": 13298
    },
    {
      "epoch": 1.8212818405916187,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006822448992159122,
      "loss": 4.836,
      "step": 13299
    },
    {
      "epoch": 1.8214187893727747,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006822098102038419,
      "loss": 4.8066,
      "step": 13300
    },
    {
      "epoch": 1.8215557381539305,
      "grad_norm": 0.375,
      "learning_rate": 0.0006821747192673935,
      "loss": 4.7348,
      "step": 13301
    },
    {
      "epoch": 1.8216926869350862,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0006821396264068578,
      "loss": 4.7769,
      "step": 13302
    },
    {
      "epoch": 1.821829635716242,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006821045316225257,
      "loss": 4.7642,
      "step": 13303
    },
    {
      "epoch": 1.821966584497398,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006820694349146882,
      "loss": 4.6716,
      "step": 13304
    },
    {
      "epoch": 1.8221035332785538,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006820343362836358,
      "loss": 4.8298,
      "step": 13305
    },
    {
      "epoch": 1.8222404820597098,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006819992357296597,
      "loss": 4.7382,
      "step": 13306
    },
    {
      "epoch": 1.8223774308408656,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006819641332530506,
      "loss": 4.8313,
      "step": 13307
    },
    {
      "epoch": 1.8225143796220213,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0006819290288540996,
      "loss": 4.8325,
      "step": 13308
    },
    {
      "epoch": 1.8226513284031771,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006818939225330973,
      "loss": 4.7703,
      "step": 13309
    },
    {
      "epoch": 1.822788277184333,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000681858814290335,
      "loss": 4.797,
      "step": 13310
    },
    {
      "epoch": 1.822925225965489,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006818237041261031,
      "loss": 4.7345,
      "step": 13311
    },
    {
      "epoch": 1.823062174746645,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006817885920406932,
      "loss": 4.8168,
      "step": 13312
    },
    {
      "epoch": 1.8231991235278007,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006817534780343959,
      "loss": 4.8114,
      "step": 13313
    },
    {
      "epoch": 1.8233360723089564,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006817183621075024,
      "loss": 4.797,
      "step": 13314
    },
    {
      "epoch": 1.8234730210901122,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006816832442603036,
      "loss": 4.792,
      "step": 13315
    },
    {
      "epoch": 1.823609969871268,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0006816481244930905,
      "loss": 4.7883,
      "step": 13316
    },
    {
      "epoch": 1.823746918652424,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006816130028061543,
      "loss": 4.7562,
      "step": 13317
    },
    {
      "epoch": 1.82388386743358,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006815778791997857,
      "loss": 4.9225,
      "step": 13318
    },
    {
      "epoch": 1.8240208162147358,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006815427536742762,
      "loss": 4.7421,
      "step": 13319
    },
    {
      "epoch": 1.8241577649958916,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006815076262299168,
      "loss": 4.7708,
      "step": 13320
    },
    {
      "epoch": 1.8242947137770473,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006814724968669983,
      "loss": 4.7747,
      "step": 13321
    },
    {
      "epoch": 1.824431662558203,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006814373655858124,
      "loss": 4.8041,
      "step": 13322
    },
    {
      "epoch": 1.824568611339359,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006814022323866496,
      "loss": 4.8031,
      "step": 13323
    },
    {
      "epoch": 1.8247055601205149,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006813670972698016,
      "loss": 4.7788,
      "step": 13324
    },
    {
      "epoch": 1.8248425089016709,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0006813319602355591,
      "loss": 4.7968,
      "step": 13325
    },
    {
      "epoch": 1.8249794576828267,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006812968212842138,
      "loss": 4.8039,
      "step": 13326
    },
    {
      "epoch": 1.8251164064639824,
      "grad_norm": 0.375,
      "learning_rate": 0.0006812616804160565,
      "loss": 4.7848,
      "step": 13327
    },
    {
      "epoch": 1.8252533552451382,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006812265376313784,
      "loss": 4.7926,
      "step": 13328
    },
    {
      "epoch": 1.8253903040262942,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006811913929304711,
      "loss": 4.8699,
      "step": 13329
    },
    {
      "epoch": 1.82552725280745,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0006811562463136256,
      "loss": 4.7441,
      "step": 13330
    },
    {
      "epoch": 1.825664201588606,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000681121097781133,
      "loss": 4.773,
      "step": 13331
    },
    {
      "epoch": 1.8258011503697618,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006810859473332848,
      "loss": 4.8232,
      "step": 13332
    },
    {
      "epoch": 1.8259380991509175,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006810507949703726,
      "loss": 4.8262,
      "step": 13333
    },
    {
      "epoch": 1.8260750479320733,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006810156406926871,
      "loss": 4.7135,
      "step": 13334
    },
    {
      "epoch": 1.826211996713229,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006809804845005202,
      "loss": 4.836,
      "step": 13335
    },
    {
      "epoch": 1.826348945494385,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006809453263941629,
      "loss": 4.7469,
      "step": 13336
    },
    {
      "epoch": 1.826485894275541,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006809101663739066,
      "loss": 4.7918,
      "step": 13337
    },
    {
      "epoch": 1.8266228430566969,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006808750044400427,
      "loss": 4.7435,
      "step": 13338
    },
    {
      "epoch": 1.8267597918378526,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006808398405928626,
      "loss": 4.8003,
      "step": 13339
    },
    {
      "epoch": 1.8268967406190084,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006808046748326578,
      "loss": 4.7181,
      "step": 13340
    },
    {
      "epoch": 1.8270336894001642,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006807695071597196,
      "loss": 4.6949,
      "step": 13341
    },
    {
      "epoch": 1.8271706381813202,
      "grad_norm": 0.4375,
      "learning_rate": 0.0006807343375743396,
      "loss": 4.7293,
      "step": 13342
    },
    {
      "epoch": 1.8273075869624762,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006806991660768091,
      "loss": 4.8367,
      "step": 13343
    },
    {
      "epoch": 1.827444535743632,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006806639926674196,
      "loss": 4.8285,
      "step": 13344
    },
    {
      "epoch": 1.8275814845247877,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006806288173464627,
      "loss": 4.7244,
      "step": 13345
    },
    {
      "epoch": 1.8277184333059435,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006805936401142299,
      "loss": 4.8474,
      "step": 13346
    },
    {
      "epoch": 1.8278553820870993,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006805584609710125,
      "loss": 4.7186,
      "step": 13347
    },
    {
      "epoch": 1.8279923308682553,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006805232799171022,
      "loss": 4.7667,
      "step": 13348
    },
    {
      "epoch": 1.828129279649411,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006804880969527907,
      "loss": 4.8009,
      "step": 13349
    },
    {
      "epoch": 1.828266228430567,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006804529120783692,
      "loss": 4.8318,
      "step": 13350
    },
    {
      "epoch": 1.8284031772117229,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0006804177252941298,
      "loss": 4.7885,
      "step": 13351
    },
    {
      "epoch": 1.8285401259928786,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006803825366003636,
      "loss": 4.7518,
      "step": 13352
    },
    {
      "epoch": 1.8286770747740344,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006803473459973625,
      "loss": 4.8413,
      "step": 13353
    },
    {
      "epoch": 1.8288140235551904,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000680312153485418,
      "loss": 4.8086,
      "step": 13354
    },
    {
      "epoch": 1.8289509723363462,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0006802769590648219,
      "loss": 4.7481,
      "step": 13355
    },
    {
      "epoch": 1.8290879211175022,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006802417627358656,
      "loss": 4.7244,
      "step": 13356
    },
    {
      "epoch": 1.829224869898658,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000680206564498841,
      "loss": 4.7997,
      "step": 13357
    },
    {
      "epoch": 1.8293618186798137,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006801713643540398,
      "loss": 4.7636,
      "step": 13358
    },
    {
      "epoch": 1.8294987674609695,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006801361623017537,
      "loss": 4.8255,
      "step": 13359
    },
    {
      "epoch": 1.8296357162421253,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006801009583422743,
      "loss": 4.8003,
      "step": 13360
    },
    {
      "epoch": 1.8297726650232813,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006800657524758934,
      "loss": 4.7873,
      "step": 13361
    },
    {
      "epoch": 1.8299096138044373,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000680030544702903,
      "loss": 4.8022,
      "step": 13362
    },
    {
      "epoch": 1.830046562585593,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006799953350235944,
      "loss": 4.7736,
      "step": 13363
    },
    {
      "epoch": 1.8301835113667488,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006799601234382597,
      "loss": 4.8425,
      "step": 13364
    },
    {
      "epoch": 1.8303204601479046,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006799249099471908,
      "loss": 4.8536,
      "step": 13365
    },
    {
      "epoch": 1.8304574089290604,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006798896945506793,
      "loss": 4.8328,
      "step": 13366
    },
    {
      "epoch": 1.8305943577102164,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006798544772490171,
      "loss": 4.7,
      "step": 13367
    },
    {
      "epoch": 1.8307313064913724,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000679819258042496,
      "loss": 4.7468,
      "step": 13368
    },
    {
      "epoch": 1.8308682552725282,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000679784036931408,
      "loss": 4.7519,
      "step": 13369
    },
    {
      "epoch": 1.831005204053684,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006797488139160451,
      "loss": 4.8011,
      "step": 13370
    },
    {
      "epoch": 1.8311421528348397,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006797135889966989,
      "loss": 4.8249,
      "step": 13371
    },
    {
      "epoch": 1.8312791016159955,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006796783621736615,
      "loss": 4.8082,
      "step": 13372
    },
    {
      "epoch": 1.8314160503971515,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006796431334472247,
      "loss": 4.9039,
      "step": 13373
    },
    {
      "epoch": 1.8315529991783073,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006796079028176806,
      "loss": 4.836,
      "step": 13374
    },
    {
      "epoch": 1.8316899479594633,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006795726702853211,
      "loss": 4.8302,
      "step": 13375
    },
    {
      "epoch": 1.831826896740619,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006795374358504381,
      "loss": 4.7878,
      "step": 13376
    },
    {
      "epoch": 1.8319638455217748,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006795021995133237,
      "loss": 4.7899,
      "step": 13377
    },
    {
      "epoch": 1.8321007943029306,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006794669612742699,
      "loss": 4.8275,
      "step": 13378
    },
    {
      "epoch": 1.8322377430840866,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006794317211335686,
      "loss": 4.781,
      "step": 13379
    },
    {
      "epoch": 1.8323746918652424,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006793964790915123,
      "loss": 4.7998,
      "step": 13380
    },
    {
      "epoch": 1.8325116406463984,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006793612351483925,
      "loss": 4.8022,
      "step": 13381
    },
    {
      "epoch": 1.8326485894275542,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006793259893045014,
      "loss": 4.8383,
      "step": 13382
    },
    {
      "epoch": 1.83278553820871,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006792907415601313,
      "loss": 4.7734,
      "step": 13383
    },
    {
      "epoch": 1.8329224869898657,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006792554919155742,
      "loss": 4.7751,
      "step": 13384
    },
    {
      "epoch": 1.8330594357710215,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000679220240371122,
      "loss": 4.8004,
      "step": 13385
    },
    {
      "epoch": 1.8331963845521775,
      "grad_norm": 0.328125,
      "learning_rate": 0.0006791849869270674,
      "loss": 4.8624,
      "step": 13386
    },
    {
      "epoch": 1.8333333333333335,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000679149731583702,
      "loss": 4.8324,
      "step": 13387
    },
    {
      "epoch": 1.8334702821144893,
      "grad_norm": 0.318359375,
      "learning_rate": 0.000679114474341318,
      "loss": 4.765,
      "step": 13388
    },
    {
      "epoch": 1.833607230895645,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006790792152002081,
      "loss": 4.7873,
      "step": 13389
    },
    {
      "epoch": 1.8337441796768008,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0006790439541606639,
      "loss": 4.8073,
      "step": 13390
    },
    {
      "epoch": 1.8338811284579566,
      "grad_norm": 0.31640625,
      "learning_rate": 0.000679008691222978,
      "loss": 4.8182,
      "step": 13391
    },
    {
      "epoch": 1.8340180772391126,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006789734263874425,
      "loss": 4.7704,
      "step": 13392
    },
    {
      "epoch": 1.8341550260202684,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006789381596543497,
      "loss": 4.847,
      "step": 13393
    },
    {
      "epoch": 1.8342919748014244,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006789028910239917,
      "loss": 4.7637,
      "step": 13394
    },
    {
      "epoch": 1.8344289235825801,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000678867620496661,
      "loss": 4.7854,
      "step": 13395
    },
    {
      "epoch": 1.834565872363736,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006788323480726496,
      "loss": 4.7844,
      "step": 13396
    },
    {
      "epoch": 1.8347028211448917,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006787970737522502,
      "loss": 4.7818,
      "step": 13397
    },
    {
      "epoch": 1.8348397699260477,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006787617975357548,
      "loss": 4.7651,
      "step": 13398
    },
    {
      "epoch": 1.8349767187072035,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000678726519423456,
      "loss": 4.7929,
      "step": 13399
    },
    {
      "epoch": 1.8351136674883595,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000678691239415646,
      "loss": 4.6702,
      "step": 13400
    },
    {
      "epoch": 1.8352506162695152,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006786559575126172,
      "loss": 4.7872,
      "step": 13401
    },
    {
      "epoch": 1.835387565050671,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000678620673714662,
      "loss": 4.7198,
      "step": 13402
    },
    {
      "epoch": 1.8355245138318268,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006785853880220728,
      "loss": 4.7412,
      "step": 13403
    },
    {
      "epoch": 1.8356614626129828,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000678550100435142,
      "loss": 4.7684,
      "step": 13404
    },
    {
      "epoch": 1.8357984113941386,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000678514810954162,
      "loss": 4.8465,
      "step": 13405
    },
    {
      "epoch": 1.8359353601752946,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006784795195794252,
      "loss": 4.8229,
      "step": 13406
    },
    {
      "epoch": 1.8360723089564503,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006784442263112245,
      "loss": 4.7286,
      "step": 13407
    },
    {
      "epoch": 1.8362092577376061,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006784089311498518,
      "loss": 4.8149,
      "step": 13408
    },
    {
      "epoch": 1.836346206518762,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006783736340956,
      "loss": 4.7857,
      "step": 13409
    },
    {
      "epoch": 1.8364831552999177,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006783383351487614,
      "loss": 4.862,
      "step": 13410
    },
    {
      "epoch": 1.8366201040810737,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006783030343096288,
      "loss": 4.6886,
      "step": 13411
    },
    {
      "epoch": 1.8367570528622297,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006782677315784943,
      "loss": 4.7681,
      "step": 13412
    },
    {
      "epoch": 1.8368940016433855,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006782324269556508,
      "loss": 4.7581,
      "step": 13413
    },
    {
      "epoch": 1.8370309504245412,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006781971204413908,
      "loss": 4.7783,
      "step": 13414
    },
    {
      "epoch": 1.837167899205697,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006781618120360068,
      "loss": 4.8333,
      "step": 13415
    },
    {
      "epoch": 1.8373048479868528,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0006781265017397917,
      "loss": 4.7182,
      "step": 13416
    },
    {
      "epoch": 1.8374417967680088,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006780911895530377,
      "loss": 4.7507,
      "step": 13417
    },
    {
      "epoch": 1.8375787455491646,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006780558754760379,
      "loss": 4.8628,
      "step": 13418
    },
    {
      "epoch": 1.8377156943303206,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006780205595090846,
      "loss": 4.7911,
      "step": 13419
    },
    {
      "epoch": 1.8378526431114763,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006779852416524706,
      "loss": 4.8132,
      "step": 13420
    },
    {
      "epoch": 1.8379895918926321,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006779499219064885,
      "loss": 4.7927,
      "step": 13421
    },
    {
      "epoch": 1.838126540673788,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006779146002714311,
      "loss": 4.8477,
      "step": 13422
    },
    {
      "epoch": 1.8382634894549439,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006778792767475911,
      "loss": 4.651,
      "step": 13423
    },
    {
      "epoch": 1.8384004382360997,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006778439513352612,
      "loss": 4.8167,
      "step": 13424
    },
    {
      "epoch": 1.8385373870172557,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006778086240347343,
      "loss": 4.8486,
      "step": 13425
    },
    {
      "epoch": 1.8386743357984114,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000677773294846303,
      "loss": 4.7721,
      "step": 13426
    },
    {
      "epoch": 1.8388112845795672,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006777379637702601,
      "loss": 4.77,
      "step": 13427
    },
    {
      "epoch": 1.838948233360723,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006777026308068984,
      "loss": 4.756,
      "step": 13428
    },
    {
      "epoch": 1.8390851821418788,
      "grad_norm": 0.359375,
      "learning_rate": 0.000677667295956511,
      "loss": 4.8306,
      "step": 13429
    },
    {
      "epoch": 1.8392221309230348,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006776319592193902,
      "loss": 4.744,
      "step": 13430
    },
    {
      "epoch": 1.8393590797041908,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006775966205958291,
      "loss": 4.7891,
      "step": 13431
    },
    {
      "epoch": 1.8394960284853465,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006775612800861207,
      "loss": 4.774,
      "step": 13432
    },
    {
      "epoch": 1.8396329772665023,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006775259376905577,
      "loss": 4.8528,
      "step": 13433
    },
    {
      "epoch": 1.839769926047658,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000677490593409433,
      "loss": 4.8299,
      "step": 13434
    },
    {
      "epoch": 1.8399068748288139,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006774552472430395,
      "loss": 4.7816,
      "step": 13435
    },
    {
      "epoch": 1.8400438236099699,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006774198991916704,
      "loss": 4.7328,
      "step": 13436
    },
    {
      "epoch": 1.8401807723911259,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006773845492556182,
      "loss": 4.8587,
      "step": 13437
    },
    {
      "epoch": 1.8403177211722817,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006773491974351761,
      "loss": 4.7896,
      "step": 13438
    },
    {
      "epoch": 1.8404546699534374,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006773138437306369,
      "loss": 4.7668,
      "step": 13439
    },
    {
      "epoch": 1.8405916187345932,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006772784881422939,
      "loss": 4.8556,
      "step": 13440
    },
    {
      "epoch": 1.840728567515749,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006772431306704399,
      "loss": 4.7511,
      "step": 13441
    },
    {
      "epoch": 1.840865516296905,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006772077713153678,
      "loss": 4.8392,
      "step": 13442
    },
    {
      "epoch": 1.8410024650780608,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006771724100773709,
      "loss": 4.7508,
      "step": 13443
    },
    {
      "epoch": 1.8411394138592168,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000677137046956742,
      "loss": 4.7845,
      "step": 13444
    },
    {
      "epoch": 1.8412763626403725,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006771016819537744,
      "loss": 4.8064,
      "step": 13445
    },
    {
      "epoch": 1.8414133114215283,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006770663150687608,
      "loss": 4.7743,
      "step": 13446
    },
    {
      "epoch": 1.841550260202684,
      "grad_norm": 0.375,
      "learning_rate": 0.0006770309463019949,
      "loss": 4.7677,
      "step": 13447
    },
    {
      "epoch": 1.84168720898384,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006769955756537691,
      "loss": 4.7885,
      "step": 13448
    },
    {
      "epoch": 1.8418241577649959,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006769602031243769,
      "loss": 4.8032,
      "step": 13449
    },
    {
      "epoch": 1.8419611065461519,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006769248287141117,
      "loss": 4.8392,
      "step": 13450
    },
    {
      "epoch": 1.8420980553273076,
      "grad_norm": 0.375,
      "learning_rate": 0.0006768894524232661,
      "loss": 4.7806,
      "step": 13451
    },
    {
      "epoch": 1.8422350041084634,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006768540742521338,
      "loss": 4.8064,
      "step": 13452
    },
    {
      "epoch": 1.8423719528896192,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006768186942010075,
      "loss": 4.7715,
      "step": 13453
    },
    {
      "epoch": 1.842508901670775,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006767833122701808,
      "loss": 4.7649,
      "step": 13454
    },
    {
      "epoch": 1.842645850451931,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006767479284599465,
      "loss": 4.7817,
      "step": 13455
    },
    {
      "epoch": 1.842782799233087,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006767125427705983,
      "loss": 4.8082,
      "step": 13456
    },
    {
      "epoch": 1.8429197480142427,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006766771552024292,
      "loss": 4.7407,
      "step": 13457
    },
    {
      "epoch": 1.8430566967953985,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006766417657557323,
      "loss": 4.7952,
      "step": 13458
    },
    {
      "epoch": 1.8431936455765543,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006766063744308015,
      "loss": 4.7963,
      "step": 13459
    },
    {
      "epoch": 1.84333059435771,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006765709812279293,
      "loss": 4.8281,
      "step": 13460
    },
    {
      "epoch": 1.843467543138866,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006765355861474096,
      "loss": 4.7783,
      "step": 13461
    },
    {
      "epoch": 1.843604491920022,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006765001891895355,
      "loss": 4.7429,
      "step": 13462
    },
    {
      "epoch": 1.8437414407011778,
      "grad_norm": 0.375,
      "learning_rate": 0.0006764647903546001,
      "loss": 4.7911,
      "step": 13463
    },
    {
      "epoch": 1.8438783894823336,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006764293896428974,
      "loss": 4.8573,
      "step": 13464
    },
    {
      "epoch": 1.8440153382634894,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006763939870547201,
      "loss": 4.8012,
      "step": 13465
    },
    {
      "epoch": 1.8441522870446452,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006763585825903622,
      "loss": 4.7636,
      "step": 13466
    },
    {
      "epoch": 1.8442892358258012,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006763231762501165,
      "loss": 4.8424,
      "step": 13467
    },
    {
      "epoch": 1.844426184606957,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006762877680342768,
      "loss": 4.7931,
      "step": 13468
    },
    {
      "epoch": 1.844563133388113,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006762523579431364,
      "loss": 4.7845,
      "step": 13469
    },
    {
      "epoch": 1.8447000821692687,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006762169459769889,
      "loss": 4.8313,
      "step": 13470
    },
    {
      "epoch": 1.8448370309504245,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006761815321361276,
      "loss": 4.7793,
      "step": 13471
    },
    {
      "epoch": 1.8449739797315803,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000676146116420846,
      "loss": 4.7642,
      "step": 13472
    },
    {
      "epoch": 1.8451109285127363,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006761106988314378,
      "loss": 4.8344,
      "step": 13473
    },
    {
      "epoch": 1.845247877293892,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006760752793681962,
      "loss": 4.8136,
      "step": 13474
    },
    {
      "epoch": 1.845384826075048,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006760398580314149,
      "loss": 4.7279,
      "step": 13475
    },
    {
      "epoch": 1.8455217748562038,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006760044348213875,
      "loss": 4.7281,
      "step": 13476
    },
    {
      "epoch": 1.8456587236373596,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006759690097384075,
      "loss": 4.7877,
      "step": 13477
    },
    {
      "epoch": 1.8457956724185154,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006759335827827684,
      "loss": 4.7211,
      "step": 13478
    },
    {
      "epoch": 1.8459326211996712,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006758981539547641,
      "loss": 4.7981,
      "step": 13479
    },
    {
      "epoch": 1.8460695699808272,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006758627232546876,
      "loss": 4.7882,
      "step": 13480
    },
    {
      "epoch": 1.8462065187619832,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006758272906828331,
      "loss": 4.7887,
      "step": 13481
    },
    {
      "epoch": 1.846343467543139,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000675791856239494,
      "loss": 4.8521,
      "step": 13482
    },
    {
      "epoch": 1.8464804163242947,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000675756419924964,
      "loss": 4.783,
      "step": 13483
    },
    {
      "epoch": 1.8466173651054505,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006757209817395368,
      "loss": 4.8081,
      "step": 13484
    },
    {
      "epoch": 1.8467543138866063,
      "grad_norm": 0.390625,
      "learning_rate": 0.000675685541683506,
      "loss": 4.8251,
      "step": 13485
    },
    {
      "epoch": 1.8468912626677623,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006756500997571652,
      "loss": 4.7796,
      "step": 13486
    },
    {
      "epoch": 1.8470282114489183,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006756146559608085,
      "loss": 4.8407,
      "step": 13487
    },
    {
      "epoch": 1.847165160230074,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006755792102947292,
      "loss": 4.8006,
      "step": 13488
    },
    {
      "epoch": 1.8473021090112298,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006755437627592214,
      "loss": 4.7789,
      "step": 13489
    },
    {
      "epoch": 1.8474390577923856,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006755083133545784,
      "loss": 4.8199,
      "step": 13490
    },
    {
      "epoch": 1.8475760065735414,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006754728620810944,
      "loss": 4.7985,
      "step": 13491
    },
    {
      "epoch": 1.8477129553546974,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006754374089390633,
      "loss": 4.8014,
      "step": 13492
    },
    {
      "epoch": 1.8478499041358531,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006754019539287784,
      "loss": 4.7516,
      "step": 13493
    },
    {
      "epoch": 1.8479868529170091,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006753664970505339,
      "loss": 4.7722,
      "step": 13494
    },
    {
      "epoch": 1.848123801698165,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006753310383046234,
      "loss": 4.8014,
      "step": 13495
    },
    {
      "epoch": 1.8482607504793207,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006752955776913411,
      "loss": 4.8223,
      "step": 13496
    },
    {
      "epoch": 1.8483976992604765,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006752601152109805,
      "loss": 4.8666,
      "step": 13497
    },
    {
      "epoch": 1.8485346480416325,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006752246508638359,
      "loss": 4.7395,
      "step": 13498
    },
    {
      "epoch": 1.8486715968227883,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006751891846502007,
      "loss": 4.6957,
      "step": 13499
    },
    {
      "epoch": 1.8488085456039443,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006751537165703691,
      "loss": 4.7882,
      "step": 13500
    },
    {
      "epoch": 1.8489454943851,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000675118246624635,
      "loss": 4.8399,
      "step": 13501
    },
    {
      "epoch": 1.8490824431662558,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006750827748132924,
      "loss": 4.8557,
      "step": 13502
    },
    {
      "epoch": 1.8492193919474116,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006750473011366353,
      "loss": 4.736,
      "step": 13503
    },
    {
      "epoch": 1.8493563407285674,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006750118255949574,
      "loss": 4.7945,
      "step": 13504
    },
    {
      "epoch": 1.8494932895097234,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006749763481885531,
      "loss": 4.7655,
      "step": 13505
    },
    {
      "epoch": 1.8496302382908794,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000674940868917716,
      "loss": 4.743,
      "step": 13506
    },
    {
      "epoch": 1.8497671870720351,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006749053877827404,
      "loss": 4.8578,
      "step": 13507
    },
    {
      "epoch": 1.849904135853191,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006748699047839202,
      "loss": 4.7843,
      "step": 13508
    },
    {
      "epoch": 1.8500410846343467,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006748344199215495,
      "loss": 4.7839,
      "step": 13509
    },
    {
      "epoch": 1.8501780334155025,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006747989331959226,
      "loss": 4.6917,
      "step": 13510
    },
    {
      "epoch": 1.8503149821966585,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006747634446073331,
      "loss": 4.8405,
      "step": 13511
    },
    {
      "epoch": 1.8504519309778145,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006747279541560755,
      "loss": 4.7582,
      "step": 13512
    },
    {
      "epoch": 1.8505888797589702,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006746924618424438,
      "loss": 4.8262,
      "step": 13513
    },
    {
      "epoch": 1.850725828540126,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006746569676667323,
      "loss": 4.7765,
      "step": 13514
    },
    {
      "epoch": 1.8508627773212818,
      "grad_norm": 0.4375,
      "learning_rate": 0.0006746214716292346,
      "loss": 4.7851,
      "step": 13515
    },
    {
      "epoch": 1.8509997261024376,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006745859737302456,
      "loss": 4.8602,
      "step": 13516
    },
    {
      "epoch": 1.8511366748835936,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006745504739700588,
      "loss": 4.6859,
      "step": 13517
    },
    {
      "epoch": 1.8512736236647493,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006745149723489689,
      "loss": 4.7696,
      "step": 13518
    },
    {
      "epoch": 1.8514105724459053,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00067447946886727,
      "loss": 4.7805,
      "step": 13519
    },
    {
      "epoch": 1.8515475212270611,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006744439635252562,
      "loss": 4.7531,
      "step": 13520
    },
    {
      "epoch": 1.851684470008217,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006744084563232218,
      "loss": 4.8039,
      "step": 13521
    },
    {
      "epoch": 1.8518214187893727,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006743729472614608,
      "loss": 4.8133,
      "step": 13522
    },
    {
      "epoch": 1.8519583675705287,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000674337436340268,
      "loss": 4.7804,
      "step": 13523
    },
    {
      "epoch": 1.8520953163516845,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006743019235599373,
      "loss": 4.7459,
      "step": 13524
    },
    {
      "epoch": 1.8522322651328404,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006742664089207631,
      "loss": 4.7869,
      "step": 13525
    },
    {
      "epoch": 1.8523692139139962,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006742308924230398,
      "loss": 4.7264,
      "step": 13526
    },
    {
      "epoch": 1.852506162695152,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006741953740670616,
      "loss": 4.8318,
      "step": 13527
    },
    {
      "epoch": 1.8526431114763078,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006741598538531229,
      "loss": 4.798,
      "step": 13528
    },
    {
      "epoch": 1.8527800602574636,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006741243317815181,
      "loss": 4.8585,
      "step": 13529
    },
    {
      "epoch": 1.8529170090386196,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006740888078525417,
      "loss": 4.6914,
      "step": 13530
    },
    {
      "epoch": 1.8530539578197756,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006740532820664878,
      "loss": 4.7296,
      "step": 13531
    },
    {
      "epoch": 1.8531909066009313,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006740177544236509,
      "loss": 4.7939,
      "step": 13532
    },
    {
      "epoch": 1.853327855382087,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006739822249243255,
      "loss": 4.8228,
      "step": 13533
    },
    {
      "epoch": 1.8534648041632429,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006739466935688063,
      "loss": 4.7929,
      "step": 13534
    },
    {
      "epoch": 1.8536017529443987,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006739111603573872,
      "loss": 4.7256,
      "step": 13535
    },
    {
      "epoch": 1.8537387017255547,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000673875625290363,
      "loss": 4.8556,
      "step": 13536
    },
    {
      "epoch": 1.8538756505067104,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006738400883680283,
      "loss": 4.863,
      "step": 13537
    },
    {
      "epoch": 1.8540125992878664,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006738045495906772,
      "loss": 4.8284,
      "step": 13538
    },
    {
      "epoch": 1.8541495480690222,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006737690089586046,
      "loss": 4.7378,
      "step": 13539
    },
    {
      "epoch": 1.854286496850178,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000673733466472105,
      "loss": 4.8048,
      "step": 13540
    },
    {
      "epoch": 1.8544234456313338,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006736979221314729,
      "loss": 4.7763,
      "step": 13541
    },
    {
      "epoch": 1.8545603944124898,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006736623759370026,
      "loss": 4.6661,
      "step": 13542
    },
    {
      "epoch": 1.8546973431936455,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006736268278889889,
      "loss": 4.7784,
      "step": 13543
    },
    {
      "epoch": 1.8548342919748015,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006735912779877264,
      "loss": 4.8253,
      "step": 13544
    },
    {
      "epoch": 1.8549712407559573,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006735557262335098,
      "loss": 4.7752,
      "step": 13545
    },
    {
      "epoch": 1.855108189537113,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006735201726266336,
      "loss": 4.8549,
      "step": 13546
    },
    {
      "epoch": 1.8552451383182689,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006734846171673924,
      "loss": 4.8389,
      "step": 13547
    },
    {
      "epoch": 1.8553820870994249,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006734490598560811,
      "loss": 4.7909,
      "step": 13548
    },
    {
      "epoch": 1.8555190358805806,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000673413500692994,
      "loss": 4.8073,
      "step": 13549
    },
    {
      "epoch": 1.8556559846617366,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006733779396784261,
      "loss": 4.77,
      "step": 13550
    },
    {
      "epoch": 1.8557929334428924,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006733423768126719,
      "loss": 4.7979,
      "step": 13551
    },
    {
      "epoch": 1.8559298822240482,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006733068120960263,
      "loss": 4.7055,
      "step": 13552
    },
    {
      "epoch": 1.856066831005204,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006732712455287839,
      "loss": 4.7411,
      "step": 13553
    },
    {
      "epoch": 1.8562037797863598,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006732356771112395,
      "loss": 4.7737,
      "step": 13554
    },
    {
      "epoch": 1.8563407285675158,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006732001068436878,
      "loss": 4.8476,
      "step": 13555
    },
    {
      "epoch": 1.8564776773486718,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006731645347264236,
      "loss": 4.8256,
      "step": 13556
    },
    {
      "epoch": 1.8566146261298275,
      "grad_norm": 0.375,
      "learning_rate": 0.0006731289607597419,
      "loss": 4.7261,
      "step": 13557
    },
    {
      "epoch": 1.8567515749109833,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006730933849439371,
      "loss": 4.8024,
      "step": 13558
    },
    {
      "epoch": 1.856888523692139,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006730578072793044,
      "loss": 4.8462,
      "step": 13559
    },
    {
      "epoch": 1.8570254724732949,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006730222277661384,
      "loss": 4.7897,
      "step": 13560
    },
    {
      "epoch": 1.8571624212544509,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0006729866464047342,
      "loss": 4.8443,
      "step": 13561
    },
    {
      "epoch": 1.8572993700356066,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006729510631953864,
      "loss": 4.7903,
      "step": 13562
    },
    {
      "epoch": 1.8574363188167626,
      "grad_norm": 0.34375,
      "learning_rate": 0.00067291547813839,
      "loss": 4.7638,
      "step": 13563
    },
    {
      "epoch": 1.8575732675979184,
      "grad_norm": 0.328125,
      "learning_rate": 0.00067287989123404,
      "loss": 4.8227,
      "step": 13564
    },
    {
      "epoch": 1.8577102163790742,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006728443024826312,
      "loss": 4.8472,
      "step": 13565
    },
    {
      "epoch": 1.85784716516023,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006728087118844586,
      "loss": 4.8205,
      "step": 13566
    },
    {
      "epoch": 1.857984113941386,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006727731194398171,
      "loss": 4.7576,
      "step": 13567
    },
    {
      "epoch": 1.8581210627225417,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006727375251490017,
      "loss": 4.8287,
      "step": 13568
    },
    {
      "epoch": 1.8582580115036977,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006727019290123073,
      "loss": 4.8386,
      "step": 13569
    },
    {
      "epoch": 1.8583949602848535,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000672666331030029,
      "loss": 4.8432,
      "step": 13570
    },
    {
      "epoch": 1.8585319090660093,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006726307312024617,
      "loss": 4.7507,
      "step": 13571
    },
    {
      "epoch": 1.858668857847165,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006725951295299005,
      "loss": 4.8246,
      "step": 13572
    },
    {
      "epoch": 1.8588058066283208,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006725595260126404,
      "loss": 4.7974,
      "step": 13573
    },
    {
      "epoch": 1.8589427554094768,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006725239206509764,
      "loss": 4.7423,
      "step": 13574
    },
    {
      "epoch": 1.8590797041906328,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006724883134452038,
      "loss": 4.8388,
      "step": 13575
    },
    {
      "epoch": 1.8592166529717886,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006724527043956175,
      "loss": 4.6986,
      "step": 13576
    },
    {
      "epoch": 1.8593536017529444,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006724170935025126,
      "loss": 4.8414,
      "step": 13577
    },
    {
      "epoch": 1.8594905505341002,
      "grad_norm": 0.375,
      "learning_rate": 0.0006723814807661841,
      "loss": 4.8442,
      "step": 13578
    },
    {
      "epoch": 1.859627499315256,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006723458661869276,
      "loss": 4.8048,
      "step": 13579
    },
    {
      "epoch": 1.859764448096412,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006723102497650377,
      "loss": 4.7844,
      "step": 13580
    },
    {
      "epoch": 1.859901396877568,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006722746315008098,
      "loss": 4.7713,
      "step": 13581
    },
    {
      "epoch": 1.8600383456587237,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006722390113945392,
      "loss": 4.7488,
      "step": 13582
    },
    {
      "epoch": 1.8601752944398795,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006722033894465207,
      "loss": 4.8256,
      "step": 13583
    },
    {
      "epoch": 1.8603122432210353,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00067216776565705,
      "loss": 4.8116,
      "step": 13584
    },
    {
      "epoch": 1.860449192002191,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006721321400264221,
      "loss": 4.7895,
      "step": 13585
    },
    {
      "epoch": 1.860586140783347,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000672096512554932,
      "loss": 4.6792,
      "step": 13586
    },
    {
      "epoch": 1.8607230895645028,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006720608832428754,
      "loss": 4.8068,
      "step": 13587
    },
    {
      "epoch": 1.8608600383456588,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006720252520905473,
      "loss": 4.7442,
      "step": 13588
    },
    {
      "epoch": 1.8609969871268146,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000671989619098243,
      "loss": 4.8266,
      "step": 13589
    },
    {
      "epoch": 1.8611339359079704,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006719539842662577,
      "loss": 4.8554,
      "step": 13590
    },
    {
      "epoch": 1.8612708846891262,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000671918347594887,
      "loss": 4.7215,
      "step": 13591
    },
    {
      "epoch": 1.8614078334702822,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0006718827090844259,
      "loss": 4.9357,
      "step": 13592
    },
    {
      "epoch": 1.861544782251438,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006718470687351702,
      "loss": 4.8136,
      "step": 13593
    },
    {
      "epoch": 1.861681731032594,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0006718114265474148,
      "loss": 4.8126,
      "step": 13594
    },
    {
      "epoch": 1.8618186798137497,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006717757825214553,
      "loss": 4.8308,
      "step": 13595
    },
    {
      "epoch": 1.8619556285949055,
      "grad_norm": 0.484375,
      "learning_rate": 0.0006717401366575869,
      "loss": 4.77,
      "step": 13596
    },
    {
      "epoch": 1.8620925773760613,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006717044889561054,
      "loss": 4.8588,
      "step": 13597
    },
    {
      "epoch": 1.862229526157217,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0006716688394173058,
      "loss": 4.7118,
      "step": 13598
    },
    {
      "epoch": 1.862366474938373,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006716331880414838,
      "loss": 4.7589,
      "step": 13599
    },
    {
      "epoch": 1.862503423719529,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006715975348289346,
      "loss": 4.7558,
      "step": 13600
    },
    {
      "epoch": 1.8626403725006848,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000671561879779954,
      "loss": 4.7092,
      "step": 13601
    },
    {
      "epoch": 1.8627773212818406,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006715262228948372,
      "loss": 4.8397,
      "step": 13602
    },
    {
      "epoch": 1.8629142700629964,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006714905641738798,
      "loss": 4.844,
      "step": 13603
    },
    {
      "epoch": 1.8630512188441521,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006714549036173775,
      "loss": 4.7265,
      "step": 13604
    },
    {
      "epoch": 1.8631881676253081,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006714192412256253,
      "loss": 4.7914,
      "step": 13605
    },
    {
      "epoch": 1.8633251164064641,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006713835769989195,
      "loss": 4.7537,
      "step": 13606
    },
    {
      "epoch": 1.86346206518762,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000671347910937555,
      "loss": 4.7598,
      "step": 13607
    },
    {
      "epoch": 1.8635990139687757,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006713122430418276,
      "loss": 4.8275,
      "step": 13608
    },
    {
      "epoch": 1.8637359627499315,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000671276573312033,
      "loss": 4.7739,
      "step": 13609
    },
    {
      "epoch": 1.8638729115310873,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006712409017484666,
      "loss": 4.7781,
      "step": 13610
    },
    {
      "epoch": 1.8640098603122432,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006712052283514243,
      "loss": 4.7707,
      "step": 13611
    },
    {
      "epoch": 1.864146809093399,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006711695531212013,
      "loss": 4.7524,
      "step": 13612
    },
    {
      "epoch": 1.864283757874555,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006711338760580938,
      "loss": 4.7504,
      "step": 13613
    },
    {
      "epoch": 1.8644207066557108,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000671098197162397,
      "loss": 4.801,
      "step": 13614
    },
    {
      "epoch": 1.8645576554368666,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006710625164344068,
      "loss": 4.7618,
      "step": 13615
    },
    {
      "epoch": 1.8646946042180224,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006710268338744189,
      "loss": 4.8402,
      "step": 13616
    },
    {
      "epoch": 1.8648315529991784,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006709911494827289,
      "loss": 4.7598,
      "step": 13617
    },
    {
      "epoch": 1.8649685017803341,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006709554632596326,
      "loss": 4.7552,
      "step": 13618
    },
    {
      "epoch": 1.8651054505614901,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006709197752054257,
      "loss": 4.7582,
      "step": 13619
    },
    {
      "epoch": 1.865242399342646,
      "grad_norm": 0.359375,
      "learning_rate": 0.000670884085320404,
      "loss": 4.8175,
      "step": 13620
    },
    {
      "epoch": 1.8653793481238017,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006708483936048633,
      "loss": 4.8118,
      "step": 13621
    },
    {
      "epoch": 1.8655162969049575,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006708127000590992,
      "loss": 4.8657,
      "step": 13622
    },
    {
      "epoch": 1.8656532456861132,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006707770046834077,
      "loss": 4.877,
      "step": 13623
    },
    {
      "epoch": 1.8657901944672692,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006707413074780846,
      "loss": 4.7355,
      "step": 13624
    },
    {
      "epoch": 1.8659271432484252,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006707056084434257,
      "loss": 4.7651,
      "step": 13625
    },
    {
      "epoch": 1.866064092029581,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006706699075797269,
      "loss": 4.7634,
      "step": 13626
    },
    {
      "epoch": 1.8662010408107368,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006706342048872838,
      "loss": 4.7753,
      "step": 13627
    },
    {
      "epoch": 1.8663379895918926,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006705985003663926,
      "loss": 4.7326,
      "step": 13628
    },
    {
      "epoch": 1.8664749383730483,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000670562794017349,
      "loss": 4.7551,
      "step": 13629
    },
    {
      "epoch": 1.8666118871542043,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000670527085840449,
      "loss": 4.7314,
      "step": 13630
    },
    {
      "epoch": 1.8667488359353603,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006704913758359886,
      "loss": 4.7312,
      "step": 13631
    },
    {
      "epoch": 1.8668857847165161,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006704556640042634,
      "loss": 4.7622,
      "step": 13632
    },
    {
      "epoch": 1.867022733497672,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006704199503455698,
      "loss": 4.7568,
      "step": 13633
    },
    {
      "epoch": 1.8671596822788277,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006703842348602033,
      "loss": 4.763,
      "step": 13634
    },
    {
      "epoch": 1.8672966310599834,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006703485175484603,
      "loss": 4.8142,
      "step": 13635
    },
    {
      "epoch": 1.8674335798411394,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006703127984106366,
      "loss": 4.8154,
      "step": 13636
    },
    {
      "epoch": 1.8675705286222952,
      "grad_norm": 0.375,
      "learning_rate": 0.0006702770774470283,
      "loss": 4.8763,
      "step": 13637
    },
    {
      "epoch": 1.8677074774034512,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006702413546579313,
      "loss": 4.8558,
      "step": 13638
    },
    {
      "epoch": 1.867844426184607,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006702056300436416,
      "loss": 4.8002,
      "step": 13639
    },
    {
      "epoch": 1.8679813749657628,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006701699036044556,
      "loss": 4.7516,
      "step": 13640
    },
    {
      "epoch": 1.8681183237469186,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000670134175340669,
      "loss": 4.8653,
      "step": 13641
    },
    {
      "epoch": 1.8682552725280746,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006700984452525781,
      "loss": 4.7552,
      "step": 13642
    },
    {
      "epoch": 1.8683922213092303,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006700627133404789,
      "loss": 4.6994,
      "step": 13643
    },
    {
      "epoch": 1.8685291700903863,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006700269796046677,
      "loss": 4.8057,
      "step": 13644
    },
    {
      "epoch": 1.868666118871542,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006699912440454402,
      "loss": 4.8374,
      "step": 13645
    },
    {
      "epoch": 1.8688030676526979,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006699555066630931,
      "loss": 4.8111,
      "step": 13646
    },
    {
      "epoch": 1.8689400164338537,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006699197674579221,
      "loss": 4.8993,
      "step": 13647
    },
    {
      "epoch": 1.8690769652150094,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006698840264302237,
      "loss": 4.8209,
      "step": 13648
    },
    {
      "epoch": 1.8692139139961654,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000669848283580294,
      "loss": 4.721,
      "step": 13649
    },
    {
      "epoch": 1.8693508627773214,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006698125389084292,
      "loss": 4.7302,
      "step": 13650
    },
    {
      "epoch": 1.8694878115584772,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006697767924149254,
      "loss": 4.8636,
      "step": 13651
    },
    {
      "epoch": 1.869624760339633,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006697410441000791,
      "loss": 4.7991,
      "step": 13652
    },
    {
      "epoch": 1.8697617091207888,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006697052939641863,
      "loss": 4.7713,
      "step": 13653
    },
    {
      "epoch": 1.8698986579019445,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006696695420075434,
      "loss": 4.7624,
      "step": 13654
    },
    {
      "epoch": 1.8700356066831005,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006696337882304467,
      "loss": 4.8058,
      "step": 13655
    },
    {
      "epoch": 1.8701725554642563,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006695980326331924,
      "loss": 4.7318,
      "step": 13656
    },
    {
      "epoch": 1.8703095042454123,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006695622752160769,
      "loss": 4.8495,
      "step": 13657
    },
    {
      "epoch": 1.870446453026568,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006695265159793965,
      "loss": 4.8279,
      "step": 13658
    },
    {
      "epoch": 1.8705834018077239,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006694907549234476,
      "loss": 4.8089,
      "step": 13659
    },
    {
      "epoch": 1.8707203505888796,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006694549920485263,
      "loss": 4.7888,
      "step": 13660
    },
    {
      "epoch": 1.8708572993700356,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006694192273549294,
      "loss": 4.8389,
      "step": 13661
    },
    {
      "epoch": 1.8709942481511914,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006693834608429531,
      "loss": 4.9103,
      "step": 13662
    },
    {
      "epoch": 1.8711311969323474,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006693476925128936,
      "loss": 4.7461,
      "step": 13663
    },
    {
      "epoch": 1.8712681457135032,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006693119223650477,
      "loss": 4.803,
      "step": 13664
    },
    {
      "epoch": 1.871405094494659,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006692761503997116,
      "loss": 4.7937,
      "step": 13665
    },
    {
      "epoch": 1.8715420432758147,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006692403766171818,
      "loss": 4.8264,
      "step": 13666
    },
    {
      "epoch": 1.8716789920569707,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006692046010177546,
      "loss": 4.8433,
      "step": 13667
    },
    {
      "epoch": 1.8718159408381265,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006691688236017268,
      "loss": 4.8798,
      "step": 13668
    },
    {
      "epoch": 1.8719528896192825,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006691330443693947,
      "loss": 4.7774,
      "step": 13669
    },
    {
      "epoch": 1.8720898384004383,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006690972633210547,
      "loss": 4.8356,
      "step": 13670
    },
    {
      "epoch": 1.872226787181594,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006690614804570036,
      "loss": 4.8284,
      "step": 13671
    },
    {
      "epoch": 1.8723637359627499,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006690256957775378,
      "loss": 4.8561,
      "step": 13672
    },
    {
      "epoch": 1.8725006847439056,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000668989909282954,
      "loss": 4.82,
      "step": 13673
    },
    {
      "epoch": 1.8726376335250616,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006689541209735484,
      "loss": 4.8113,
      "step": 13674
    },
    {
      "epoch": 1.8727745823062176,
      "grad_norm": 0.375,
      "learning_rate": 0.000668918330849618,
      "loss": 4.8251,
      "step": 13675
    },
    {
      "epoch": 1.8729115310873734,
      "grad_norm": 0.375,
      "learning_rate": 0.0006688825389114592,
      "loss": 4.7693,
      "step": 13676
    },
    {
      "epoch": 1.8730484798685292,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006688467451593685,
      "loss": 4.7799,
      "step": 13677
    },
    {
      "epoch": 1.873185428649685,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006688109495936429,
      "loss": 4.8095,
      "step": 13678
    },
    {
      "epoch": 1.8733223774308407,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006687751522145786,
      "loss": 4.756,
      "step": 13679
    },
    {
      "epoch": 1.8734593262119967,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006687393530224727,
      "loss": 4.7779,
      "step": 13680
    },
    {
      "epoch": 1.8735962749931525,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006687035520176215,
      "loss": 4.833,
      "step": 13681
    },
    {
      "epoch": 1.8737332237743085,
      "grad_norm": 0.359375,
      "learning_rate": 0.000668667749200322,
      "loss": 4.7754,
      "step": 13682
    },
    {
      "epoch": 1.8738701725554643,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006686319445708706,
      "loss": 4.6577,
      "step": 13683
    },
    {
      "epoch": 1.87400712133662,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006685961381295641,
      "loss": 4.7867,
      "step": 13684
    },
    {
      "epoch": 1.8741440701177758,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006685603298766995,
      "loss": 4.8598,
      "step": 13685
    },
    {
      "epoch": 1.8742810188989318,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006685245198125733,
      "loss": 4.7526,
      "step": 13686
    },
    {
      "epoch": 1.8744179676800876,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006684887079374825,
      "loss": 4.8236,
      "step": 13687
    },
    {
      "epoch": 1.8745549164612436,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006684528942517234,
      "loss": 4.7485,
      "step": 13688
    },
    {
      "epoch": 1.8746918652423994,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006684170787555933,
      "loss": 4.8063,
      "step": 13689
    },
    {
      "epoch": 1.8748288140235552,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006683812614493888,
      "loss": 4.8465,
      "step": 13690
    },
    {
      "epoch": 1.874965762804711,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006683454423334067,
      "loss": 4.7931,
      "step": 13691
    },
    {
      "epoch": 1.8751027115858667,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000668309621407944,
      "loss": 4.7895,
      "step": 13692
    },
    {
      "epoch": 1.8752396603670227,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006682737986732974,
      "loss": 4.7759,
      "step": 13693
    },
    {
      "epoch": 1.8753766091481787,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006682379741297637,
      "loss": 4.8299,
      "step": 13694
    },
    {
      "epoch": 1.8755135579293345,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006682021477776401,
      "loss": 4.8227,
      "step": 13695
    },
    {
      "epoch": 1.8756505067104903,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006681663196172232,
      "loss": 4.788,
      "step": 13696
    },
    {
      "epoch": 1.875787455491646,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006681304896488101,
      "loss": 4.726,
      "step": 13697
    },
    {
      "epoch": 1.8759244042728018,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006680946578726974,
      "loss": 4.8016,
      "step": 13698
    },
    {
      "epoch": 1.8760613530539578,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006680588242891826,
      "loss": 4.7545,
      "step": 13699
    },
    {
      "epoch": 1.8761983018351138,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006680229888985622,
      "loss": 4.8435,
      "step": 13700
    },
    {
      "epoch": 1.8763352506162696,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006679871517011334,
      "loss": 4.7341,
      "step": 13701
    },
    {
      "epoch": 1.8764721993974254,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006679513126971931,
      "loss": 4.8091,
      "step": 13702
    },
    {
      "epoch": 1.8766091481785812,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006679154718870384,
      "loss": 4.8234,
      "step": 13703
    },
    {
      "epoch": 1.876746096959737,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006678796292709662,
      "loss": 4.7684,
      "step": 13704
    },
    {
      "epoch": 1.876883045740893,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006678437848492736,
      "loss": 4.797,
      "step": 13705
    },
    {
      "epoch": 1.8770199945220487,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006678079386222577,
      "loss": 4.7365,
      "step": 13706
    },
    {
      "epoch": 1.8771569433032047,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006677720905902153,
      "loss": 4.8012,
      "step": 13707
    },
    {
      "epoch": 1.8772938920843605,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006677362407534439,
      "loss": 4.7665,
      "step": 13708
    },
    {
      "epoch": 1.8774308408655163,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006677003891122404,
      "loss": 4.6985,
      "step": 13709
    },
    {
      "epoch": 1.877567789646672,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006676645356669018,
      "loss": 4.7475,
      "step": 13710
    },
    {
      "epoch": 1.877704738427828,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006676286804177254,
      "loss": 4.8692,
      "step": 13711
    },
    {
      "epoch": 1.8778416872089838,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006675928233650082,
      "loss": 4.6951,
      "step": 13712
    },
    {
      "epoch": 1.8779786359901398,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006675569645090476,
      "loss": 4.6932,
      "step": 13713
    },
    {
      "epoch": 1.8781155847712956,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006675211038501403,
      "loss": 4.7695,
      "step": 13714
    },
    {
      "epoch": 1.8782525335524514,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000667485241388584,
      "loss": 4.7746,
      "step": 13715
    },
    {
      "epoch": 1.8783894823336071,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006674493771246756,
      "loss": 4.8743,
      "step": 13716
    },
    {
      "epoch": 1.878526431114763,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006674135110587124,
      "loss": 4.7451,
      "step": 13717
    },
    {
      "epoch": 1.878663379895919,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006673776431909917,
      "loss": 4.8642,
      "step": 13718
    },
    {
      "epoch": 1.878800328677075,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006673417735218105,
      "loss": 4.8105,
      "step": 13719
    },
    {
      "epoch": 1.8789372774582307,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006673059020514665,
      "loss": 4.8045,
      "step": 13720
    },
    {
      "epoch": 1.8790742262393865,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006672700287802565,
      "loss": 4.7601,
      "step": 13721
    },
    {
      "epoch": 1.8792111750205422,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006672341537084781,
      "loss": 4.7853,
      "step": 13722
    },
    {
      "epoch": 1.879348123801698,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006671982768364284,
      "loss": 4.8098,
      "step": 13723
    },
    {
      "epoch": 1.879485072582854,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006671623981644049,
      "loss": 4.7466,
      "step": 13724
    },
    {
      "epoch": 1.87962202136401,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006671265176927049,
      "loss": 4.7252,
      "step": 13725
    },
    {
      "epoch": 1.8797589701451658,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006670906354216256,
      "loss": 4.8049,
      "step": 13726
    },
    {
      "epoch": 1.8798959189263216,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006670547513514645,
      "loss": 4.7389,
      "step": 13727
    },
    {
      "epoch": 1.8800328677074774,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000667018865482519,
      "loss": 4.7574,
      "step": 13728
    },
    {
      "epoch": 1.8801698164886331,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006669829778150865,
      "loss": 4.7823,
      "step": 13729
    },
    {
      "epoch": 1.8803067652697891,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006669470883494642,
      "loss": 4.7961,
      "step": 13730
    },
    {
      "epoch": 1.880443714050945,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00066691119708595,
      "loss": 4.9009,
      "step": 13731
    },
    {
      "epoch": 1.880580662832101,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0006668753040248406,
      "loss": 4.8337,
      "step": 13732
    },
    {
      "epoch": 1.8807176116132567,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000666839409166434,
      "loss": 4.7489,
      "step": 13733
    },
    {
      "epoch": 1.8808545603944125,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006668035125110277,
      "loss": 4.7536,
      "step": 13734
    },
    {
      "epoch": 1.8809915091755682,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006667676140589189,
      "loss": 4.8172,
      "step": 13735
    },
    {
      "epoch": 1.8811284579567242,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006667317138104054,
      "loss": 4.7509,
      "step": 13736
    },
    {
      "epoch": 1.88126540673788,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006666958117657843,
      "loss": 4.7846,
      "step": 13737
    },
    {
      "epoch": 1.881402355519036,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006666599079253536,
      "loss": 4.8043,
      "step": 13738
    },
    {
      "epoch": 1.8815393043001918,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006666240022894105,
      "loss": 4.8436,
      "step": 13739
    },
    {
      "epoch": 1.8816762530813476,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006665880948582527,
      "loss": 4.7802,
      "step": 13740
    },
    {
      "epoch": 1.8818132018625033,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006665521856321778,
      "loss": 4.889,
      "step": 13741
    },
    {
      "epoch": 1.8819501506436591,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006665162746114831,
      "loss": 4.7906,
      "step": 13742
    },
    {
      "epoch": 1.8820870994248151,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006664803617964667,
      "loss": 4.6353,
      "step": 13743
    },
    {
      "epoch": 1.8822240482059711,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006664444471874257,
      "loss": 4.8542,
      "step": 13744
    },
    {
      "epoch": 1.8823609969871269,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006664085307846583,
      "loss": 4.7951,
      "step": 13745
    },
    {
      "epoch": 1.8824979457682827,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006663726125884617,
      "loss": 4.8087,
      "step": 13746
    },
    {
      "epoch": 1.8826348945494384,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006663366925991338,
      "loss": 4.7789,
      "step": 13747
    },
    {
      "epoch": 1.8827718433305942,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000666300770816972,
      "loss": 4.812,
      "step": 13748
    },
    {
      "epoch": 1.8829087921117502,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006662648472422742,
      "loss": 4.8373,
      "step": 13749
    },
    {
      "epoch": 1.8830457408929062,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006662289218753382,
      "loss": 4.7274,
      "step": 13750
    },
    {
      "epoch": 1.883182689674062,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006661929947164615,
      "loss": 4.7714,
      "step": 13751
    },
    {
      "epoch": 1.8833196384552178,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006661570657659419,
      "loss": 4.776,
      "step": 13752
    },
    {
      "epoch": 1.8834565872363735,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006661211350240772,
      "loss": 4.7274,
      "step": 13753
    },
    {
      "epoch": 1.8835935360175293,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006660852024911653,
      "loss": 4.7594,
      "step": 13754
    },
    {
      "epoch": 1.8837304847986853,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006660492681675037,
      "loss": 4.8389,
      "step": 13755
    },
    {
      "epoch": 1.883867433579841,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006660133320533904,
      "loss": 4.8035,
      "step": 13756
    },
    {
      "epoch": 1.884004382360997,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006659773941491231,
      "loss": 4.8066,
      "step": 13757
    },
    {
      "epoch": 1.8841413311421529,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006659414544549998,
      "loss": 4.785,
      "step": 13758
    },
    {
      "epoch": 1.8842782799233087,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006659055129713179,
      "loss": 4.7912,
      "step": 13759
    },
    {
      "epoch": 1.8844152287044644,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006658695696983757,
      "loss": 4.765,
      "step": 13760
    },
    {
      "epoch": 1.8845521774856204,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006658336246364711,
      "loss": 4.7018,
      "step": 13761
    },
    {
      "epoch": 1.8846891262667762,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006657976777859016,
      "loss": 4.8335,
      "step": 13762
    },
    {
      "epoch": 1.8848260750479322,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006657617291469656,
      "loss": 4.8737,
      "step": 13763
    },
    {
      "epoch": 1.884963023829088,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006657257787199605,
      "loss": 4.8407,
      "step": 13764
    },
    {
      "epoch": 1.8850999726102438,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006656898265051846,
      "loss": 4.7849,
      "step": 13765
    },
    {
      "epoch": 1.8852369213913995,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006656538725029356,
      "loss": 4.8235,
      "step": 13766
    },
    {
      "epoch": 1.8853738701725553,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006656179167135118,
      "loss": 4.7909,
      "step": 13767
    },
    {
      "epoch": 1.8855108189537113,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006655819591372107,
      "loss": 4.7566,
      "step": 13768
    },
    {
      "epoch": 1.8856477677348673,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006655459997743305,
      "loss": 4.8423,
      "step": 13769
    },
    {
      "epoch": 1.885784716516023,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006655100386251694,
      "loss": 4.751,
      "step": 13770
    },
    {
      "epoch": 1.8859216652971789,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006654740756900252,
      "loss": 4.7827,
      "step": 13771
    },
    {
      "epoch": 1.8860586140783346,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000665438110969196,
      "loss": 4.8326,
      "step": 13772
    },
    {
      "epoch": 1.8861955628594904,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006654021444629798,
      "loss": 4.7705,
      "step": 13773
    },
    {
      "epoch": 1.8863325116406464,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006653661761716748,
      "loss": 4.7975,
      "step": 13774
    },
    {
      "epoch": 1.8864694604218022,
      "grad_norm": 0.375,
      "learning_rate": 0.0006653302060955789,
      "loss": 4.8605,
      "step": 13775
    },
    {
      "epoch": 1.8866064092029582,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006652942342349903,
      "loss": 4.7828,
      "step": 13776
    },
    {
      "epoch": 1.886743357984114,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006652582605902071,
      "loss": 4.7598,
      "step": 13777
    },
    {
      "epoch": 1.8868803067652697,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006652222851615273,
      "loss": 4.799,
      "step": 13778
    },
    {
      "epoch": 1.8870172555464255,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006651863079492492,
      "loss": 4.8246,
      "step": 13779
    },
    {
      "epoch": 1.8871542043275815,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006651503289536709,
      "loss": 4.8024,
      "step": 13780
    },
    {
      "epoch": 1.8872911531087373,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006651143481750906,
      "loss": 4.7883,
      "step": 13781
    },
    {
      "epoch": 1.8874281018898933,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006650783656138063,
      "loss": 4.7904,
      "step": 13782
    },
    {
      "epoch": 1.887565050671049,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006650423812701165,
      "loss": 4.7689,
      "step": 13783
    },
    {
      "epoch": 1.8877019994522048,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006650063951443192,
      "loss": 4.7609,
      "step": 13784
    },
    {
      "epoch": 1.8878389482333606,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006649704072367126,
      "loss": 4.6642,
      "step": 13785
    },
    {
      "epoch": 1.8879758970145166,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006649344175475951,
      "loss": 4.7991,
      "step": 13786
    },
    {
      "epoch": 1.8881128457956724,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006648984260772649,
      "loss": 4.7187,
      "step": 13787
    },
    {
      "epoch": 1.8882497945768284,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00066486243282602,
      "loss": 4.82,
      "step": 13788
    },
    {
      "epoch": 1.8883867433579842,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000664826437794159,
      "loss": 4.7784,
      "step": 13789
    },
    {
      "epoch": 1.88852369213914,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006647904409819803,
      "loss": 4.7347,
      "step": 13790
    },
    {
      "epoch": 1.8886606409202957,
      "grad_norm": 0.375,
      "learning_rate": 0.000664754442389782,
      "loss": 4.7857,
      "step": 13791
    },
    {
      "epoch": 1.8887975897014515,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006647184420178623,
      "loss": 4.6903,
      "step": 13792
    },
    {
      "epoch": 1.8889345384826075,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006646824398665197,
      "loss": 4.819,
      "step": 13793
    },
    {
      "epoch": 1.8890714872637635,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006646464359360527,
      "loss": 4.7633,
      "step": 13794
    },
    {
      "epoch": 1.8892084360449193,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006646104302267594,
      "loss": 4.78,
      "step": 13795
    },
    {
      "epoch": 1.889345384826075,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006645744227389386,
      "loss": 4.8452,
      "step": 13796
    },
    {
      "epoch": 1.8894823336072308,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006645384134728882,
      "loss": 4.8552,
      "step": 13797
    },
    {
      "epoch": 1.8896192823883866,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006645024024289069,
      "loss": 4.8685,
      "step": 13798
    },
    {
      "epoch": 1.8897562311695426,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006644663896072931,
      "loss": 4.8282,
      "step": 13799
    },
    {
      "epoch": 1.8898931799506984,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006644303750083452,
      "loss": 4.884,
      "step": 13800
    },
    {
      "epoch": 1.8900301287318544,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006643943586323617,
      "loss": 4.8043,
      "step": 13801
    },
    {
      "epoch": 1.8901670775130102,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000664358340479641,
      "loss": 4.7819,
      "step": 13802
    },
    {
      "epoch": 1.890304026294166,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006643223205504817,
      "loss": 4.7048,
      "step": 13803
    },
    {
      "epoch": 1.8904409750753217,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006642862988451823,
      "loss": 4.8245,
      "step": 13804
    },
    {
      "epoch": 1.8905779238564777,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0006642502753640413,
      "loss": 4.7472,
      "step": 13805
    },
    {
      "epoch": 1.8907148726376335,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006642142501073571,
      "loss": 4.8119,
      "step": 13806
    },
    {
      "epoch": 1.8908518214187895,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006641782230754283,
      "loss": 4.856,
      "step": 13807
    },
    {
      "epoch": 1.8909887701999453,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006641421942685538,
      "loss": 4.8234,
      "step": 13808
    },
    {
      "epoch": 1.891125718981101,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006641061636870317,
      "loss": 4.8417,
      "step": 13809
    },
    {
      "epoch": 1.8912626677622568,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006640701313311609,
      "loss": 4.8498,
      "step": 13810
    },
    {
      "epoch": 1.8913996165434126,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006640340972012398,
      "loss": 4.7654,
      "step": 13811
    },
    {
      "epoch": 1.8915365653245686,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006639980612975672,
      "loss": 4.8714,
      "step": 13812
    },
    {
      "epoch": 1.8916735141057246,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006639620236204417,
      "loss": 4.8049,
      "step": 13813
    },
    {
      "epoch": 1.8918104628868804,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006639259841701619,
      "loss": 4.8226,
      "step": 13814
    },
    {
      "epoch": 1.8919474116680361,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006638899429470265,
      "loss": 4.7718,
      "step": 13815
    },
    {
      "epoch": 1.892084360449192,
      "grad_norm": 0.375,
      "learning_rate": 0.0006638538999513341,
      "loss": 4.8419,
      "step": 13816
    },
    {
      "epoch": 1.8922213092303477,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006638178551833837,
      "loss": 4.792,
      "step": 13817
    },
    {
      "epoch": 1.8923582580115037,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006637818086434735,
      "loss": 4.8035,
      "step": 13818
    },
    {
      "epoch": 1.8924952067926597,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006637457603319028,
      "loss": 4.8049,
      "step": 13819
    },
    {
      "epoch": 1.8926321555738155,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006637097102489698,
      "loss": 4.6988,
      "step": 13820
    },
    {
      "epoch": 1.8927691043549713,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006636736583949737,
      "loss": 4.7268,
      "step": 13821
    },
    {
      "epoch": 1.892906053136127,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006636376047702129,
      "loss": 4.7932,
      "step": 13822
    },
    {
      "epoch": 1.8930430019172828,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006636015493749865,
      "loss": 4.7567,
      "step": 13823
    },
    {
      "epoch": 1.8931799506984388,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000663565492209593,
      "loss": 4.7711,
      "step": 13824
    },
    {
      "epoch": 1.8933168994795946,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006635294332743314,
      "loss": 4.7233,
      "step": 13825
    },
    {
      "epoch": 1.8934538482607506,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006634933725695007,
      "loss": 4.7708,
      "step": 13826
    },
    {
      "epoch": 1.8935907970419064,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006634573100953993,
      "loss": 4.7933,
      "step": 13827
    },
    {
      "epoch": 1.8937277458230621,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006634212458523265,
      "loss": 4.74,
      "step": 13828
    },
    {
      "epoch": 1.893864694604218,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000663385179840581,
      "loss": 4.8148,
      "step": 13829
    },
    {
      "epoch": 1.894001643385374,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006633491120604616,
      "loss": 4.833,
      "step": 13830
    },
    {
      "epoch": 1.8941385921665297,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006633130425122672,
      "loss": 4.8534,
      "step": 13831
    },
    {
      "epoch": 1.8942755409476857,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006632769711962969,
      "loss": 4.7643,
      "step": 13832
    },
    {
      "epoch": 1.8944124897288415,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006632408981128493,
      "loss": 4.7704,
      "step": 13833
    },
    {
      "epoch": 1.8945494385099972,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006632048232622238,
      "loss": 4.7799,
      "step": 13834
    },
    {
      "epoch": 1.894686387291153,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000663168746644719,
      "loss": 4.6954,
      "step": 13835
    },
    {
      "epoch": 1.8948233360723088,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006631326682606341,
      "loss": 4.7938,
      "step": 13836
    },
    {
      "epoch": 1.8949602848534648,
      "grad_norm": 0.416015625,
      "learning_rate": 0.000663096588110268,
      "loss": 4.8349,
      "step": 13837
    },
    {
      "epoch": 1.8950972336346208,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006630605061939196,
      "loss": 4.7485,
      "step": 13838
    },
    {
      "epoch": 1.8952341824157766,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0006630244225118881,
      "loss": 4.843,
      "step": 13839
    },
    {
      "epoch": 1.8953711311969323,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006629883370644724,
      "loss": 4.7495,
      "step": 13840
    },
    {
      "epoch": 1.8955080799780881,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006629522498519716,
      "loss": 4.8215,
      "step": 13841
    },
    {
      "epoch": 1.895645028759244,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006629161608746847,
      "loss": 4.7655,
      "step": 13842
    },
    {
      "epoch": 1.8957819775404,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000662880070132911,
      "loss": 4.869,
      "step": 13843
    },
    {
      "epoch": 1.895918926321556,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006628439776269493,
      "loss": 4.8085,
      "step": 13844
    },
    {
      "epoch": 1.8960558751027117,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006628078833570988,
      "loss": 4.7226,
      "step": 13845
    },
    {
      "epoch": 1.8961928238838675,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006627717873236589,
      "loss": 4.8303,
      "step": 13846
    },
    {
      "epoch": 1.8963297726650232,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006627356895269282,
      "loss": 4.8632,
      "step": 13847
    },
    {
      "epoch": 1.896466721446179,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006626995899672064,
      "loss": 4.773,
      "step": 13848
    },
    {
      "epoch": 1.896603670227335,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006626634886447922,
      "loss": 4.8433,
      "step": 13849
    },
    {
      "epoch": 1.8967406190084908,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006626273855599852,
      "loss": 4.7854,
      "step": 13850
    },
    {
      "epoch": 1.8968775677896468,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006625912807130842,
      "loss": 4.7903,
      "step": 13851
    },
    {
      "epoch": 1.8970145165708026,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006625551741043887,
      "loss": 4.7828,
      "step": 13852
    },
    {
      "epoch": 1.8971514653519583,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0006625190657341978,
      "loss": 4.8256,
      "step": 13853
    },
    {
      "epoch": 1.897288414133114,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006624829556028108,
      "loss": 4.7734,
      "step": 13854
    },
    {
      "epoch": 1.89742536291427,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0006624468437105269,
      "loss": 4.7054,
      "step": 13855
    },
    {
      "epoch": 1.8975623116954259,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006624107300576453,
      "loss": 4.7714,
      "step": 13856
    },
    {
      "epoch": 1.8976992604765819,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006623746146444655,
      "loss": 4.8021,
      "step": 13857
    },
    {
      "epoch": 1.8978362092577377,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006623384974712865,
      "loss": 4.7914,
      "step": 13858
    },
    {
      "epoch": 1.8979731580388934,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006623023785384079,
      "loss": 4.6992,
      "step": 13859
    },
    {
      "epoch": 1.8981101068200492,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006622662578461289,
      "loss": 4.7467,
      "step": 13860
    },
    {
      "epoch": 1.898247055601205,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006622301353947489,
      "loss": 4.7663,
      "step": 13861
    },
    {
      "epoch": 1.898384004382361,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000662194011184567,
      "loss": 4.8235,
      "step": 13862
    },
    {
      "epoch": 1.898520953163517,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006621578852158828,
      "loss": 4.7179,
      "step": 13863
    },
    {
      "epoch": 1.8986579019446728,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006621217574889958,
      "loss": 4.7332,
      "step": 13864
    },
    {
      "epoch": 1.8987948507258285,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006620856280042051,
      "loss": 4.8144,
      "step": 13865
    },
    {
      "epoch": 1.8989317995069843,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006620494967618103,
      "loss": 4.7979,
      "step": 13866
    },
    {
      "epoch": 1.89906874828814,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006620133637621109,
      "loss": 4.8149,
      "step": 13867
    },
    {
      "epoch": 1.899205697069296,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006619772290054062,
      "loss": 4.73,
      "step": 13868
    },
    {
      "epoch": 1.899342645850452,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006619410924919956,
      "loss": 4.8348,
      "step": 13869
    },
    {
      "epoch": 1.8994795946316079,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006619049542221788,
      "loss": 4.8729,
      "step": 13870
    },
    {
      "epoch": 1.8996165434127636,
      "grad_norm": 0.390625,
      "learning_rate": 0.000661868814196255,
      "loss": 4.8388,
      "step": 13871
    },
    {
      "epoch": 1.8997534921939194,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006618326724145239,
      "loss": 4.7158,
      "step": 13872
    },
    {
      "epoch": 1.8998904409750752,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006617965288772849,
      "loss": 4.8201,
      "step": 13873
    },
    {
      "epoch": 1.9000273897562312,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006617603835848378,
      "loss": 4.7137,
      "step": 13874
    },
    {
      "epoch": 1.900164338537387,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006617242365374818,
      "loss": 4.8209,
      "step": 13875
    },
    {
      "epoch": 1.900301287318543,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006616880877355167,
      "loss": 4.8013,
      "step": 13876
    },
    {
      "epoch": 1.9004382360996988,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006616519371792418,
      "loss": 4.8517,
      "step": 13877
    },
    {
      "epoch": 1.9005751848808545,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006616157848689571,
      "loss": 4.7667,
      "step": 13878
    },
    {
      "epoch": 1.9007121336620103,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006615796308049618,
      "loss": 4.7924,
      "step": 13879
    },
    {
      "epoch": 1.9008490824431663,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006615434749875556,
      "loss": 4.7563,
      "step": 13880
    },
    {
      "epoch": 1.900986031224322,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006615073174170384,
      "loss": 4.8266,
      "step": 13881
    },
    {
      "epoch": 1.901122980005478,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006614711580937096,
      "loss": 4.7461,
      "step": 13882
    },
    {
      "epoch": 1.9012599287866339,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006614349970178688,
      "loss": 4.8077,
      "step": 13883
    },
    {
      "epoch": 1.9013968775677896,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006613988341898161,
      "loss": 4.7224,
      "step": 13884
    },
    {
      "epoch": 1.9015338263489454,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006613626696098505,
      "loss": 4.8303,
      "step": 13885
    },
    {
      "epoch": 1.9016707751301012,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006613265032782724,
      "loss": 4.6835,
      "step": 13886
    },
    {
      "epoch": 1.9018077239112572,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000661290335195381,
      "loss": 4.7689,
      "step": 13887
    },
    {
      "epoch": 1.9019446726924132,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006612541653614764,
      "loss": 4.7838,
      "step": 13888
    },
    {
      "epoch": 1.902081621473569,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006612179937768581,
      "loss": 4.7434,
      "step": 13889
    },
    {
      "epoch": 1.9022185702547247,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006611818204418261,
      "loss": 4.767,
      "step": 13890
    },
    {
      "epoch": 1.9023555190358805,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006611456453566798,
      "loss": 4.7355,
      "step": 13891
    },
    {
      "epoch": 1.9024924678170363,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006611094685217194,
      "loss": 4.8098,
      "step": 13892
    },
    {
      "epoch": 1.9026294165981923,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006610732899372444,
      "loss": 4.7618,
      "step": 13893
    },
    {
      "epoch": 1.9027663653793483,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006610371096035549,
      "loss": 4.7756,
      "step": 13894
    },
    {
      "epoch": 1.902903314160504,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006610009275209505,
      "loss": 4.809,
      "step": 13895
    },
    {
      "epoch": 1.9030402629416598,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006609647436897311,
      "loss": 4.7803,
      "step": 13896
    },
    {
      "epoch": 1.9031772117228156,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006609285581101967,
      "loss": 4.7292,
      "step": 13897
    },
    {
      "epoch": 1.9033141605039714,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006608923707826469,
      "loss": 4.7681,
      "step": 13898
    },
    {
      "epoch": 1.9034511092851274,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000660856181707382,
      "loss": 4.765,
      "step": 13899
    },
    {
      "epoch": 1.9035880580662832,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006608199908847015,
      "loss": 4.7555,
      "step": 13900
    },
    {
      "epoch": 1.9037250068474392,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006607837983149056,
      "loss": 4.8415,
      "step": 13901
    },
    {
      "epoch": 1.903861955628595,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006607476039982941,
      "loss": 4.8482,
      "step": 13902
    },
    {
      "epoch": 1.9039989044097507,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006607114079351669,
      "loss": 4.7967,
      "step": 13903
    },
    {
      "epoch": 1.9041358531909065,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006606752101258242,
      "loss": 4.7701,
      "step": 13904
    },
    {
      "epoch": 1.9042728019720625,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006606390105705659,
      "loss": 4.8862,
      "step": 13905
    },
    {
      "epoch": 1.9044097507532183,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006606028092696918,
      "loss": 4.8231,
      "step": 13906
    },
    {
      "epoch": 1.9045466995343743,
      "grad_norm": 0.390625,
      "learning_rate": 0.000660566606223502,
      "loss": 4.7748,
      "step": 13907
    },
    {
      "epoch": 1.90468364831553,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006605304014322966,
      "loss": 4.7013,
      "step": 13908
    },
    {
      "epoch": 1.9048205970966858,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006604941948963755,
      "loss": 4.7412,
      "step": 13909
    },
    {
      "epoch": 1.9049575458778416,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006604579866160389,
      "loss": 4.8192,
      "step": 13910
    },
    {
      "epoch": 1.9050944946589974,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000660421776591587,
      "loss": 4.8118,
      "step": 13911
    },
    {
      "epoch": 1.9052314434401534,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006603855648233195,
      "loss": 4.782,
      "step": 13912
    },
    {
      "epoch": 1.9053683922213094,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0006603493513115368,
      "loss": 4.8232,
      "step": 13913
    },
    {
      "epoch": 1.9055053410024652,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006603131360565387,
      "loss": 4.8541,
      "step": 13914
    },
    {
      "epoch": 1.905642289783621,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0006602769190586259,
      "loss": 4.8168,
      "step": 13915
    },
    {
      "epoch": 1.9057792385647767,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006602407003180979,
      "loss": 4.8678,
      "step": 13916
    },
    {
      "epoch": 1.9059161873459325,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000660204479835255,
      "loss": 4.7405,
      "step": 13917
    },
    {
      "epoch": 1.9060531361270885,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006601682576103978,
      "loss": 4.7824,
      "step": 13918
    },
    {
      "epoch": 1.9061900849082443,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006601320336438259,
      "loss": 4.8078,
      "step": 13919
    },
    {
      "epoch": 1.9063270336894003,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006600958079358398,
      "loss": 4.7527,
      "step": 13920
    },
    {
      "epoch": 1.906463982470556,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006600595804867399,
      "loss": 4.7123,
      "step": 13921
    },
    {
      "epoch": 1.9066009312517118,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000660023351296826,
      "loss": 4.726,
      "step": 13922
    },
    {
      "epoch": 1.9067378800328676,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006599871203663986,
      "loss": 4.7704,
      "step": 13923
    },
    {
      "epoch": 1.9068748288140236,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006599508876957579,
      "loss": 4.7955,
      "step": 13924
    },
    {
      "epoch": 1.9070117775951794,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006599146532852042,
      "loss": 4.8058,
      "step": 13925
    },
    {
      "epoch": 1.9071487263763354,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006598784171350377,
      "loss": 4.8626,
      "step": 13926
    },
    {
      "epoch": 1.9072856751574911,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006598421792455588,
      "loss": 4.6825,
      "step": 13927
    },
    {
      "epoch": 1.907422623938647,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006598059396170676,
      "loss": 4.7672,
      "step": 13928
    },
    {
      "epoch": 1.9075595727198027,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006597696982498649,
      "loss": 4.8113,
      "step": 13929
    },
    {
      "epoch": 1.9076965215009585,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006597334551442503,
      "loss": 4.7687,
      "step": 13930
    },
    {
      "epoch": 1.9078334702821145,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006596972103005248,
      "loss": 4.7714,
      "step": 13931
    },
    {
      "epoch": 1.9079704190632705,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006596609637189886,
      "loss": 4.8012,
      "step": 13932
    },
    {
      "epoch": 1.9081073678444262,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000659624715399942,
      "loss": 4.7644,
      "step": 13933
    },
    {
      "epoch": 1.908244316625582,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006595884653436855,
      "loss": 4.8184,
      "step": 13934
    },
    {
      "epoch": 1.9083812654067378,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006595522135505193,
      "loss": 4.7629,
      "step": 13935
    },
    {
      "epoch": 1.9085182141878936,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006595159600207441,
      "loss": 4.7616,
      "step": 13936
    },
    {
      "epoch": 1.9086551629690496,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00065947970475466,
      "loss": 4.7764,
      "step": 13937
    },
    {
      "epoch": 1.9087921117502056,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006594434477525679,
      "loss": 4.7493,
      "step": 13938
    },
    {
      "epoch": 1.9089290605313614,
      "grad_norm": 0.375,
      "learning_rate": 0.000659407189014768,
      "loss": 4.8565,
      "step": 13939
    },
    {
      "epoch": 1.9090660093125171,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006593709285415607,
      "loss": 4.7773,
      "step": 13940
    },
    {
      "epoch": 1.909202958093673,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006593346663332469,
      "loss": 4.8328,
      "step": 13941
    },
    {
      "epoch": 1.9093399068748287,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006592984023901265,
      "loss": 4.7916,
      "step": 13942
    },
    {
      "epoch": 1.9094768556559847,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006592621367125006,
      "loss": 4.7381,
      "step": 13943
    },
    {
      "epoch": 1.9096138044371405,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006592258693006694,
      "loss": 4.757,
      "step": 13944
    },
    {
      "epoch": 1.9097507532182965,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006591896001549336,
      "loss": 4.8021,
      "step": 13945
    },
    {
      "epoch": 1.9098877019994522,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006591533292755937,
      "loss": 4.8162,
      "step": 13946
    },
    {
      "epoch": 1.910024650780608,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006591170566629503,
      "loss": 4.749,
      "step": 13947
    },
    {
      "epoch": 1.9101615995617638,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006590807823173042,
      "loss": 4.9106,
      "step": 13948
    },
    {
      "epoch": 1.9102985483429198,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006590445062389556,
      "loss": 4.8056,
      "step": 13949
    },
    {
      "epoch": 1.9104354971240756,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006590082284282056,
      "loss": 4.8092,
      "step": 13950
    },
    {
      "epoch": 1.9105724459052316,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006589719488853544,
      "loss": 4.8516,
      "step": 13951
    },
    {
      "epoch": 1.9107093946863873,
      "grad_norm": 0.375,
      "learning_rate": 0.0006589356676107029,
      "loss": 4.7003,
      "step": 13952
    },
    {
      "epoch": 1.9108463434675431,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006588993846045518,
      "loss": 4.8222,
      "step": 13953
    },
    {
      "epoch": 1.910983292248699,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006588630998672017,
      "loss": 4.7648,
      "step": 13954
    },
    {
      "epoch": 1.9111202410298547,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006588268133989533,
      "loss": 4.8201,
      "step": 13955
    },
    {
      "epoch": 1.9112571898110107,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006587905252001074,
      "loss": 4.7986,
      "step": 13956
    },
    {
      "epoch": 1.9113941385921667,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006587542352709647,
      "loss": 4.7959,
      "step": 13957
    },
    {
      "epoch": 1.9115310873733224,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006587179436118258,
      "loss": 4.6956,
      "step": 13958
    },
    {
      "epoch": 1.9116680361544782,
      "grad_norm": 0.375,
      "learning_rate": 0.0006586816502229915,
      "loss": 4.7531,
      "step": 13959
    },
    {
      "epoch": 1.911804984935634,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006586453551047629,
      "loss": 4.7855,
      "step": 13960
    },
    {
      "epoch": 1.9119419337167898,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006586090582574403,
      "loss": 4.7827,
      "step": 13961
    },
    {
      "epoch": 1.9120788824979458,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006585727596813249,
      "loss": 4.8044,
      "step": 13962
    },
    {
      "epoch": 1.9122158312791018,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006585364593767172,
      "loss": 4.7697,
      "step": 13963
    },
    {
      "epoch": 1.9123527800602576,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006585001573439181,
      "loss": 4.7756,
      "step": 13964
    },
    {
      "epoch": 1.9124897288414133,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006584638535832287,
      "loss": 4.8425,
      "step": 13965
    },
    {
      "epoch": 1.912626677622569,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006584275480949495,
      "loss": 4.8243,
      "step": 13966
    },
    {
      "epoch": 1.9127636264037249,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006583912408793816,
      "loss": 4.7626,
      "step": 13967
    },
    {
      "epoch": 1.9129005751848809,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006583549319368259,
      "loss": 4.8307,
      "step": 13968
    },
    {
      "epoch": 1.9130375239660367,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006583186212675831,
      "loss": 4.8216,
      "step": 13969
    },
    {
      "epoch": 1.9131744727471927,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006582823088719544,
      "loss": 4.8013,
      "step": 13970
    },
    {
      "epoch": 1.9133114215283484,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006582459947502405,
      "loss": 4.8113,
      "step": 13971
    },
    {
      "epoch": 1.9134483703095042,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006582096789027424,
      "loss": 4.7416,
      "step": 13972
    },
    {
      "epoch": 1.91358531909066,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006581733613297612,
      "loss": 4.7602,
      "step": 13973
    },
    {
      "epoch": 1.913722267871816,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006581370420315976,
      "loss": 4.7816,
      "step": 13974
    },
    {
      "epoch": 1.9138592166529718,
      "grad_norm": 0.375,
      "learning_rate": 0.0006581007210085529,
      "loss": 4.715,
      "step": 13975
    },
    {
      "epoch": 1.9139961654341278,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006580643982609278,
      "loss": 4.811,
      "step": 13976
    },
    {
      "epoch": 1.9141331142152835,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006580280737890233,
      "loss": 4.7486,
      "step": 13977
    },
    {
      "epoch": 1.9142700629964393,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000657991747593141,
      "loss": 4.8025,
      "step": 13978
    },
    {
      "epoch": 1.914407011777595,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006579554196735811,
      "loss": 4.7784,
      "step": 13979
    },
    {
      "epoch": 1.9145439605587509,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006579190900306455,
      "loss": 4.7589,
      "step": 13980
    },
    {
      "epoch": 1.9146809093399069,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006578827586646344,
      "loss": 4.826,
      "step": 13981
    },
    {
      "epoch": 1.9148178581210629,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006578464255758496,
      "loss": 4.7561,
      "step": 13982
    },
    {
      "epoch": 1.9149548069022186,
      "grad_norm": 0.359375,
      "learning_rate": 0.000657810090764592,
      "loss": 4.7525,
      "step": 13983
    },
    {
      "epoch": 1.9150917556833744,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006577737542311626,
      "loss": 4.7949,
      "step": 13984
    },
    {
      "epoch": 1.9152287044645302,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006577374159758625,
      "loss": 4.7681,
      "step": 13985
    },
    {
      "epoch": 1.915365653245686,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000657701075998993,
      "loss": 4.7849,
      "step": 13986
    },
    {
      "epoch": 1.915502602026842,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006576647343008552,
      "loss": 4.8325,
      "step": 13987
    },
    {
      "epoch": 1.915639550807998,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006576283908817502,
      "loss": 4.7871,
      "step": 13988
    },
    {
      "epoch": 1.9157764995891537,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006575920457419795,
      "loss": 4.8101,
      "step": 13989
    },
    {
      "epoch": 1.9159134483703095,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006575556988818437,
      "loss": 4.688,
      "step": 13990
    },
    {
      "epoch": 1.9160503971514653,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006575193503016446,
      "loss": 4.7974,
      "step": 13991
    },
    {
      "epoch": 1.916187345932621,
      "grad_norm": 0.328125,
      "learning_rate": 0.0006574830000016831,
      "loss": 4.8498,
      "step": 13992
    },
    {
      "epoch": 1.916324294713777,
      "grad_norm": 0.375,
      "learning_rate": 0.0006574466479822605,
      "loss": 4.7191,
      "step": 13993
    },
    {
      "epoch": 1.9164612434949329,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006574102942436781,
      "loss": 4.7657,
      "step": 13994
    },
    {
      "epoch": 1.9165981922760889,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006573739387862373,
      "loss": 4.844,
      "step": 13995
    },
    {
      "epoch": 1.9167351410572446,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006573375816102392,
      "loss": 4.8283,
      "step": 13996
    },
    {
      "epoch": 1.9168720898384004,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006573012227159851,
      "loss": 4.7592,
      "step": 13997
    },
    {
      "epoch": 1.9170090386195562,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006572648621037765,
      "loss": 4.7974,
      "step": 13998
    },
    {
      "epoch": 1.9171459874007122,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006572284997739145,
      "loss": 4.8252,
      "step": 13999
    },
    {
      "epoch": 1.917282936181868,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006571921357267005,
      "loss": 4.7065,
      "step": 14000
    },
    {
      "epoch": 1.917419884963024,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006571557699624359,
      "loss": 4.9292,
      "step": 14001
    },
    {
      "epoch": 1.9175568337441797,
      "grad_norm": 0.375,
      "learning_rate": 0.0006571194024814222,
      "loss": 4.7953,
      "step": 14002
    },
    {
      "epoch": 1.9176937825253355,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006570830332839605,
      "loss": 4.7377,
      "step": 14003
    },
    {
      "epoch": 1.9178307313064913,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006570466623703524,
      "loss": 4.7239,
      "step": 14004
    },
    {
      "epoch": 1.917967680087647,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006570102897408993,
      "loss": 4.8057,
      "step": 14005
    },
    {
      "epoch": 1.918104628868803,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006569739153959026,
      "loss": 4.7688,
      "step": 14006
    },
    {
      "epoch": 1.918241577649959,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006569375393356638,
      "loss": 4.7548,
      "step": 14007
    },
    {
      "epoch": 1.9183785264311148,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006569011615604842,
      "loss": 4.7947,
      "step": 14008
    },
    {
      "epoch": 1.9185154752122706,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006568647820706654,
      "loss": 4.8076,
      "step": 14009
    },
    {
      "epoch": 1.9186524239934264,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006568284008665089,
      "loss": 4.6697,
      "step": 14010
    },
    {
      "epoch": 1.9187893727745822,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006567920179483161,
      "loss": 4.7524,
      "step": 14011
    },
    {
      "epoch": 1.9189263215557382,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006567556333163886,
      "loss": 4.6608,
      "step": 14012
    },
    {
      "epoch": 1.9190632703368942,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000656719246971028,
      "loss": 4.8042,
      "step": 14013
    },
    {
      "epoch": 1.91920021911805,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006566828589125356,
      "loss": 4.7876,
      "step": 14014
    },
    {
      "epoch": 1.9193371678992057,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006566464691412132,
      "loss": 4.7434,
      "step": 14015
    },
    {
      "epoch": 1.9194741166803615,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006566100776573622,
      "loss": 4.739,
      "step": 14016
    },
    {
      "epoch": 1.9196110654615173,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006565736844612842,
      "loss": 4.7775,
      "step": 14017
    },
    {
      "epoch": 1.9197480142426733,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000656537289553281,
      "loss": 4.8007,
      "step": 14018
    },
    {
      "epoch": 1.919884963023829,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000656500892933654,
      "loss": 4.7905,
      "step": 14019
    },
    {
      "epoch": 1.920021911804985,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006564644946027049,
      "loss": 4.7868,
      "step": 14020
    },
    {
      "epoch": 1.9201588605861408,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006564280945607352,
      "loss": 4.7583,
      "step": 14021
    },
    {
      "epoch": 1.9202958093672966,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006563916928080468,
      "loss": 4.7925,
      "step": 14022
    },
    {
      "epoch": 1.9204327581484524,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006563552893449413,
      "loss": 4.8287,
      "step": 14023
    },
    {
      "epoch": 1.9205697069296084,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0006563188841717202,
      "loss": 4.7346,
      "step": 14024
    },
    {
      "epoch": 1.9207066557107642,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006562824772886857,
      "loss": 4.8655,
      "step": 14025
    },
    {
      "epoch": 1.9208436044919202,
      "grad_norm": 0.46875,
      "learning_rate": 0.0006562460686961388,
      "loss": 4.7929,
      "step": 14026
    },
    {
      "epoch": 1.920980553273076,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006562096583943817,
      "loss": 4.7627,
      "step": 14027
    },
    {
      "epoch": 1.9211175020542317,
      "grad_norm": 0.470703125,
      "learning_rate": 0.000656173246383716,
      "loss": 4.7516,
      "step": 14028
    },
    {
      "epoch": 1.9212544508353875,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006561368326644435,
      "loss": 4.8066,
      "step": 14029
    },
    {
      "epoch": 1.9213913996165433,
      "grad_norm": 0.515625,
      "learning_rate": 0.0006561004172368659,
      "loss": 4.7499,
      "step": 14030
    },
    {
      "epoch": 1.9215283483976993,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006560640001012851,
      "loss": 4.7533,
      "step": 14031
    },
    {
      "epoch": 1.9216652971788553,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0006560275812580028,
      "loss": 4.7014,
      "step": 14032
    },
    {
      "epoch": 1.921802245960011,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006559911607073208,
      "loss": 4.7832,
      "step": 14033
    },
    {
      "epoch": 1.9219391947411668,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006559547384495411,
      "loss": 4.7773,
      "step": 14034
    },
    {
      "epoch": 1.9220761435223226,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006559183144849653,
      "loss": 4.771,
      "step": 14035
    },
    {
      "epoch": 1.9222130923034784,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006558818888138954,
      "loss": 4.7472,
      "step": 14036
    },
    {
      "epoch": 1.9223500410846344,
      "grad_norm": 0.4375,
      "learning_rate": 0.0006558454614366332,
      "loss": 4.796,
      "step": 14037
    },
    {
      "epoch": 1.9224869898657901,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006558090323534806,
      "loss": 4.7499,
      "step": 14038
    },
    {
      "epoch": 1.9226239386469461,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006557726015647395,
      "loss": 4.7407,
      "step": 14039
    },
    {
      "epoch": 1.922760887428102,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006557361690707119,
      "loss": 4.7244,
      "step": 14040
    },
    {
      "epoch": 1.9228978362092577,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006556997348716996,
      "loss": 4.7876,
      "step": 14041
    },
    {
      "epoch": 1.9230347849904135,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006556632989680046,
      "loss": 4.8027,
      "step": 14042
    },
    {
      "epoch": 1.9231717337715695,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006556268613599288,
      "loss": 4.7753,
      "step": 14043
    },
    {
      "epoch": 1.9233086825527252,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006555904220477742,
      "loss": 4.7628,
      "step": 14044
    },
    {
      "epoch": 1.9234456313338812,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006555539810318427,
      "loss": 4.8211,
      "step": 14045
    },
    {
      "epoch": 1.923582580115037,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0006555175383124365,
      "loss": 4.8206,
      "step": 14046
    },
    {
      "epoch": 1.9237195288961928,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0006554810938898574,
      "loss": 4.7522,
      "step": 14047
    },
    {
      "epoch": 1.9238564776773486,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006554446477644077,
      "loss": 4.7754,
      "step": 14048
    },
    {
      "epoch": 1.9239934264585046,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000655408199936389,
      "loss": 4.6731,
      "step": 14049
    },
    {
      "epoch": 1.9241303752396604,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006553717504061038,
      "loss": 4.7225,
      "step": 14050
    },
    {
      "epoch": 1.9242673240208163,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006553352991738538,
      "loss": 4.7454,
      "step": 14051
    },
    {
      "epoch": 1.9244042728019721,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006552988462399414,
      "loss": 4.8291,
      "step": 14052
    },
    {
      "epoch": 1.924541221583128,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006552623916046685,
      "loss": 4.7677,
      "step": 14053
    },
    {
      "epoch": 1.9246781703642837,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006552259352683372,
      "loss": 4.7971,
      "step": 14054
    },
    {
      "epoch": 1.9248151191454395,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006551894772312496,
      "loss": 4.6791,
      "step": 14055
    },
    {
      "epoch": 1.9249520679265955,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006551530174937081,
      "loss": 4.8336,
      "step": 14056
    },
    {
      "epoch": 1.9250890167077515,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006551165560560146,
      "loss": 4.7917,
      "step": 14057
    },
    {
      "epoch": 1.9252259654889072,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006550800929184713,
      "loss": 4.8347,
      "step": 14058
    },
    {
      "epoch": 1.925362914270063,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006550436280813802,
      "loss": 4.7555,
      "step": 14059
    },
    {
      "epoch": 1.9254998630512188,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006550071615450439,
      "loss": 4.8116,
      "step": 14060
    },
    {
      "epoch": 1.9256368118323746,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006549706933097645,
      "loss": 4.7989,
      "step": 14061
    },
    {
      "epoch": 1.9257737606135306,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006549342233758439,
      "loss": 4.7197,
      "step": 14062
    },
    {
      "epoch": 1.9259107093946863,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006548977517435848,
      "loss": 4.7449,
      "step": 14063
    },
    {
      "epoch": 1.9260476581758423,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006548612784132889,
      "loss": 4.8084,
      "step": 14064
    },
    {
      "epoch": 1.9261846069569981,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006548248033852591,
      "loss": 4.773,
      "step": 14065
    },
    {
      "epoch": 1.926321555738154,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006547883266597971,
      "loss": 4.7086,
      "step": 14066
    },
    {
      "epoch": 1.9264585045193097,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006547518482372056,
      "loss": 4.744,
      "step": 14067
    },
    {
      "epoch": 1.9265954533004657,
      "grad_norm": 0.375,
      "learning_rate": 0.0006547153681177866,
      "loss": 4.7347,
      "step": 14068
    },
    {
      "epoch": 1.9267324020816214,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006546788863018427,
      "loss": 4.8549,
      "step": 14069
    },
    {
      "epoch": 1.9268693508627774,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006546424027896761,
      "loss": 4.7379,
      "step": 14070
    },
    {
      "epoch": 1.9270062996439332,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000654605917581589,
      "loss": 4.7415,
      "step": 14071
    },
    {
      "epoch": 1.927143248425089,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000654569430677884,
      "loss": 4.8109,
      "step": 14072
    },
    {
      "epoch": 1.9272801972062448,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006545329420788633,
      "loss": 4.7104,
      "step": 14073
    },
    {
      "epoch": 1.9274171459874005,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006544964517848295,
      "loss": 4.8325,
      "step": 14074
    },
    {
      "epoch": 1.9275540947685565,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006544599597960848,
      "loss": 4.733,
      "step": 14075
    },
    {
      "epoch": 1.9276910435497125,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006544234661129317,
      "loss": 4.783,
      "step": 14076
    },
    {
      "epoch": 1.9278279923308683,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006543869707356726,
      "loss": 4.7262,
      "step": 14077
    },
    {
      "epoch": 1.927964941112024,
      "grad_norm": 0.390625,
      "learning_rate": 0.00065435047366461,
      "loss": 4.7044,
      "step": 14078
    },
    {
      "epoch": 1.9281018898931799,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006543139749000462,
      "loss": 4.7064,
      "step": 14079
    },
    {
      "epoch": 1.9282388386743357,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000654277474442284,
      "loss": 4.8375,
      "step": 14080
    },
    {
      "epoch": 1.9283757874554917,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006542409722916257,
      "loss": 4.7572,
      "step": 14081
    },
    {
      "epoch": 1.9285127362366477,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006542044684483737,
      "loss": 4.7205,
      "step": 14082
    },
    {
      "epoch": 1.9286496850178034,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006541679629128306,
      "loss": 4.8131,
      "step": 14083
    },
    {
      "epoch": 1.9287866337989592,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006541314556852989,
      "loss": 4.7714,
      "step": 14084
    },
    {
      "epoch": 1.928923582580115,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006540949467660814,
      "loss": 4.7133,
      "step": 14085
    },
    {
      "epoch": 1.9290605313612708,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006540584361554802,
      "loss": 4.7754,
      "step": 14086
    },
    {
      "epoch": 1.9291974801424268,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006540219238537982,
      "loss": 4.7346,
      "step": 14087
    },
    {
      "epoch": 1.9293344289235825,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000653985409861338,
      "loss": 4.8088,
      "step": 14088
    },
    {
      "epoch": 1.9294713777047385,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006539488941784019,
      "loss": 4.7495,
      "step": 14089
    },
    {
      "epoch": 1.9296083264858943,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006539123768052929,
      "loss": 4.7623,
      "step": 14090
    },
    {
      "epoch": 1.92974527526705,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006538758577423133,
      "loss": 4.7004,
      "step": 14091
    },
    {
      "epoch": 1.9298822240482059,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000653839336989766,
      "loss": 4.8144,
      "step": 14092
    },
    {
      "epoch": 1.9300191728293619,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006538028145479535,
      "loss": 4.7668,
      "step": 14093
    },
    {
      "epoch": 1.9301561216105176,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006537662904171785,
      "loss": 4.8449,
      "step": 14094
    },
    {
      "epoch": 1.9302930703916736,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006537297645977438,
      "loss": 4.7085,
      "step": 14095
    },
    {
      "epoch": 1.9304300191728294,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006536932370899518,
      "loss": 4.7182,
      "step": 14096
    },
    {
      "epoch": 1.9305669679539852,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006536567078941056,
      "loss": 4.7983,
      "step": 14097
    },
    {
      "epoch": 1.930703916735141,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006536201770105075,
      "loss": 4.8579,
      "step": 14098
    },
    {
      "epoch": 1.9308408655162967,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006535836444394608,
      "loss": 4.7791,
      "step": 14099
    },
    {
      "epoch": 1.9309778142974527,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006535471101812676,
      "loss": 4.7847,
      "step": 14100
    },
    {
      "epoch": 1.9311147630786087,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006535105742362312,
      "loss": 4.7767,
      "step": 14101
    },
    {
      "epoch": 1.9312517118597645,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006534740366046541,
      "loss": 4.7828,
      "step": 14102
    },
    {
      "epoch": 1.9313886606409203,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006534374972868393,
      "loss": 4.6906,
      "step": 14103
    },
    {
      "epoch": 1.931525609422076,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006534009562830892,
      "loss": 4.79,
      "step": 14104
    },
    {
      "epoch": 1.9316625582032318,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006533644135937072,
      "loss": 4.8071,
      "step": 14105
    },
    {
      "epoch": 1.9317995069843878,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006533278692189958,
      "loss": 4.7948,
      "step": 14106
    },
    {
      "epoch": 1.9319364557655438,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006532913231592577,
      "loss": 4.684,
      "step": 14107
    },
    {
      "epoch": 1.9320734045466996,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006532547754147962,
      "loss": 4.8255,
      "step": 14108
    },
    {
      "epoch": 1.9322103533278554,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006532182259859138,
      "loss": 4.7293,
      "step": 14109
    },
    {
      "epoch": 1.9323473021090112,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006531816748729136,
      "loss": 4.7974,
      "step": 14110
    },
    {
      "epoch": 1.932484250890167,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006531451220760983,
      "loss": 4.7731,
      "step": 14111
    },
    {
      "epoch": 1.932621199671323,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006531085675957713,
      "loss": 4.8083,
      "step": 14112
    },
    {
      "epoch": 1.9327581484524787,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006530720114322348,
      "loss": 4.751,
      "step": 14113
    },
    {
      "epoch": 1.9328950972336347,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006530354535857924,
      "loss": 4.7451,
      "step": 14114
    },
    {
      "epoch": 1.9330320460147905,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006529988940567467,
      "loss": 4.7854,
      "step": 14115
    },
    {
      "epoch": 1.9331689947959463,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006529623328454009,
      "loss": 4.7891,
      "step": 14116
    },
    {
      "epoch": 1.933305943577102,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006529257699520578,
      "loss": 4.8259,
      "step": 14117
    },
    {
      "epoch": 1.933442892358258,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006528892053770205,
      "loss": 4.8048,
      "step": 14118
    },
    {
      "epoch": 1.9335798411394138,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000652852639120592,
      "loss": 4.7158,
      "step": 14119
    },
    {
      "epoch": 1.9337167899205698,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006528160711830753,
      "loss": 4.815,
      "step": 14120
    },
    {
      "epoch": 1.9338537387017256,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006527795015647737,
      "loss": 4.7555,
      "step": 14121
    },
    {
      "epoch": 1.9339906874828814,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006527429302659899,
      "loss": 4.7761,
      "step": 14122
    },
    {
      "epoch": 1.9341276362640372,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006527063572870272,
      "loss": 4.7954,
      "step": 14123
    },
    {
      "epoch": 1.934264585045193,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006526697826281884,
      "loss": 4.8491,
      "step": 14124
    },
    {
      "epoch": 1.934401533826349,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000652633206289777,
      "loss": 4.8352,
      "step": 14125
    },
    {
      "epoch": 1.934538482607505,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000652596628272096,
      "loss": 4.8128,
      "step": 14126
    },
    {
      "epoch": 1.9346754313886607,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006525600485754483,
      "loss": 4.7164,
      "step": 14127
    },
    {
      "epoch": 1.9348123801698165,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006525234672001373,
      "loss": 4.76,
      "step": 14128
    },
    {
      "epoch": 1.9349493289509723,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000652486884146466,
      "loss": 4.6816,
      "step": 14129
    },
    {
      "epoch": 1.935086277732128,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006524502994147377,
      "loss": 4.7157,
      "step": 14130
    },
    {
      "epoch": 1.935223226513284,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006524137130052555,
      "loss": 4.8096,
      "step": 14131
    },
    {
      "epoch": 1.93536017529444,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006523771249183227,
      "loss": 4.7511,
      "step": 14132
    },
    {
      "epoch": 1.9354971240755958,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006523405351542424,
      "loss": 4.7023,
      "step": 14133
    },
    {
      "epoch": 1.9356340728567516,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006523039437133178,
      "loss": 4.7976,
      "step": 14134
    },
    {
      "epoch": 1.9357710216379074,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006522673505958523,
      "loss": 4.7913,
      "step": 14135
    },
    {
      "epoch": 1.9359079704190632,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000652230755802149,
      "loss": 4.8078,
      "step": 14136
    },
    {
      "epoch": 1.9360449192002191,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006521941593325113,
      "loss": 4.7633,
      "step": 14137
    },
    {
      "epoch": 1.936181867981375,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006521575611872424,
      "loss": 4.7499,
      "step": 14138
    },
    {
      "epoch": 1.936318816762531,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006521209613666456,
      "loss": 4.858,
      "step": 14139
    },
    {
      "epoch": 1.9364557655436867,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006520843598710242,
      "loss": 4.7719,
      "step": 14140
    },
    {
      "epoch": 1.9365927143248425,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006520477567006817,
      "loss": 4.8342,
      "step": 14141
    },
    {
      "epoch": 1.9367296631059983,
      "grad_norm": 0.34375,
      "learning_rate": 0.000652011151855921,
      "loss": 4.7613,
      "step": 14142
    },
    {
      "epoch": 1.9368666118871543,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006519745453370461,
      "loss": 4.7562,
      "step": 14143
    },
    {
      "epoch": 1.93700356066831,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006519379371443598,
      "loss": 4.7808,
      "step": 14144
    },
    {
      "epoch": 1.937140509449466,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006519013272781657,
      "loss": 4.8639,
      "step": 14145
    },
    {
      "epoch": 1.9372774582306218,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006518647157387673,
      "loss": 4.8101,
      "step": 14146
    },
    {
      "epoch": 1.9374144070117776,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006518281025264679,
      "loss": 4.7916,
      "step": 14147
    },
    {
      "epoch": 1.9375513557929334,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006517914876415709,
      "loss": 4.7608,
      "step": 14148
    },
    {
      "epoch": 1.9376883045740891,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006517548710843796,
      "loss": 4.7621,
      "step": 14149
    },
    {
      "epoch": 1.9378252533552451,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006517182528551978,
      "loss": 4.7548,
      "step": 14150
    },
    {
      "epoch": 1.9379622021364011,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006516816329543287,
      "loss": 4.739,
      "step": 14151
    },
    {
      "epoch": 1.938099150917557,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006516450113820758,
      "loss": 4.8259,
      "step": 14152
    },
    {
      "epoch": 1.9382360996987127,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006516083881387427,
      "loss": 4.8121,
      "step": 14153
    },
    {
      "epoch": 1.9383730484798685,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006515717632246329,
      "loss": 4.8098,
      "step": 14154
    },
    {
      "epoch": 1.9385099972610242,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006515351366400498,
      "loss": 4.8229,
      "step": 14155
    },
    {
      "epoch": 1.9386469460421802,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000651498508385297,
      "loss": 4.719,
      "step": 14156
    },
    {
      "epoch": 1.938783894823336,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006514618784606781,
      "loss": 4.7927,
      "step": 14157
    },
    {
      "epoch": 1.938920843604492,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006514252468664965,
      "loss": 4.7876,
      "step": 14158
    },
    {
      "epoch": 1.9390577923856478,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000651388613603056,
      "loss": 4.7163,
      "step": 14159
    },
    {
      "epoch": 1.9391947411668036,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006513519786706599,
      "loss": 4.7816,
      "step": 14160
    },
    {
      "epoch": 1.9393316899479593,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006513153420696121,
      "loss": 4.7602,
      "step": 14161
    },
    {
      "epoch": 1.9394686387291153,
      "grad_norm": 0.375,
      "learning_rate": 0.0006512787038002162,
      "loss": 4.8001,
      "step": 14162
    },
    {
      "epoch": 1.9396055875102711,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006512420638627754,
      "loss": 4.777,
      "step": 14163
    },
    {
      "epoch": 1.9397425362914271,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000651205422257594,
      "loss": 4.8116,
      "step": 14164
    },
    {
      "epoch": 1.939879485072583,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006511687789849751,
      "loss": 4.778,
      "step": 14165
    },
    {
      "epoch": 1.9400164338537387,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0006511321340452226,
      "loss": 4.7923,
      "step": 14166
    },
    {
      "epoch": 1.9401533826348945,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006510954874386403,
      "loss": 4.7725,
      "step": 14167
    },
    {
      "epoch": 1.9402903314160505,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006510588391655317,
      "loss": 4.7628,
      "step": 14168
    },
    {
      "epoch": 1.9404272801972062,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006510221892262007,
      "loss": 4.8122,
      "step": 14169
    },
    {
      "epoch": 1.9405642289783622,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006509855376209508,
      "loss": 4.7521,
      "step": 14170
    },
    {
      "epoch": 1.940701177759518,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006509488843500859,
      "loss": 4.8133,
      "step": 14171
    },
    {
      "epoch": 1.9408381265406738,
      "grad_norm": 0.4375,
      "learning_rate": 0.0006509122294139098,
      "loss": 4.7824,
      "step": 14172
    },
    {
      "epoch": 1.9409750753218296,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006508755728127258,
      "loss": 4.8284,
      "step": 14173
    },
    {
      "epoch": 1.9411120241029853,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006508389145468385,
      "loss": 4.7393,
      "step": 14174
    },
    {
      "epoch": 1.9412489728841413,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006508022546165511,
      "loss": 4.8224,
      "step": 14175
    },
    {
      "epoch": 1.9413859216652973,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0006507655930221676,
      "loss": 4.7513,
      "step": 14176
    },
    {
      "epoch": 1.941522870446453,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006507289297639918,
      "loss": 4.7574,
      "step": 14177
    },
    {
      "epoch": 1.9416598192276089,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006506922648423273,
      "loss": 4.7816,
      "step": 14178
    },
    {
      "epoch": 1.9417967680087647,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006506555982574785,
      "loss": 4.8457,
      "step": 14179
    },
    {
      "epoch": 1.9419337167899204,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006506189300097487,
      "loss": 4.7512,
      "step": 14180
    },
    {
      "epoch": 1.9420706655710764,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006505822600994422,
      "loss": 4.7636,
      "step": 14181
    },
    {
      "epoch": 1.9422076143522322,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006505455885268626,
      "loss": 4.7789,
      "step": 14182
    },
    {
      "epoch": 1.9423445631333882,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000650508915292314,
      "loss": 4.7906,
      "step": 14183
    },
    {
      "epoch": 1.942481511914544,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006504722403961002,
      "loss": 4.7246,
      "step": 14184
    },
    {
      "epoch": 1.9426184606956998,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006504355638385252,
      "loss": 4.7052,
      "step": 14185
    },
    {
      "epoch": 1.9427554094768555,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006503988856198929,
      "loss": 4.7512,
      "step": 14186
    },
    {
      "epoch": 1.9428923582580115,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006503622057405072,
      "loss": 4.7862,
      "step": 14187
    },
    {
      "epoch": 1.9430293070391673,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006503255242006723,
      "loss": 4.8575,
      "step": 14188
    },
    {
      "epoch": 1.9431662558203233,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000650288841000692,
      "loss": 4.7822,
      "step": 14189
    },
    {
      "epoch": 1.943303204601479,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006502521561408703,
      "loss": 4.7978,
      "step": 14190
    },
    {
      "epoch": 1.9434401533826349,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006502154696215114,
      "loss": 4.8346,
      "step": 14191
    },
    {
      "epoch": 1.9435771021637906,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000650178781442919,
      "loss": 4.7416,
      "step": 14192
    },
    {
      "epoch": 1.9437140509449464,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006501420916053975,
      "loss": 4.7634,
      "step": 14193
    },
    {
      "epoch": 1.9438509997261024,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006501054001092507,
      "loss": 4.7634,
      "step": 14194
    },
    {
      "epoch": 1.9439879485072584,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006500687069547829,
      "loss": 4.7131,
      "step": 14195
    },
    {
      "epoch": 1.9441248972884142,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0006500320121422978,
      "loss": 4.7739,
      "step": 14196
    },
    {
      "epoch": 1.94426184606957,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006499953156721,
      "loss": 4.7169,
      "step": 14197
    },
    {
      "epoch": 1.9443987948507258,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006499586175444933,
      "loss": 4.8091,
      "step": 14198
    },
    {
      "epoch": 1.9445357436318815,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006499219177597817,
      "loss": 4.7334,
      "step": 14199
    },
    {
      "epoch": 1.9446726924130375,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006498852163182697,
      "loss": 4.7526,
      "step": 14200
    },
    {
      "epoch": 1.9448096411941935,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006498485132202613,
      "loss": 4.8113,
      "step": 14201
    },
    {
      "epoch": 1.9449465899753493,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006498118084660606,
      "loss": 4.8192,
      "step": 14202
    },
    {
      "epoch": 1.945083538756505,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006497751020559719,
      "loss": 4.7355,
      "step": 14203
    },
    {
      "epoch": 1.9452204875376609,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006497383939902992,
      "loss": 4.7111,
      "step": 14204
    },
    {
      "epoch": 1.9453574363188166,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000649701684269347,
      "loss": 4.724,
      "step": 14205
    },
    {
      "epoch": 1.9454943850999726,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006496649728934192,
      "loss": 4.7329,
      "step": 14206
    },
    {
      "epoch": 1.9456313338811284,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006496282598628201,
      "loss": 4.7502,
      "step": 14207
    },
    {
      "epoch": 1.9457682826622844,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006495915451778543,
      "loss": 4.678,
      "step": 14208
    },
    {
      "epoch": 1.9459052314434402,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006495548288388257,
      "loss": 4.727,
      "step": 14209
    },
    {
      "epoch": 1.946042180224596,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006495181108460387,
      "loss": 4.8307,
      "step": 14210
    },
    {
      "epoch": 1.9461791290057517,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006494813911997975,
      "loss": 4.789,
      "step": 14211
    },
    {
      "epoch": 1.9463160777869077,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006494446699004065,
      "loss": 4.7796,
      "step": 14212
    },
    {
      "epoch": 1.9464530265680635,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00064940794694817,
      "loss": 4.813,
      "step": 14213
    },
    {
      "epoch": 1.9465899753492195,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006493712223433924,
      "loss": 4.7558,
      "step": 14214
    },
    {
      "epoch": 1.9467269241303753,
      "grad_norm": 0.34375,
      "learning_rate": 0.000649334496086378,
      "loss": 4.7736,
      "step": 14215
    },
    {
      "epoch": 1.946863872911531,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000649297768177431,
      "loss": 4.776,
      "step": 14216
    },
    {
      "epoch": 1.9470008216926868,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000649261038616856,
      "loss": 4.7375,
      "step": 14217
    },
    {
      "epoch": 1.9471377704738426,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006492243074049573,
      "loss": 4.7433,
      "step": 14218
    },
    {
      "epoch": 1.9472747192549986,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006491875745420392,
      "loss": 4.7657,
      "step": 14219
    },
    {
      "epoch": 1.9474116680361546,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006491508400284062,
      "loss": 4.8624,
      "step": 14220
    },
    {
      "epoch": 1.9475486168173104,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006491141038643628,
      "loss": 4.8093,
      "step": 14221
    },
    {
      "epoch": 1.9476855655984662,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006490773660502133,
      "loss": 4.6963,
      "step": 14222
    },
    {
      "epoch": 1.947822514379622,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006490406265862624,
      "loss": 4.7874,
      "step": 14223
    },
    {
      "epoch": 1.9479594631607777,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006490038854728143,
      "loss": 4.7064,
      "step": 14224
    },
    {
      "epoch": 1.9480964119419337,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006489671427101736,
      "loss": 4.8007,
      "step": 14225
    },
    {
      "epoch": 1.9482333607230897,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006489303982986448,
      "loss": 4.6908,
      "step": 14226
    },
    {
      "epoch": 1.9483703095042455,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006488936522385322,
      "loss": 4.774,
      "step": 14227
    },
    {
      "epoch": 1.9485072582854013,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006488569045301405,
      "loss": 4.8054,
      "step": 14228
    },
    {
      "epoch": 1.948644207066557,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006488201551737746,
      "loss": 4.7853,
      "step": 14229
    },
    {
      "epoch": 1.9487811558477128,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006487834041697384,
      "loss": 4.6731,
      "step": 14230
    },
    {
      "epoch": 1.9489181046288688,
      "grad_norm": 0.34375,
      "learning_rate": 0.000648746651518337,
      "loss": 4.7941,
      "step": 14231
    },
    {
      "epoch": 1.9490550534100246,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006487098972198744,
      "loss": 4.7277,
      "step": 14232
    },
    {
      "epoch": 1.9491920021911806,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006486731412746558,
      "loss": 4.8147,
      "step": 14233
    },
    {
      "epoch": 1.9493289509723364,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006486363836829854,
      "loss": 4.7467,
      "step": 14234
    },
    {
      "epoch": 1.9494658997534922,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000648599624445168,
      "loss": 4.7625,
      "step": 14235
    },
    {
      "epoch": 1.949602848534648,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006485628635615082,
      "loss": 4.7784,
      "step": 14236
    },
    {
      "epoch": 1.949739797315804,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006485261010323106,
      "loss": 4.8211,
      "step": 14237
    },
    {
      "epoch": 1.9498767460969597,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006484893368578801,
      "loss": 4.6991,
      "step": 14238
    },
    {
      "epoch": 1.9500136948781157,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006484525710385209,
      "loss": 4.7194,
      "step": 14239
    },
    {
      "epoch": 1.9501506436592715,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006484158035745381,
      "loss": 4.8012,
      "step": 14240
    },
    {
      "epoch": 1.9502875924404273,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006483790344662362,
      "loss": 4.7214,
      "step": 14241
    },
    {
      "epoch": 1.950424541221583,
      "grad_norm": 0.421875,
      "learning_rate": 0.00064834226371392,
      "loss": 4.752,
      "step": 14242
    },
    {
      "epoch": 1.9505614900027388,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0006483054913178941,
      "loss": 4.765,
      "step": 14243
    },
    {
      "epoch": 1.9506984387838948,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006482687172784635,
      "loss": 4.8151,
      "step": 14244
    },
    {
      "epoch": 1.9508353875650508,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0006482319415959327,
      "loss": 4.7758,
      "step": 14245
    },
    {
      "epoch": 1.9509723363462066,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006481951642706065,
      "loss": 4.7118,
      "step": 14246
    },
    {
      "epoch": 1.9511092851273624,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006481583853027899,
      "loss": 4.8658,
      "step": 14247
    },
    {
      "epoch": 1.9512462339085181,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006481216046927874,
      "loss": 4.8002,
      "step": 14248
    },
    {
      "epoch": 1.951383182689674,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006480848224409041,
      "loss": 4.7518,
      "step": 14249
    },
    {
      "epoch": 1.95152013147083,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006480480385474446,
      "loss": 4.8376,
      "step": 14250
    },
    {
      "epoch": 1.951657080251986,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006480112530127138,
      "loss": 4.7508,
      "step": 14251
    },
    {
      "epoch": 1.9517940290331417,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006479744658370166,
      "loss": 4.7919,
      "step": 14252
    },
    {
      "epoch": 1.9519309778142975,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0006479376770206578,
      "loss": 4.7403,
      "step": 14253
    },
    {
      "epoch": 1.9520679265954533,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006479008865639422,
      "loss": 4.7498,
      "step": 14254
    },
    {
      "epoch": 1.952204875376609,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006478640944671749,
      "loss": 4.7667,
      "step": 14255
    },
    {
      "epoch": 1.952341824157765,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006478273007306608,
      "loss": 4.7097,
      "step": 14256
    },
    {
      "epoch": 1.9524787729389208,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006477905053547046,
      "loss": 4.7415,
      "step": 14257
    },
    {
      "epoch": 1.9526157217200768,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006477537083396113,
      "loss": 4.711,
      "step": 14258
    },
    {
      "epoch": 1.9527526705012326,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006477169096856859,
      "loss": 4.8067,
      "step": 14259
    },
    {
      "epoch": 1.9528896192823884,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006476801093932335,
      "loss": 4.7729,
      "step": 14260
    },
    {
      "epoch": 1.9530265680635441,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006476433074625588,
      "loss": 4.8621,
      "step": 14261
    },
    {
      "epoch": 1.9531635168447001,
      "grad_norm": 0.375,
      "learning_rate": 0.0006476065038939671,
      "loss": 4.7432,
      "step": 14262
    },
    {
      "epoch": 1.953300465625856,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000647569698687763,
      "loss": 4.7535,
      "step": 14263
    },
    {
      "epoch": 1.953437414407012,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006475328918442517,
      "loss": 4.7735,
      "step": 14264
    },
    {
      "epoch": 1.9535743631881677,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006474960833637386,
      "loss": 4.8081,
      "step": 14265
    },
    {
      "epoch": 1.9537113119693235,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000647459273246528,
      "loss": 4.7915,
      "step": 14266
    },
    {
      "epoch": 1.9538482607504792,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006474224614929256,
      "loss": 4.7038,
      "step": 14267
    },
    {
      "epoch": 1.953985209531635,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006473856481032361,
      "loss": 4.8427,
      "step": 14268
    },
    {
      "epoch": 1.954122158312791,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006473488330777647,
      "loss": 4.801,
      "step": 14269
    },
    {
      "epoch": 1.954259107093947,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006473120164168165,
      "loss": 4.7972,
      "step": 14270
    },
    {
      "epoch": 1.9543960558751028,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006472751981206967,
      "loss": 4.8152,
      "step": 14271
    },
    {
      "epoch": 1.9545330046562586,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006472383781897102,
      "loss": 4.7558,
      "step": 14272
    },
    {
      "epoch": 1.9546699534374143,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006472015566241623,
      "loss": 4.8275,
      "step": 14273
    },
    {
      "epoch": 1.9548069022185701,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006471647334243581,
      "loss": 4.7207,
      "step": 14274
    },
    {
      "epoch": 1.9549438509997261,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006471279085906028,
      "loss": 4.8535,
      "step": 14275
    },
    {
      "epoch": 1.955080799780882,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006470910821232016,
      "loss": 4.8398,
      "step": 14276
    },
    {
      "epoch": 1.955217748562038,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006470542540224596,
      "loss": 4.7182,
      "step": 14277
    },
    {
      "epoch": 1.9553546973431937,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006470174242886822,
      "loss": 4.789,
      "step": 14278
    },
    {
      "epoch": 1.9554916461243494,
      "grad_norm": 0.34375,
      "learning_rate": 0.000646980592922174,
      "loss": 4.8344,
      "step": 14279
    },
    {
      "epoch": 1.9556285949055052,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006469437599232412,
      "loss": 4.8249,
      "step": 14280
    },
    {
      "epoch": 1.9557655436866612,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006469069252921882,
      "loss": 4.8456,
      "step": 14281
    },
    {
      "epoch": 1.955902492467817,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006468700890293207,
      "loss": 4.8198,
      "step": 14282
    },
    {
      "epoch": 1.956039441248973,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006468332511349439,
      "loss": 4.8687,
      "step": 14283
    },
    {
      "epoch": 1.9561763900301288,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006467964116093629,
      "loss": 4.7588,
      "step": 14284
    },
    {
      "epoch": 1.9563133388112846,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006467595704528833,
      "loss": 4.7734,
      "step": 14285
    },
    {
      "epoch": 1.9564502875924403,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006467227276658102,
      "loss": 4.8099,
      "step": 14286
    },
    {
      "epoch": 1.9565872363735963,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006466858832484489,
      "loss": 4.7763,
      "step": 14287
    },
    {
      "epoch": 1.956724185154752,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006466490372011049,
      "loss": 4.8463,
      "step": 14288
    },
    {
      "epoch": 1.956861133935908,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006466121895240835,
      "loss": 4.738,
      "step": 14289
    },
    {
      "epoch": 1.9569980827170639,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00064657534021769,
      "loss": 4.7354,
      "step": 14290
    },
    {
      "epoch": 1.9571350314982197,
      "grad_norm": 0.375,
      "learning_rate": 0.0006465384892822299,
      "loss": 4.7395,
      "step": 14291
    },
    {
      "epoch": 1.9572719802793754,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006465016367180084,
      "loss": 4.7692,
      "step": 14292
    },
    {
      "epoch": 1.9574089290605312,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006464647825253309,
      "loss": 4.6677,
      "step": 14293
    },
    {
      "epoch": 1.9575458778416872,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006464279267045032,
      "loss": 4.7395,
      "step": 14294
    },
    {
      "epoch": 1.9576828266228432,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006463910692558302,
      "loss": 4.7825,
      "step": 14295
    },
    {
      "epoch": 1.957819775403999,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006463542101796178,
      "loss": 4.8969,
      "step": 14296
    },
    {
      "epoch": 1.9579567241851548,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006463173494761711,
      "loss": 4.8231,
      "step": 14297
    },
    {
      "epoch": 1.9580936729663105,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000646280487145796,
      "loss": 4.7673,
      "step": 14298
    },
    {
      "epoch": 1.9582306217474663,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006462436231887975,
      "loss": 4.8653,
      "step": 14299
    },
    {
      "epoch": 1.9583675705286223,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006462067576054816,
      "loss": 4.7551,
      "step": 14300
    },
    {
      "epoch": 1.958504519309778,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006461698903961532,
      "loss": 4.6944,
      "step": 14301
    },
    {
      "epoch": 1.958641468090934,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006461330215611184,
      "loss": 4.8358,
      "step": 14302
    },
    {
      "epoch": 1.9587784168720899,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006460961511006824,
      "loss": 4.6917,
      "step": 14303
    },
    {
      "epoch": 1.9589153656532456,
      "grad_norm": 0.375,
      "learning_rate": 0.0006460592790151509,
      "loss": 4.7873,
      "step": 14304
    },
    {
      "epoch": 1.9590523144344014,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006460224053048295,
      "loss": 4.8853,
      "step": 14305
    },
    {
      "epoch": 1.9591892632155574,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006459855299700237,
      "loss": 4.7192,
      "step": 14306
    },
    {
      "epoch": 1.9593262119967132,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006459486530110392,
      "loss": 4.778,
      "step": 14307
    },
    {
      "epoch": 1.9594631607778692,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0006459117744281813,
      "loss": 4.7701,
      "step": 14308
    },
    {
      "epoch": 1.959600109559025,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0006458748942217559,
      "loss": 4.7626,
      "step": 14309
    },
    {
      "epoch": 1.9597370583401807,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006458380123920686,
      "loss": 4.7965,
      "step": 14310
    },
    {
      "epoch": 1.9598740071213365,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006458011289394251,
      "loss": 4.7763,
      "step": 14311
    },
    {
      "epoch": 1.9600109559024923,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000645764243864131,
      "loss": 4.8081,
      "step": 14312
    },
    {
      "epoch": 1.9601479046836483,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006457273571664918,
      "loss": 4.8984,
      "step": 14313
    },
    {
      "epoch": 1.9602848534648043,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006456904688468136,
      "loss": 4.7384,
      "step": 14314
    },
    {
      "epoch": 1.96042180224596,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006456535789054016,
      "loss": 4.7444,
      "step": 14315
    },
    {
      "epoch": 1.9605587510271159,
      "grad_norm": 0.34375,
      "learning_rate": 0.000645616687342562,
      "loss": 4.7731,
      "step": 14316
    },
    {
      "epoch": 1.9606956998082716,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006455797941586001,
      "loss": 4.8252,
      "step": 14317
    },
    {
      "epoch": 1.9608326485894274,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006455428993538219,
      "loss": 4.7473,
      "step": 14318
    },
    {
      "epoch": 1.9609695973705834,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006455060029285331,
      "loss": 4.7163,
      "step": 14319
    },
    {
      "epoch": 1.9611065461517394,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006454691048830395,
      "loss": 4.7244,
      "step": 14320
    },
    {
      "epoch": 1.9612434949328952,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000645432205217647,
      "loss": 4.7481,
      "step": 14321
    },
    {
      "epoch": 1.961380443714051,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006453953039326609,
      "loss": 4.7889,
      "step": 14322
    },
    {
      "epoch": 1.9615173924952067,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006453584010283876,
      "loss": 4.8613,
      "step": 14323
    },
    {
      "epoch": 1.9616543412763625,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006453214965051326,
      "loss": 4.7714,
      "step": 14324
    },
    {
      "epoch": 1.9617912900575185,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006452845903632019,
      "loss": 4.8368,
      "step": 14325
    },
    {
      "epoch": 1.9619282388386743,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006452476826029011,
      "loss": 4.7521,
      "step": 14326
    },
    {
      "epoch": 1.9620651876198303,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006452107732245365,
      "loss": 4.8039,
      "step": 14327
    },
    {
      "epoch": 1.962202136400986,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0006451738622284135,
      "loss": 4.7024,
      "step": 14328
    },
    {
      "epoch": 1.9623390851821418,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006451369496148382,
      "loss": 4.7677,
      "step": 14329
    },
    {
      "epoch": 1.9624760339632976,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006451000353841164,
      "loss": 4.771,
      "step": 14330
    },
    {
      "epoch": 1.9626129827444536,
      "grad_norm": 0.375,
      "learning_rate": 0.0006450631195365543,
      "loss": 4.8169,
      "step": 14331
    },
    {
      "epoch": 1.9627499315256094,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006450262020724577,
      "loss": 4.8282,
      "step": 14332
    },
    {
      "epoch": 1.9628868803067654,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006449892829921323,
      "loss": 4.7518,
      "step": 14333
    },
    {
      "epoch": 1.9630238290879212,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006449523622958843,
      "loss": 4.7438,
      "step": 14334
    },
    {
      "epoch": 1.963160777869077,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006449154399840197,
      "loss": 4.7665,
      "step": 14335
    },
    {
      "epoch": 1.9632977266502327,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006448785160568445,
      "loss": 4.811,
      "step": 14336
    },
    {
      "epoch": 1.9634346754313885,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006448415905146643,
      "loss": 4.7732,
      "step": 14337
    },
    {
      "epoch": 1.9635716242125445,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006448046633577856,
      "loss": 4.8932,
      "step": 14338
    },
    {
      "epoch": 1.9637085729937005,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006447677345865142,
      "loss": 4.8463,
      "step": 14339
    },
    {
      "epoch": 1.9638455217748563,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006447308042011562,
      "loss": 4.7304,
      "step": 14340
    },
    {
      "epoch": 1.963982470556012,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006446938722020176,
      "loss": 4.8937,
      "step": 14341
    },
    {
      "epoch": 1.9641194193371678,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006446569385894044,
      "loss": 4.7485,
      "step": 14342
    },
    {
      "epoch": 1.9642563681183236,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000644620003363623,
      "loss": 4.8685,
      "step": 14343
    },
    {
      "epoch": 1.9643933168994796,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006445830665249791,
      "loss": 4.8131,
      "step": 14344
    },
    {
      "epoch": 1.9645302656806356,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000644546128073779,
      "loss": 4.8375,
      "step": 14345
    },
    {
      "epoch": 1.9646672144617914,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006445091880103287,
      "loss": 4.8069,
      "step": 14346
    },
    {
      "epoch": 1.9648041632429472,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006444722463349344,
      "loss": 4.7459,
      "step": 14347
    },
    {
      "epoch": 1.964941112024103,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006444353030479024,
      "loss": 4.7638,
      "step": 14348
    },
    {
      "epoch": 1.9650780608052587,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006443983581495386,
      "loss": 4.7501,
      "step": 14349
    },
    {
      "epoch": 1.9652150095864147,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006443614116401494,
      "loss": 4.8424,
      "step": 14350
    },
    {
      "epoch": 1.9653519583675705,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006443244635200408,
      "loss": 4.7968,
      "step": 14351
    },
    {
      "epoch": 1.9654889071487265,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000644287513789519,
      "loss": 4.746,
      "step": 14352
    },
    {
      "epoch": 1.9656258559298823,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006442505624488903,
      "loss": 4.7872,
      "step": 14353
    },
    {
      "epoch": 1.965762804711038,
      "grad_norm": 0.34375,
      "learning_rate": 0.000644213609498461,
      "loss": 4.858,
      "step": 14354
    },
    {
      "epoch": 1.9658997534921938,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006441766549385371,
      "loss": 4.8583,
      "step": 14355
    },
    {
      "epoch": 1.9660367022733498,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006441396987694251,
      "loss": 4.7779,
      "step": 14356
    },
    {
      "epoch": 1.9661736510545056,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000644102740991431,
      "loss": 4.8073,
      "step": 14357
    },
    {
      "epoch": 1.9663105998356616,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006440657816048614,
      "loss": 4.7815,
      "step": 14358
    },
    {
      "epoch": 1.9664475486168174,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0006440288206100224,
      "loss": 4.7165,
      "step": 14359
    },
    {
      "epoch": 1.9665844973979731,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006439918580072203,
      "loss": 4.8203,
      "step": 14360
    },
    {
      "epoch": 1.966721446179129,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006439548937967614,
      "loss": 4.7614,
      "step": 14361
    },
    {
      "epoch": 1.9668583949602847,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006439179279789521,
      "loss": 4.7544,
      "step": 14362
    },
    {
      "epoch": 1.9669953437414407,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006438809605540988,
      "loss": 4.8331,
      "step": 14363
    },
    {
      "epoch": 1.9671322925225967,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006438439915225077,
      "loss": 4.7072,
      "step": 14364
    },
    {
      "epoch": 1.9672692413037525,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006438070208844853,
      "loss": 4.8536,
      "step": 14365
    },
    {
      "epoch": 1.9674061900849082,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006437700486403378,
      "loss": 4.6231,
      "step": 14366
    },
    {
      "epoch": 1.967543138866064,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0006437330747903719,
      "loss": 4.8076,
      "step": 14367
    },
    {
      "epoch": 1.9676800876472198,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006436960993348939,
      "loss": 4.7565,
      "step": 14368
    },
    {
      "epoch": 1.9678170364283758,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00064365912227421,
      "loss": 4.743,
      "step": 14369
    },
    {
      "epoch": 1.9679539852095318,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006436221436086269,
      "loss": 4.7036,
      "step": 14370
    },
    {
      "epoch": 1.9680909339906876,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006435851633384508,
      "loss": 4.7541,
      "step": 14371
    },
    {
      "epoch": 1.9682278827718434,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006435481814639886,
      "loss": 4.8297,
      "step": 14372
    },
    {
      "epoch": 1.9683648315529991,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006435111979855463,
      "loss": 4.7999,
      "step": 14373
    },
    {
      "epoch": 1.968501780334155,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006434742129034307,
      "loss": 4.7411,
      "step": 14374
    },
    {
      "epoch": 1.968638729115311,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000643437226217948,
      "loss": 4.7616,
      "step": 14375
    },
    {
      "epoch": 1.9687756778964667,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000643400237929405,
      "loss": 4.7793,
      "step": 14376
    },
    {
      "epoch": 1.9689126266776227,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006433632480381083,
      "loss": 4.8562,
      "step": 14377
    },
    {
      "epoch": 1.9690495754587785,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006433262565443641,
      "loss": 4.8256,
      "step": 14378
    },
    {
      "epoch": 1.9691865242399342,
      "grad_norm": 0.375,
      "learning_rate": 0.0006432892634484793,
      "loss": 4.7246,
      "step": 14379
    },
    {
      "epoch": 1.96932347302109,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006432522687507602,
      "loss": 4.6846,
      "step": 14380
    },
    {
      "epoch": 1.969460421802246,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006432152724515136,
      "loss": 4.832,
      "step": 14381
    },
    {
      "epoch": 1.9695973705834018,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000643178274551046,
      "loss": 4.7784,
      "step": 14382
    },
    {
      "epoch": 1.9697343193645578,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000643141275049664,
      "loss": 4.8836,
      "step": 14383
    },
    {
      "epoch": 1.9698712681457136,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000643104273947674,
      "loss": 4.7703,
      "step": 14384
    },
    {
      "epoch": 1.9700082169268693,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000643067271245383,
      "loss": 4.8783,
      "step": 14385
    },
    {
      "epoch": 1.9701451657080251,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006430302669430977,
      "loss": 4.7267,
      "step": 14386
    },
    {
      "epoch": 1.970282114489181,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006429932610411244,
      "loss": 4.7729,
      "step": 14387
    },
    {
      "epoch": 1.970419063270337,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006429562535397701,
      "loss": 4.6766,
      "step": 14388
    },
    {
      "epoch": 1.970556012051493,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006429192444393411,
      "loss": 4.7238,
      "step": 14389
    },
    {
      "epoch": 1.9706929608326487,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006428822337401446,
      "loss": 4.8352,
      "step": 14390
    },
    {
      "epoch": 1.9708299096138044,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006428452214424869,
      "loss": 4.7652,
      "step": 14391
    },
    {
      "epoch": 1.9709668583949602,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000642808207546675,
      "loss": 4.7579,
      "step": 14392
    },
    {
      "epoch": 1.971103807176116,
      "grad_norm": 0.375,
      "learning_rate": 0.0006427711920530155,
      "loss": 4.8659,
      "step": 14393
    },
    {
      "epoch": 1.971240755957272,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006427341749618152,
      "loss": 4.736,
      "step": 14394
    },
    {
      "epoch": 1.971377704738428,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006426971562733809,
      "loss": 4.8353,
      "step": 14395
    },
    {
      "epoch": 1.9715146535195838,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006426601359880192,
      "loss": 4.8429,
      "step": 14396
    },
    {
      "epoch": 1.9716516023007395,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006426231141060373,
      "loss": 4.8035,
      "step": 14397
    },
    {
      "epoch": 1.9717885510818953,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006425860906277415,
      "loss": 4.7381,
      "step": 14398
    },
    {
      "epoch": 1.971925499863051,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006425490655534389,
      "loss": 4.7983,
      "step": 14399
    },
    {
      "epoch": 1.972062448644207,
      "grad_norm": 0.375,
      "learning_rate": 0.0006425120388834365,
      "loss": 4.7721,
      "step": 14400
    },
    {
      "epoch": 1.9721993974253629,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006424750106180408,
      "loss": 4.7643,
      "step": 14401
    },
    {
      "epoch": 1.9723363462065189,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006424379807575591,
      "loss": 4.7969,
      "step": 14402
    },
    {
      "epoch": 1.9724732949876747,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006424009493022976,
      "loss": 4.775,
      "step": 14403
    },
    {
      "epoch": 1.9726102437688304,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006423639162525638,
      "loss": 4.8587,
      "step": 14404
    },
    {
      "epoch": 1.9727471925499862,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006423268816086645,
      "loss": 4.7524,
      "step": 14405
    },
    {
      "epoch": 1.9728841413311422,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006422898453709062,
      "loss": 4.7193,
      "step": 14406
    },
    {
      "epoch": 1.973021090112298,
      "grad_norm": 0.328125,
      "learning_rate": 0.0006422528075395963,
      "loss": 4.7054,
      "step": 14407
    },
    {
      "epoch": 1.973158038893454,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006422157681150416,
      "loss": 4.7725,
      "step": 14408
    },
    {
      "epoch": 1.9732949876746098,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006421787270975489,
      "loss": 4.7719,
      "step": 14409
    },
    {
      "epoch": 1.9734319364557655,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006421416844874254,
      "loss": 4.8136,
      "step": 14410
    },
    {
      "epoch": 1.9735688852369213,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000642104640284978,
      "loss": 4.774,
      "step": 14411
    },
    {
      "epoch": 1.973705834018077,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006420675944905136,
      "loss": 4.7003,
      "step": 14412
    },
    {
      "epoch": 1.973842782799233,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006420305471043394,
      "loss": 4.7718,
      "step": 14413
    },
    {
      "epoch": 1.973979731580389,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006419934981267622,
      "loss": 4.7613,
      "step": 14414
    },
    {
      "epoch": 1.9741166803615449,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006419564475580892,
      "loss": 4.6993,
      "step": 14415
    },
    {
      "epoch": 1.9742536291427006,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006419193953986273,
      "loss": 4.8312,
      "step": 14416
    },
    {
      "epoch": 1.9743905779238564,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006418823416486836,
      "loss": 4.7531,
      "step": 14417
    },
    {
      "epoch": 1.9745275267050122,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006418452863085652,
      "loss": 4.7873,
      "step": 14418
    },
    {
      "epoch": 1.9746644754861682,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006418082293785793,
      "loss": 4.776,
      "step": 14419
    },
    {
      "epoch": 1.974801424267324,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006417711708590328,
      "loss": 4.8464,
      "step": 14420
    },
    {
      "epoch": 1.97493837304848,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000641734110750233,
      "loss": 4.892,
      "step": 14421
    },
    {
      "epoch": 1.9750753218296357,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006416970490524869,
      "loss": 4.8389,
      "step": 14422
    },
    {
      "epoch": 1.9752122706107915,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006416599857661017,
      "loss": 4.7903,
      "step": 14423
    },
    {
      "epoch": 1.9753492193919473,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006416229208913845,
      "loss": 4.7793,
      "step": 14424
    },
    {
      "epoch": 1.9754861681731033,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006415858544286425,
      "loss": 4.7671,
      "step": 14425
    },
    {
      "epoch": 1.975623116954259,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000641548786378183,
      "loss": 4.6412,
      "step": 14426
    },
    {
      "epoch": 1.975760065735415,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006415117167403128,
      "loss": 4.7702,
      "step": 14427
    },
    {
      "epoch": 1.9758970145165708,
      "grad_norm": 0.375,
      "learning_rate": 0.0006414746455153396,
      "loss": 4.7238,
      "step": 14428
    },
    {
      "epoch": 1.9760339632977266,
      "grad_norm": 0.375,
      "learning_rate": 0.0006414375727035703,
      "loss": 4.8045,
      "step": 14429
    },
    {
      "epoch": 1.9761709120788824,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006414004983053122,
      "loss": 4.7098,
      "step": 14430
    },
    {
      "epoch": 1.9763078608600382,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006413634223208725,
      "loss": 4.83,
      "step": 14431
    },
    {
      "epoch": 1.9764448096411942,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006413263447505587,
      "loss": 4.849,
      "step": 14432
    },
    {
      "epoch": 1.9765817584223502,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006412892655946778,
      "loss": 4.7664,
      "step": 14433
    },
    {
      "epoch": 1.976718707203506,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006412521848535371,
      "loss": 4.7463,
      "step": 14434
    },
    {
      "epoch": 1.9768556559846617,
      "grad_norm": 0.390625,
      "learning_rate": 0.000641215102527444,
      "loss": 4.7405,
      "step": 14435
    },
    {
      "epoch": 1.9769926047658175,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006411780186167058,
      "loss": 4.7813,
      "step": 14436
    },
    {
      "epoch": 1.9771295535469733,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00064114093312163,
      "loss": 4.6637,
      "step": 14437
    },
    {
      "epoch": 1.9772665023281293,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006411038460425233,
      "loss": 4.7457,
      "step": 14438
    },
    {
      "epoch": 1.9774034511092853,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006410667573796939,
      "loss": 4.7529,
      "step": 14439
    },
    {
      "epoch": 1.977540399890441,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006410296671334485,
      "loss": 4.6819,
      "step": 14440
    },
    {
      "epoch": 1.9776773486715968,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006409925753040949,
      "loss": 4.8066,
      "step": 14441
    },
    {
      "epoch": 1.9778142974527526,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006409554818919403,
      "loss": 4.7131,
      "step": 14442
    },
    {
      "epoch": 1.9779512462339084,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006409183868972921,
      "loss": 4.6807,
      "step": 14443
    },
    {
      "epoch": 1.9780881950150644,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006408812903204576,
      "loss": 4.739,
      "step": 14444
    },
    {
      "epoch": 1.9782251437962202,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006408441921617445,
      "loss": 4.7411,
      "step": 14445
    },
    {
      "epoch": 1.9783620925773762,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006408070924214601,
      "loss": 4.704,
      "step": 14446
    },
    {
      "epoch": 1.978499041358532,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006407699910999118,
      "loss": 4.7287,
      "step": 14447
    },
    {
      "epoch": 1.9786359901396877,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006407328881974072,
      "loss": 4.8263,
      "step": 14448
    },
    {
      "epoch": 1.9787729389208435,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006406957837142537,
      "loss": 4.7625,
      "step": 14449
    },
    {
      "epoch": 1.9789098877019995,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006406586776507586,
      "loss": 4.7799,
      "step": 14450
    },
    {
      "epoch": 1.9790468364831553,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006406215700072299,
      "loss": 4.7677,
      "step": 14451
    },
    {
      "epoch": 1.9791837852643113,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006405844607839746,
      "loss": 4.7422,
      "step": 14452
    },
    {
      "epoch": 1.979320734045467,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006405473499813007,
      "loss": 4.7905,
      "step": 14453
    },
    {
      "epoch": 1.9794576828266228,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006405102375995152,
      "loss": 4.7424,
      "step": 14454
    },
    {
      "epoch": 1.9795946316077786,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006404731236389262,
      "loss": 4.769,
      "step": 14455
    },
    {
      "epoch": 1.9797315803889344,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000640436008099841,
      "loss": 4.7826,
      "step": 14456
    },
    {
      "epoch": 1.9798685291700904,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006403988909825672,
      "loss": 4.8515,
      "step": 14457
    },
    {
      "epoch": 1.9800054779512464,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006403617722874123,
      "loss": 4.7932,
      "step": 14458
    },
    {
      "epoch": 1.9801424267324021,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006403246520146843,
      "loss": 4.7543,
      "step": 14459
    },
    {
      "epoch": 1.980279375513558,
      "grad_norm": 0.375,
      "learning_rate": 0.0006402875301646904,
      "loss": 4.7354,
      "step": 14460
    },
    {
      "epoch": 1.9804163242947137,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006402504067377382,
      "loss": 4.8036,
      "step": 14461
    },
    {
      "epoch": 1.9805532730758695,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006402132817341358,
      "loss": 4.8021,
      "step": 14462
    },
    {
      "epoch": 1.9806902218570255,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006401761551541906,
      "loss": 4.7673,
      "step": 14463
    },
    {
      "epoch": 1.9808271706381815,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006401390269982102,
      "loss": 4.7102,
      "step": 14464
    },
    {
      "epoch": 1.9809641194193373,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006401018972665023,
      "loss": 4.8173,
      "step": 14465
    },
    {
      "epoch": 1.981101068200493,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006400647659593749,
      "loss": 4.7385,
      "step": 14466
    },
    {
      "epoch": 1.9812380169816488,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006400276330771352,
      "loss": 4.7847,
      "step": 14467
    },
    {
      "epoch": 1.9813749657628046,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006399904986200915,
      "loss": 4.8945,
      "step": 14468
    },
    {
      "epoch": 1.9815119145439606,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006399533625885512,
      "loss": 4.7918,
      "step": 14469
    },
    {
      "epoch": 1.9816488633251164,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000639916224982822,
      "loss": 4.7353,
      "step": 14470
    },
    {
      "epoch": 1.9817858121062724,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000639879085803212,
      "loss": 4.7304,
      "step": 14471
    },
    {
      "epoch": 1.9819227608874281,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006398419450500286,
      "loss": 4.7827,
      "step": 14472
    },
    {
      "epoch": 1.982059709668584,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006398048027235798,
      "loss": 4.7872,
      "step": 14473
    },
    {
      "epoch": 1.9821966584497397,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006397676588241733,
      "loss": 4.7749,
      "step": 14474
    },
    {
      "epoch": 1.9823336072308957,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006397305133521171,
      "loss": 4.8174,
      "step": 14475
    },
    {
      "epoch": 1.9824705560120515,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000639693366307719,
      "loss": 4.7636,
      "step": 14476
    },
    {
      "epoch": 1.9826075047932075,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006396562176912866,
      "loss": 4.8765,
      "step": 14477
    },
    {
      "epoch": 1.9827444535743632,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006396190675031281,
      "loss": 4.7155,
      "step": 14478
    },
    {
      "epoch": 1.982881402355519,
      "grad_norm": 0.359375,
      "learning_rate": 0.000639581915743551,
      "loss": 4.7842,
      "step": 14479
    },
    {
      "epoch": 1.9830183511366748,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006395447624128636,
      "loss": 4.787,
      "step": 14480
    },
    {
      "epoch": 1.9831552999178306,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006395076075113733,
      "loss": 4.7602,
      "step": 14481
    },
    {
      "epoch": 1.9832922486989866,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006394704510393886,
      "loss": 4.761,
      "step": 14482
    },
    {
      "epoch": 1.9834291974801426,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006394332929972171,
      "loss": 4.769,
      "step": 14483
    },
    {
      "epoch": 1.9835661462612983,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006393961333851667,
      "loss": 4.6711,
      "step": 14484
    },
    {
      "epoch": 1.9837030950424541,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006393589722035452,
      "loss": 4.6878,
      "step": 14485
    },
    {
      "epoch": 1.98384004382361,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000639321809452661,
      "loss": 4.7965,
      "step": 14486
    },
    {
      "epoch": 1.9839769926047657,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006392846451328219,
      "loss": 4.8676,
      "step": 14487
    },
    {
      "epoch": 1.9841139413859217,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006392474792443356,
      "loss": 4.7267,
      "step": 14488
    },
    {
      "epoch": 1.9842508901670777,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006392103117875106,
      "loss": 4.8369,
      "step": 14489
    },
    {
      "epoch": 1.9843878389482335,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006391731427626546,
      "loss": 4.753,
      "step": 14490
    },
    {
      "epoch": 1.9845247877293892,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006391359721700755,
      "loss": 4.7203,
      "step": 14491
    },
    {
      "epoch": 1.984661736510545,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006390988000100818,
      "loss": 4.7841,
      "step": 14492
    },
    {
      "epoch": 1.9847986852917008,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006390616262829811,
      "loss": 4.7843,
      "step": 14493
    },
    {
      "epoch": 1.9849356340728568,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006390244509890817,
      "loss": 4.8194,
      "step": 14494
    },
    {
      "epoch": 1.9850725828540126,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006389872741286917,
      "loss": 4.8764,
      "step": 14495
    },
    {
      "epoch": 1.9852095316351686,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006389500957021191,
      "loss": 4.8138,
      "step": 14496
    },
    {
      "epoch": 1.9853464804163243,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006389129157096719,
      "loss": 4.7947,
      "step": 14497
    },
    {
      "epoch": 1.98548342919748,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006388757341516585,
      "loss": 4.7727,
      "step": 14498
    },
    {
      "epoch": 1.9856203779786359,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006388385510283868,
      "loss": 4.8271,
      "step": 14499
    },
    {
      "epoch": 1.9857573267597919,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006388013663401652,
      "loss": 4.8142,
      "step": 14500
    },
    {
      "epoch": 1.9858942755409477,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006387641800873015,
      "loss": 4.8268,
      "step": 14501
    },
    {
      "epoch": 1.9860312243221037,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006387269922701041,
      "loss": 4.8244,
      "step": 14502
    },
    {
      "epoch": 1.9861681731032594,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000638689802888881,
      "loss": 4.7563,
      "step": 14503
    },
    {
      "epoch": 1.9863051218844152,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006386526119439409,
      "loss": 4.8606,
      "step": 14504
    },
    {
      "epoch": 1.986442070665571,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006386154194355913,
      "loss": 4.7859,
      "step": 14505
    },
    {
      "epoch": 1.9865790194467268,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000638578225364141,
      "loss": 4.7921,
      "step": 14506
    },
    {
      "epoch": 1.9867159682278828,
      "grad_norm": 0.359375,
      "learning_rate": 0.000638541029729898,
      "loss": 4.7323,
      "step": 14507
    },
    {
      "epoch": 1.9868529170090388,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006385038325331703,
      "loss": 4.7372,
      "step": 14508
    },
    {
      "epoch": 1.9869898657901945,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006384666337742668,
      "loss": 4.7503,
      "step": 14509
    },
    {
      "epoch": 1.9871268145713503,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006384294334534951,
      "loss": 4.6849,
      "step": 14510
    },
    {
      "epoch": 1.987263763352506,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000638392231571164,
      "loss": 4.7216,
      "step": 14511
    },
    {
      "epoch": 1.9874007121336619,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006383550281275814,
      "loss": 4.813,
      "step": 14512
    },
    {
      "epoch": 1.9875376609148179,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006383178231230559,
      "loss": 4.771,
      "step": 14513
    },
    {
      "epoch": 1.9876746096959739,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006382806165578957,
      "loss": 4.8045,
      "step": 14514
    },
    {
      "epoch": 1.9878115584771296,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006382434084324092,
      "loss": 4.7838,
      "step": 14515
    },
    {
      "epoch": 1.9879485072582854,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006382061987469047,
      "loss": 4.7891,
      "step": 14516
    },
    {
      "epoch": 1.9880854560394412,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0006381689875016905,
      "loss": 4.7938,
      "step": 14517
    },
    {
      "epoch": 1.988222404820597,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006381317746970753,
      "loss": 4.8013,
      "step": 14518
    },
    {
      "epoch": 1.988359353601753,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000638094560333367,
      "loss": 4.7424,
      "step": 14519
    },
    {
      "epoch": 1.9884963023829088,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006380573444108744,
      "loss": 4.7322,
      "step": 14520
    },
    {
      "epoch": 1.9886332511640648,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006380201269299057,
      "loss": 4.8171,
      "step": 14521
    },
    {
      "epoch": 1.9887701999452205,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006379829078907694,
      "loss": 4.8243,
      "step": 14522
    },
    {
      "epoch": 1.9889071487263763,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000637945687293774,
      "loss": 4.7224,
      "step": 14523
    },
    {
      "epoch": 1.989044097507532,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0006379084651392279,
      "loss": 4.7015,
      "step": 14524
    },
    {
      "epoch": 1.989181046288688,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006378712414274396,
      "loss": 4.8059,
      "step": 14525
    },
    {
      "epoch": 1.9893179950698439,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006378340161587175,
      "loss": 4.7236,
      "step": 14526
    },
    {
      "epoch": 1.9894549438509999,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006377967893333701,
      "loss": 4.7466,
      "step": 14527
    },
    {
      "epoch": 1.9895918926321556,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006377595609517061,
      "loss": 4.7596,
      "step": 14528
    },
    {
      "epoch": 1.9897288414133114,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006377223310140337,
      "loss": 4.7824,
      "step": 14529
    },
    {
      "epoch": 1.9898657901944672,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006376850995206617,
      "loss": 4.7934,
      "step": 14530
    },
    {
      "epoch": 1.990002738975623,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006376478664718988,
      "loss": 4.7682,
      "step": 14531
    },
    {
      "epoch": 1.990139687756779,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006376106318680529,
      "loss": 4.8281,
      "step": 14532
    },
    {
      "epoch": 1.990276636537935,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006375733957094332,
      "loss": 4.793,
      "step": 14533
    },
    {
      "epoch": 1.9904135853190907,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006375361579963481,
      "loss": 4.7037,
      "step": 14534
    },
    {
      "epoch": 1.9905505341002465,
      "grad_norm": 0.40625,
      "learning_rate": 0.000637498918729106,
      "loss": 4.7713,
      "step": 14535
    },
    {
      "epoch": 1.9906874828814023,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006374616779080158,
      "loss": 4.7536,
      "step": 14536
    },
    {
      "epoch": 1.990824431662558,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006374244355333861,
      "loss": 4.7453,
      "step": 14537
    },
    {
      "epoch": 1.990961380443714,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006373871916055253,
      "loss": 4.7415,
      "step": 14538
    },
    {
      "epoch": 1.9910983292248698,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006373499461247423,
      "loss": 4.7442,
      "step": 14539
    },
    {
      "epoch": 1.9912352780060258,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006373126990913456,
      "loss": 4.7103,
      "step": 14540
    },
    {
      "epoch": 1.9913722267871816,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006372754505056438,
      "loss": 4.7073,
      "step": 14541
    },
    {
      "epoch": 1.9915091755683374,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006372382003679459,
      "loss": 4.8397,
      "step": 14542
    },
    {
      "epoch": 1.9916461243494932,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006372009486785604,
      "loss": 4.8158,
      "step": 14543
    },
    {
      "epoch": 1.9917830731306492,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006371636954377958,
      "loss": 4.6766,
      "step": 14544
    },
    {
      "epoch": 1.991920021911805,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006371264406459614,
      "loss": 4.7796,
      "step": 14545
    },
    {
      "epoch": 1.992056970692961,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006370891843033654,
      "loss": 4.8605,
      "step": 14546
    },
    {
      "epoch": 1.9921939194741167,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006370519264103169,
      "loss": 4.7198,
      "step": 14547
    },
    {
      "epoch": 1.9923308682552725,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006370146669671243,
      "loss": 4.7433,
      "step": 14548
    },
    {
      "epoch": 1.9924678170364283,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006369774059740967,
      "loss": 4.755,
      "step": 14549
    },
    {
      "epoch": 1.9926047658175843,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006369401434315428,
      "loss": 4.8412,
      "step": 14550
    },
    {
      "epoch": 1.99274171459874,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006369028793397714,
      "loss": 4.7572,
      "step": 14551
    },
    {
      "epoch": 1.992878663379896,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006368656136990912,
      "loss": 4.7285,
      "step": 14552
    },
    {
      "epoch": 1.9930156121610518,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006368283465098112,
      "loss": 4.7421,
      "step": 14553
    },
    {
      "epoch": 1.9931525609422076,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006367910777722401,
      "loss": 4.6708,
      "step": 14554
    },
    {
      "epoch": 1.9932895097233634,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006367538074866869,
      "loss": 4.7581,
      "step": 14555
    },
    {
      "epoch": 1.9934264585045192,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006367165356534605,
      "loss": 4.6754,
      "step": 14556
    },
    {
      "epoch": 1.9935634072856752,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006366792622728696,
      "loss": 4.8248,
      "step": 14557
    },
    {
      "epoch": 1.9937003560668312,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000636641987345223,
      "loss": 4.8156,
      "step": 14558
    },
    {
      "epoch": 1.993837304847987,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006366047108708299,
      "loss": 4.6862,
      "step": 14559
    },
    {
      "epoch": 1.9939742536291427,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006365674328499992,
      "loss": 4.7532,
      "step": 14560
    },
    {
      "epoch": 1.9941112024102985,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006365301532830396,
      "loss": 4.8218,
      "step": 14561
    },
    {
      "epoch": 1.9942481511914543,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006364928721702602,
      "loss": 4.7582,
      "step": 14562
    },
    {
      "epoch": 1.9943850999726103,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00063645558951197,
      "loss": 4.7841,
      "step": 14563
    },
    {
      "epoch": 1.994522048753766,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006364183053084779,
      "loss": 4.7153,
      "step": 14564
    },
    {
      "epoch": 1.994658997534922,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006363810195600927,
      "loss": 4.8332,
      "step": 14565
    },
    {
      "epoch": 1.9947959463160778,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006363437322671237,
      "loss": 4.8335,
      "step": 14566
    },
    {
      "epoch": 1.9949328950972336,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00063630644342988,
      "loss": 4.8165,
      "step": 14567
    },
    {
      "epoch": 1.9950698438783894,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006362691530486699,
      "loss": 4.8551,
      "step": 14568
    },
    {
      "epoch": 1.9952067926595454,
      "grad_norm": 0.375,
      "learning_rate": 0.0006362318611238035,
      "loss": 4.7335,
      "step": 14569
    },
    {
      "epoch": 1.9953437414407011,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006361945676555888,
      "loss": 4.8052,
      "step": 14570
    },
    {
      "epoch": 1.9954806902218571,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006361572726443358,
      "loss": 4.7076,
      "step": 14571
    },
    {
      "epoch": 1.995617639003013,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006361199760903528,
      "loss": 4.8072,
      "step": 14572
    },
    {
      "epoch": 1.9957545877841687,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006360826779939493,
      "loss": 4.6987,
      "step": 14573
    },
    {
      "epoch": 1.9958915365653245,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006360453783554343,
      "loss": 4.7589,
      "step": 14574
    },
    {
      "epoch": 1.9960284853464803,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000636008077175117,
      "loss": 4.7989,
      "step": 14575
    },
    {
      "epoch": 1.9961654341276363,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006359707744533064,
      "loss": 4.7834,
      "step": 14576
    },
    {
      "epoch": 1.9963023829087923,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006359334701903115,
      "loss": 4.8335,
      "step": 14577
    },
    {
      "epoch": 1.996439331689948,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006358961643864419,
      "loss": 4.7481,
      "step": 14578
    },
    {
      "epoch": 1.9965762804711038,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006358588570420063,
      "loss": 4.7058,
      "step": 14579
    },
    {
      "epoch": 1.9967132292522596,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006358215481573142,
      "loss": 4.7646,
      "step": 14580
    },
    {
      "epoch": 1.9968501780334154,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006357842377326746,
      "loss": 4.7175,
      "step": 14581
    },
    {
      "epoch": 1.9969871268145714,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006357469257683967,
      "loss": 4.7534,
      "step": 14582
    },
    {
      "epoch": 1.9971240755957274,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006357096122647898,
      "loss": 4.8223,
      "step": 14583
    },
    {
      "epoch": 1.9972610243768831,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006356722972221631,
      "loss": 4.8713,
      "step": 14584
    },
    {
      "epoch": 1.997397973158039,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006356349806408259,
      "loss": 4.8329,
      "step": 14585
    },
    {
      "epoch": 1.9975349219391947,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006355976625210873,
      "loss": 4.7823,
      "step": 14586
    },
    {
      "epoch": 1.9976718707203505,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006355603428632566,
      "loss": 4.7628,
      "step": 14587
    },
    {
      "epoch": 1.9978088195015065,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006355230216676431,
      "loss": 4.7968,
      "step": 14588
    },
    {
      "epoch": 1.9979457682826622,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006354856989345564,
      "loss": 4.776,
      "step": 14589
    },
    {
      "epoch": 1.9980827170638182,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006354483746643052,
      "loss": 4.7171,
      "step": 14590
    },
    {
      "epoch": 1.998219665844974,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006354110488571991,
      "loss": 4.7553,
      "step": 14591
    },
    {
      "epoch": 1.9983566146261298,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006353737215135476,
      "loss": 4.79,
      "step": 14592
    },
    {
      "epoch": 1.9984935634072856,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006353363926336599,
      "loss": 4.7567,
      "step": 14593
    },
    {
      "epoch": 1.9986305121884416,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006352990622178453,
      "loss": 4.8191,
      "step": 14594
    },
    {
      "epoch": 1.9987674609695973,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006352617302664133,
      "loss": 4.7581,
      "step": 14595
    },
    {
      "epoch": 1.9989044097507533,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006352243967796732,
      "loss": 4.7043,
      "step": 14596
    },
    {
      "epoch": 1.9990413585319091,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006351870617579343,
      "loss": 4.7276,
      "step": 14597
    },
    {
      "epoch": 1.999178307313065,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006351497252015062,
      "loss": 4.7226,
      "step": 14598
    },
    {
      "epoch": 1.9993152560942207,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006351123871106981,
      "loss": 4.8451,
      "step": 14599
    },
    {
      "epoch": 1.9994522048753764,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006350750474858194,
      "loss": 4.7803,
      "step": 14600
    },
    {
      "epoch": 1.9995891536565324,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00063503770632718,
      "loss": 4.775,
      "step": 14601
    },
    {
      "epoch": 1.9997261024376884,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006350003636350889,
      "loss": 4.7381,
      "step": 14602
    },
    {
      "epoch": 1.9998630512188442,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006349630194098557,
      "loss": 4.7288,
      "step": 14603
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006349256736517898,
      "loss": 4.8531,
      "step": 14604
    },
    {
      "epoch": 2.0001369487811558,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000634888326361201,
      "loss": 4.6968,
      "step": 14605
    },
    {
      "epoch": 2.0002738975623116,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006348509775383984,
      "loss": 4.7475,
      "step": 14606
    },
    {
      "epoch": 2.0004108463434673,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006348136271836917,
      "loss": 4.7173,
      "step": 14607
    },
    {
      "epoch": 2.0005477951246236,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006347762752973904,
      "loss": 4.7308,
      "step": 14608
    },
    {
      "epoch": 2.0006847439057793,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006347389218798041,
      "loss": 4.8002,
      "step": 14609
    },
    {
      "epoch": 2.000821692686935,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006347015669312425,
      "loss": 4.7533,
      "step": 14610
    },
    {
      "epoch": 2.000958641468091,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006346642104520148,
      "loss": 4.7463,
      "step": 14611
    },
    {
      "epoch": 2.0010955902492467,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006346268524424309,
      "loss": 4.7324,
      "step": 14612
    },
    {
      "epoch": 2.0012325390304024,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006345894929028002,
      "loss": 4.702,
      "step": 14613
    },
    {
      "epoch": 2.0013694878115587,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006345521318334325,
      "loss": 4.6296,
      "step": 14614
    },
    {
      "epoch": 2.0015064365927144,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006345147692346372,
      "loss": 4.7717,
      "step": 14615
    },
    {
      "epoch": 2.00164338537387,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006344774051067241,
      "loss": 4.7112,
      "step": 14616
    },
    {
      "epoch": 2.001780334155026,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006344400394500028,
      "loss": 4.7529,
      "step": 14617
    },
    {
      "epoch": 2.0019172829361818,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006344026722647829,
      "loss": 4.7575,
      "step": 14618
    },
    {
      "epoch": 2.0020542317173375,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006343653035513742,
      "loss": 4.7496,
      "step": 14619
    },
    {
      "epoch": 2.0021911804984938,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006343279333100861,
      "loss": 4.8108,
      "step": 14620
    },
    {
      "epoch": 2.0023281292796495,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006342905615412287,
      "loss": 4.7719,
      "step": 14621
    },
    {
      "epoch": 2.0024650780608053,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006342531882451113,
      "loss": 4.7429,
      "step": 14622
    },
    {
      "epoch": 2.002602026841961,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006342158134220441,
      "loss": 4.7903,
      "step": 14623
    },
    {
      "epoch": 2.002738975623117,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006341784370723362,
      "loss": 4.7376,
      "step": 14624
    },
    {
      "epoch": 2.0028759244042726,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006341410591962981,
      "loss": 4.6226,
      "step": 14625
    },
    {
      "epoch": 2.003012873185429,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006341036797942389,
      "loss": 4.7693,
      "step": 14626
    },
    {
      "epoch": 2.0031498219665846,
      "grad_norm": 0.375,
      "learning_rate": 0.0006340662988664687,
      "loss": 4.7453,
      "step": 14627
    },
    {
      "epoch": 2.0032867707477404,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006340289164132972,
      "loss": 4.7566,
      "step": 14628
    },
    {
      "epoch": 2.003423719528896,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006339915324350343,
      "loss": 4.7736,
      "step": 14629
    },
    {
      "epoch": 2.003560668310052,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006339541469319896,
      "loss": 4.8543,
      "step": 14630
    },
    {
      "epoch": 2.0036976170912077,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006339167599044732,
      "loss": 4.74,
      "step": 14631
    },
    {
      "epoch": 2.0038345658723635,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006338793713527947,
      "loss": 4.8516,
      "step": 14632
    },
    {
      "epoch": 2.0039715146535197,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000633841981277264,
      "loss": 4.7933,
      "step": 14633
    },
    {
      "epoch": 2.0041084634346755,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000633804589678191,
      "loss": 4.6609,
      "step": 14634
    },
    {
      "epoch": 2.0042454122158313,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006337671965558856,
      "loss": 4.7969,
      "step": 14635
    },
    {
      "epoch": 2.004382360996987,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006337298019106576,
      "loss": 4.7564,
      "step": 14636
    },
    {
      "epoch": 2.004519309778143,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006336924057428171,
      "loss": 4.8057,
      "step": 14637
    },
    {
      "epoch": 2.0046562585592986,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006336550080526738,
      "loss": 4.8032,
      "step": 14638
    },
    {
      "epoch": 2.004793207340455,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006336176088405376,
      "loss": 4.7656,
      "step": 14639
    },
    {
      "epoch": 2.0049301561216106,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006335802081067186,
      "loss": 4.6783,
      "step": 14640
    },
    {
      "epoch": 2.0050671049027664,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006335428058515265,
      "loss": 4.7339,
      "step": 14641
    },
    {
      "epoch": 2.005204053683922,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006335054020752717,
      "loss": 4.8277,
      "step": 14642
    },
    {
      "epoch": 2.005341002465078,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006334679967782638,
      "loss": 4.8233,
      "step": 14643
    },
    {
      "epoch": 2.0054779512462337,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006334305899608129,
      "loss": 4.7061,
      "step": 14644
    },
    {
      "epoch": 2.00561490002739,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006333931816232289,
      "loss": 4.7894,
      "step": 14645
    },
    {
      "epoch": 2.0057518488085457,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000633355771765822,
      "loss": 4.7527,
      "step": 14646
    },
    {
      "epoch": 2.0058887975897015,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006333183603889021,
      "loss": 4.8321,
      "step": 14647
    },
    {
      "epoch": 2.0060257463708573,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006332809474927793,
      "loss": 4.8122,
      "step": 14648
    },
    {
      "epoch": 2.006162695152013,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006332435330777635,
      "loss": 4.7964,
      "step": 14649
    },
    {
      "epoch": 2.006299643933169,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006332061171441649,
      "loss": 4.8314,
      "step": 14650
    },
    {
      "epoch": 2.006436592714325,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006331686996922937,
      "loss": 4.6998,
      "step": 14651
    },
    {
      "epoch": 2.006573541495481,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006331312807224597,
      "loss": 4.7087,
      "step": 14652
    },
    {
      "epoch": 2.0067104902766366,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006330938602349733,
      "loss": 4.7513,
      "step": 14653
    },
    {
      "epoch": 2.0068474390577924,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006330564382301442,
      "loss": 4.8092,
      "step": 14654
    },
    {
      "epoch": 2.006984387838948,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000633019014708283,
      "loss": 4.6881,
      "step": 14655
    },
    {
      "epoch": 2.007121336620104,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006329815896696996,
      "loss": 4.7221,
      "step": 14656
    },
    {
      "epoch": 2.0072582854012597,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000632944163114704,
      "loss": 4.7357,
      "step": 14657
    },
    {
      "epoch": 2.007395234182416,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006329067350436066,
      "loss": 4.7476,
      "step": 14658
    },
    {
      "epoch": 2.0075321829635717,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006328693054567175,
      "loss": 4.7878,
      "step": 14659
    },
    {
      "epoch": 2.0076691317447275,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006328318743543469,
      "loss": 4.658,
      "step": 14660
    },
    {
      "epoch": 2.0078060805258833,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006327944417368051,
      "loss": 4.7319,
      "step": 14661
    },
    {
      "epoch": 2.007943029307039,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000632757007604402,
      "loss": 4.7612,
      "step": 14662
    },
    {
      "epoch": 2.008079978088195,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006327195719574482,
      "loss": 4.6626,
      "step": 14663
    },
    {
      "epoch": 2.008216926869351,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006326821347962537,
      "loss": 4.7508,
      "step": 14664
    },
    {
      "epoch": 2.008353875650507,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006326446961211289,
      "loss": 4.7892,
      "step": 14665
    },
    {
      "epoch": 2.0084908244316626,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006326072559323838,
      "loss": 4.7462,
      "step": 14666
    },
    {
      "epoch": 2.0086277732128184,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006325698142303289,
      "loss": 4.7883,
      "step": 14667
    },
    {
      "epoch": 2.008764721993974,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006325323710152746,
      "loss": 4.807,
      "step": 14668
    },
    {
      "epoch": 2.00890167077513,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006324949262875309,
      "loss": 4.7048,
      "step": 14669
    },
    {
      "epoch": 2.009038619556286,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006324574800474084,
      "loss": 4.7017,
      "step": 14670
    },
    {
      "epoch": 2.009175568337442,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006324200322952171,
      "loss": 4.7019,
      "step": 14671
    },
    {
      "epoch": 2.0093125171185977,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006323825830312677,
      "loss": 4.8047,
      "step": 14672
    },
    {
      "epoch": 2.0094494658997535,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006323451322558703,
      "loss": 4.8082,
      "step": 14673
    },
    {
      "epoch": 2.0095864146809093,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006323076799693352,
      "loss": 4.7522,
      "step": 14674
    },
    {
      "epoch": 2.009723363462065,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006322702261719732,
      "loss": 4.8055,
      "step": 14675
    },
    {
      "epoch": 2.009860312243221,
      "grad_norm": 0.375,
      "learning_rate": 0.0006322327708640942,
      "loss": 4.7797,
      "step": 14676
    },
    {
      "epoch": 2.009997261024377,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006321953140460088,
      "loss": 4.8494,
      "step": 14677
    },
    {
      "epoch": 2.010134209805533,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006321578557180275,
      "loss": 4.7764,
      "step": 14678
    },
    {
      "epoch": 2.0102711585866886,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006321203958804607,
      "loss": 4.7518,
      "step": 14679
    },
    {
      "epoch": 2.0104081073678444,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006320829345336186,
      "loss": 4.7623,
      "step": 14680
    },
    {
      "epoch": 2.010545056149,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000632045471677812,
      "loss": 4.6868,
      "step": 14681
    },
    {
      "epoch": 2.010682004930156,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006320080073133511,
      "loss": 4.715,
      "step": 14682
    },
    {
      "epoch": 2.010818953711312,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006319705414405465,
      "loss": 4.749,
      "step": 14683
    },
    {
      "epoch": 2.010955902492468,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006319330740597087,
      "loss": 4.7858,
      "step": 14684
    },
    {
      "epoch": 2.0110928512736237,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000631895605171148,
      "loss": 4.7441,
      "step": 14685
    },
    {
      "epoch": 2.0112298000547795,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006318581347751753,
      "loss": 4.7861,
      "step": 14686
    },
    {
      "epoch": 2.0113667488359352,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006318206628721008,
      "loss": 4.7168,
      "step": 14687
    },
    {
      "epoch": 2.011503697617091,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006317831894622351,
      "loss": 4.8337,
      "step": 14688
    },
    {
      "epoch": 2.0116406463982472,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006317457145458889,
      "loss": 4.7677,
      "step": 14689
    },
    {
      "epoch": 2.011777595179403,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006317082381233726,
      "loss": 4.7792,
      "step": 14690
    },
    {
      "epoch": 2.011914543960559,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006316707601949969,
      "loss": 4.785,
      "step": 14691
    },
    {
      "epoch": 2.0120514927417146,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006316332807610722,
      "loss": 4.7228,
      "step": 14692
    },
    {
      "epoch": 2.0121884415228704,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006315957998219093,
      "loss": 4.6936,
      "step": 14693
    },
    {
      "epoch": 2.012325390304026,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006315583173778186,
      "loss": 4.8017,
      "step": 14694
    },
    {
      "epoch": 2.0124623390851824,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000631520833429111,
      "loss": 4.7378,
      "step": 14695
    },
    {
      "epoch": 2.012599287866338,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000631483347976097,
      "loss": 4.8024,
      "step": 14696
    },
    {
      "epoch": 2.012736236647494,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006314458610190872,
      "loss": 4.7442,
      "step": 14697
    },
    {
      "epoch": 2.0128731854286497,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006314083725583923,
      "loss": 4.762,
      "step": 14698
    },
    {
      "epoch": 2.0130101342098055,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006313708825943232,
      "loss": 4.7909,
      "step": 14699
    },
    {
      "epoch": 2.0131470829909612,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006313333911271901,
      "loss": 4.8063,
      "step": 14700
    },
    {
      "epoch": 2.013284031772117,
      "grad_norm": 0.375,
      "learning_rate": 0.0006312958981573041,
      "loss": 4.8081,
      "step": 14701
    },
    {
      "epoch": 2.0134209805532732,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006312584036849757,
      "loss": 4.764,
      "step": 14702
    },
    {
      "epoch": 2.013557929334429,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006312209077105158,
      "loss": 4.7189,
      "step": 14703
    },
    {
      "epoch": 2.013694878115585,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000631183410234235,
      "loss": 4.7913,
      "step": 14704
    },
    {
      "epoch": 2.0138318268967406,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006311459112564442,
      "loss": 4.7629,
      "step": 14705
    },
    {
      "epoch": 2.0139687756778963,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006311084107774541,
      "loss": 4.645,
      "step": 14706
    },
    {
      "epoch": 2.014105724459052,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006310709087975752,
      "loss": 4.8173,
      "step": 14707
    },
    {
      "epoch": 2.0142426732402083,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006310334053171188,
      "loss": 4.7083,
      "step": 14708
    },
    {
      "epoch": 2.014379622021364,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006309959003363953,
      "loss": 4.8296,
      "step": 14709
    },
    {
      "epoch": 2.01451657080252,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006309583938557157,
      "loss": 4.7848,
      "step": 14710
    },
    {
      "epoch": 2.0146535195836757,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006309208858753906,
      "loss": 4.726,
      "step": 14711
    },
    {
      "epoch": 2.0147904683648314,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006308833763957313,
      "loss": 4.7076,
      "step": 14712
    },
    {
      "epoch": 2.014927417145987,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006308458654170481,
      "loss": 4.7611,
      "step": 14713
    },
    {
      "epoch": 2.0150643659271434,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006308083529396523,
      "loss": 4.8018,
      "step": 14714
    },
    {
      "epoch": 2.015201314708299,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006307708389638544,
      "loss": 4.8612,
      "step": 14715
    },
    {
      "epoch": 2.015338263489455,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006307333234899657,
      "loss": 4.71,
      "step": 14716
    },
    {
      "epoch": 2.0154752122706108,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006306958065182967,
      "loss": 4.7153,
      "step": 14717
    },
    {
      "epoch": 2.0156121610517665,
      "grad_norm": 0.328125,
      "learning_rate": 0.0006306582880491586,
      "loss": 4.8443,
      "step": 14718
    },
    {
      "epoch": 2.0157491098329223,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006306207680828623,
      "loss": 4.7138,
      "step": 14719
    },
    {
      "epoch": 2.0158860586140785,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006305832466197185,
      "loss": 4.743,
      "step": 14720
    },
    {
      "epoch": 2.0160230073952343,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006305457236600383,
      "loss": 4.7551,
      "step": 14721
    },
    {
      "epoch": 2.01615995617639,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006305081992041328,
      "loss": 4.7156,
      "step": 14722
    },
    {
      "epoch": 2.016296904957546,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006304706732523127,
      "loss": 4.7076,
      "step": 14723
    },
    {
      "epoch": 2.0164338537387017,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006304331458048892,
      "loss": 4.791,
      "step": 14724
    },
    {
      "epoch": 2.0165708025198574,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006303956168621732,
      "loss": 4.7793,
      "step": 14725
    },
    {
      "epoch": 2.016707751301013,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006303580864244758,
      "loss": 4.7755,
      "step": 14726
    },
    {
      "epoch": 2.0168447000821694,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006303205544921079,
      "loss": 4.8031,
      "step": 14727
    },
    {
      "epoch": 2.016981648863325,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006302830210653806,
      "loss": 4.7487,
      "step": 14728
    },
    {
      "epoch": 2.017118597644481,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006302454861446049,
      "loss": 4.7332,
      "step": 14729
    },
    {
      "epoch": 2.0172555464256368,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000630207949730092,
      "loss": 4.739,
      "step": 14730
    },
    {
      "epoch": 2.0173924952067925,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006301704118221529,
      "loss": 4.7611,
      "step": 14731
    },
    {
      "epoch": 2.0175294439879483,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006301328724210985,
      "loss": 4.7956,
      "step": 14732
    },
    {
      "epoch": 2.0176663927691045,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006300953315272402,
      "loss": 4.6837,
      "step": 14733
    },
    {
      "epoch": 2.0178033415502603,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000630057789140889,
      "loss": 4.7254,
      "step": 14734
    },
    {
      "epoch": 2.017940290331416,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006300202452623559,
      "loss": 4.7679,
      "step": 14735
    },
    {
      "epoch": 2.018077239112572,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006299826998919521,
      "loss": 4.8331,
      "step": 14736
    },
    {
      "epoch": 2.0182141878937276,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000629945153029989,
      "loss": 4.7609,
      "step": 14737
    },
    {
      "epoch": 2.0183511366748834,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006299076046767773,
      "loss": 4.7706,
      "step": 14738
    },
    {
      "epoch": 2.0184880854560396,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006298700548326285,
      "loss": 4.7122,
      "step": 14739
    },
    {
      "epoch": 2.0186250342371954,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0006298325034978536,
      "loss": 4.787,
      "step": 14740
    },
    {
      "epoch": 2.018761983018351,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000629794950672764,
      "loss": 4.7187,
      "step": 14741
    },
    {
      "epoch": 2.018898931799507,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006297573963576709,
      "loss": 4.8712,
      "step": 14742
    },
    {
      "epoch": 2.0190358805806627,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0006297198405528852,
      "loss": 4.8428,
      "step": 14743
    },
    {
      "epoch": 2.0191728293618185,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006296822832587186,
      "loss": 4.7216,
      "step": 14744
    },
    {
      "epoch": 2.0193097781429747,
      "grad_norm": 0.46875,
      "learning_rate": 0.0006296447244754818,
      "loss": 4.756,
      "step": 14745
    },
    {
      "epoch": 2.0194467269241305,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006296071642034865,
      "loss": 4.7944,
      "step": 14746
    },
    {
      "epoch": 2.0195836757052863,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000629569602443044,
      "loss": 4.774,
      "step": 14747
    },
    {
      "epoch": 2.019720624486442,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006295320391944652,
      "loss": 4.6786,
      "step": 14748
    },
    {
      "epoch": 2.019857573267598,
      "grad_norm": 0.421875,
      "learning_rate": 0.0006294944744580617,
      "loss": 4.7772,
      "step": 14749
    },
    {
      "epoch": 2.0199945220487536,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006294569082341447,
      "loss": 4.8118,
      "step": 14750
    },
    {
      "epoch": 2.0201314708299094,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006294193405230256,
      "loss": 4.8342,
      "step": 14751
    },
    {
      "epoch": 2.0202684196110656,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006293817713250155,
      "loss": 4.6973,
      "step": 14752
    },
    {
      "epoch": 2.0204053683922214,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006293442006404261,
      "loss": 4.7673,
      "step": 14753
    },
    {
      "epoch": 2.020542317173377,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006293066284695684,
      "loss": 4.8125,
      "step": 14754
    },
    {
      "epoch": 2.020679265954533,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006292690548127542,
      "loss": 4.8225,
      "step": 14755
    },
    {
      "epoch": 2.0208162147356887,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006292314796702944,
      "loss": 4.7454,
      "step": 14756
    },
    {
      "epoch": 2.0209531635168445,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006291939030425007,
      "loss": 4.7823,
      "step": 14757
    },
    {
      "epoch": 2.0210901122980007,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006291563249296845,
      "loss": 4.7652,
      "step": 14758
    },
    {
      "epoch": 2.0212270610791565,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006291187453321569,
      "loss": 4.7349,
      "step": 14759
    },
    {
      "epoch": 2.0213640098603123,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006290811642502298,
      "loss": 4.7395,
      "step": 14760
    },
    {
      "epoch": 2.021500958641468,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006290435816842144,
      "loss": 4.7554,
      "step": 14761
    },
    {
      "epoch": 2.021637907422624,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006290059976344221,
      "loss": 4.7559,
      "step": 14762
    },
    {
      "epoch": 2.0217748562037796,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006289684121011644,
      "loss": 4.7999,
      "step": 14763
    },
    {
      "epoch": 2.021911804984936,
      "grad_norm": 0.326171875,
      "learning_rate": 0.000628930825084753,
      "loss": 4.7699,
      "step": 14764
    },
    {
      "epoch": 2.0220487537660916,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006288932365854991,
      "loss": 4.8496,
      "step": 14765
    },
    {
      "epoch": 2.0221857025472474,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006288556466037144,
      "loss": 4.774,
      "step": 14766
    },
    {
      "epoch": 2.022322651328403,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006288180551397102,
      "loss": 4.7841,
      "step": 14767
    },
    {
      "epoch": 2.022459600109559,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006287804621937982,
      "loss": 4.8301,
      "step": 14768
    },
    {
      "epoch": 2.0225965488907147,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00062874286776629,
      "loss": 4.7713,
      "step": 14769
    },
    {
      "epoch": 2.022733497671871,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006287052718574969,
      "loss": 4.7342,
      "step": 14770
    },
    {
      "epoch": 2.0228704464530267,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006286676744677309,
      "loss": 4.7842,
      "step": 14771
    },
    {
      "epoch": 2.0230073952341825,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006286300755973031,
      "loss": 4.8037,
      "step": 14772
    },
    {
      "epoch": 2.0231443440153383,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006285924752465255,
      "loss": 4.8026,
      "step": 14773
    },
    {
      "epoch": 2.023281292796494,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006285548734157093,
      "loss": 4.7602,
      "step": 14774
    },
    {
      "epoch": 2.02341824157765,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006285172701051663,
      "loss": 4.7092,
      "step": 14775
    },
    {
      "epoch": 2.0235551903588056,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006284796653152082,
      "loss": 4.7618,
      "step": 14776
    },
    {
      "epoch": 2.023692139139962,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006284420590461466,
      "loss": 4.8172,
      "step": 14777
    },
    {
      "epoch": 2.0238290879211176,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006284044512982932,
      "loss": 4.7363,
      "step": 14778
    },
    {
      "epoch": 2.0239660367022734,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006283668420719594,
      "loss": 4.8072,
      "step": 14779
    },
    {
      "epoch": 2.024102985483429,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006283292313674573,
      "loss": 4.7491,
      "step": 14780
    },
    {
      "epoch": 2.024239934264585,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006282916191850982,
      "loss": 4.6817,
      "step": 14781
    },
    {
      "epoch": 2.0243768830457407,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006282540055251941,
      "loss": 4.7826,
      "step": 14782
    },
    {
      "epoch": 2.024513831826897,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006282163903880564,
      "loss": 4.7484,
      "step": 14783
    },
    {
      "epoch": 2.0246507806080527,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006281787737739973,
      "loss": 4.7862,
      "step": 14784
    },
    {
      "epoch": 2.0247877293892085,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000628141155683328,
      "loss": 4.6667,
      "step": 14785
    },
    {
      "epoch": 2.0249246781703643,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006281035361163605,
      "loss": 4.7098,
      "step": 14786
    },
    {
      "epoch": 2.02506162695152,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006280659150734067,
      "loss": 4.8127,
      "step": 14787
    },
    {
      "epoch": 2.025198575732676,
      "grad_norm": 0.390625,
      "learning_rate": 0.000628028292554778,
      "loss": 4.8592,
      "step": 14788
    },
    {
      "epoch": 2.025335524513832,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006279906685607867,
      "loss": 4.7672,
      "step": 14789
    },
    {
      "epoch": 2.025472473294988,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006279530430917441,
      "loss": 4.8063,
      "step": 14790
    },
    {
      "epoch": 2.0256094220761436,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006279154161479624,
      "loss": 4.742,
      "step": 14791
    },
    {
      "epoch": 2.0257463708572994,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006278777877297532,
      "loss": 4.7364,
      "step": 14792
    },
    {
      "epoch": 2.025883319638455,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006278401578374283,
      "loss": 4.7401,
      "step": 14793
    },
    {
      "epoch": 2.026020268419611,
      "grad_norm": 0.375,
      "learning_rate": 0.0006278025264712997,
      "loss": 4.7737,
      "step": 14794
    },
    {
      "epoch": 2.026157217200767,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006277648936316792,
      "loss": 4.7475,
      "step": 14795
    },
    {
      "epoch": 2.026294165981923,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006277272593188786,
      "loss": 4.6858,
      "step": 14796
    },
    {
      "epoch": 2.0264311147630787,
      "grad_norm": 0.375,
      "learning_rate": 0.0006276896235332098,
      "loss": 4.7917,
      "step": 14797
    },
    {
      "epoch": 2.0265680635442345,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000627651986274985,
      "loss": 4.8047,
      "step": 14798
    },
    {
      "epoch": 2.0267050123253902,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006276143475445157,
      "loss": 4.6834,
      "step": 14799
    },
    {
      "epoch": 2.026841961106546,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006275767073421139,
      "loss": 4.8016,
      "step": 14800
    },
    {
      "epoch": 2.026978909887702,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006275390656680918,
      "loss": 4.6743,
      "step": 14801
    },
    {
      "epoch": 2.027115858668858,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0006275014225227611,
      "loss": 4.6181,
      "step": 14802
    },
    {
      "epoch": 2.027252807450014,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0006274637779064337,
      "loss": 4.7547,
      "step": 14803
    },
    {
      "epoch": 2.0273897562311696,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006274261318194219,
      "loss": 4.626,
      "step": 14804
    },
    {
      "epoch": 2.0275267050123253,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006273884842620371,
      "loss": 4.6898,
      "step": 14805
    },
    {
      "epoch": 2.027663653793481,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006273508352345919,
      "loss": 4.7694,
      "step": 14806
    },
    {
      "epoch": 2.027800602574637,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006273131847373981,
      "loss": 4.7051,
      "step": 14807
    },
    {
      "epoch": 2.027937551355793,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006272755327707676,
      "loss": 4.7626,
      "step": 14808
    },
    {
      "epoch": 2.028074500136949,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006272378793350127,
      "loss": 4.6955,
      "step": 14809
    },
    {
      "epoch": 2.0282114489181047,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000627200224430445,
      "loss": 4.6727,
      "step": 14810
    },
    {
      "epoch": 2.0283483976992605,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000627162568057377,
      "loss": 4.7782,
      "step": 14811
    },
    {
      "epoch": 2.0284853464804162,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006271249102161204,
      "loss": 4.736,
      "step": 14812
    },
    {
      "epoch": 2.028622295261572,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006270872509069877,
      "loss": 4.6973,
      "step": 14813
    },
    {
      "epoch": 2.0287592440427282,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006270495901302906,
      "loss": 4.7575,
      "step": 14814
    },
    {
      "epoch": 2.028896192823884,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006270119278863413,
      "loss": 4.7589,
      "step": 14815
    },
    {
      "epoch": 2.02903314160504,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006269742641754522,
      "loss": 4.8512,
      "step": 14816
    },
    {
      "epoch": 2.0291700903861956,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000626936598997935,
      "loss": 4.7649,
      "step": 14817
    },
    {
      "epoch": 2.0293070391673513,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006268989323541021,
      "loss": 4.7144,
      "step": 14818
    },
    {
      "epoch": 2.029443987948507,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006268612642442656,
      "loss": 4.7371,
      "step": 14819
    },
    {
      "epoch": 2.029580936729663,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006268235946687377,
      "loss": 4.7112,
      "step": 14820
    },
    {
      "epoch": 2.029717885510819,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006267859236278305,
      "loss": 4.7852,
      "step": 14821
    },
    {
      "epoch": 2.029854834291975,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006267482511218564,
      "loss": 4.7157,
      "step": 14822
    },
    {
      "epoch": 2.0299917830731307,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006267105771511271,
      "loss": 4.7216,
      "step": 14823
    },
    {
      "epoch": 2.0301287318542864,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006266729017159553,
      "loss": 4.7708,
      "step": 14824
    },
    {
      "epoch": 2.030265680635442,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006266352248166531,
      "loss": 4.7727,
      "step": 14825
    },
    {
      "epoch": 2.030402629416598,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006265975464535326,
      "loss": 4.7874,
      "step": 14826
    },
    {
      "epoch": 2.030539578197754,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006265598666269063,
      "loss": 4.7407,
      "step": 14827
    },
    {
      "epoch": 2.03067652697891,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006265221853370861,
      "loss": 4.7354,
      "step": 14828
    },
    {
      "epoch": 2.0308134757600658,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006264845025843847,
      "loss": 4.7874,
      "step": 14829
    },
    {
      "epoch": 2.0309504245412215,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000626446818369114,
      "loss": 4.7659,
      "step": 14830
    },
    {
      "epoch": 2.0310873733223773,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006264091326915866,
      "loss": 4.7581,
      "step": 14831
    },
    {
      "epoch": 2.031224322103533,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006263714455521146,
      "loss": 4.7465,
      "step": 14832
    },
    {
      "epoch": 2.0313612708846893,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006263337569510104,
      "loss": 4.8101,
      "step": 14833
    },
    {
      "epoch": 2.031498219665845,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006262960668885864,
      "loss": 4.7461,
      "step": 14834
    },
    {
      "epoch": 2.031635168447001,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006262583753651548,
      "loss": 4.7646,
      "step": 14835
    },
    {
      "epoch": 2.0317721172281566,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006262206823810282,
      "loss": 4.7671,
      "step": 14836
    },
    {
      "epoch": 2.0319090660093124,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006261829879365187,
      "loss": 4.8809,
      "step": 14837
    },
    {
      "epoch": 2.032046014790468,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006261452920319389,
      "loss": 4.7425,
      "step": 14838
    },
    {
      "epoch": 2.0321829635716244,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000626107594667601,
      "loss": 4.7772,
      "step": 14839
    },
    {
      "epoch": 2.03231991235278,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006260698958438175,
      "loss": 4.6925,
      "step": 14840
    },
    {
      "epoch": 2.032456861133936,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006260321955609008,
      "loss": 4.7236,
      "step": 14841
    },
    {
      "epoch": 2.0325938099150918,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006259944938191633,
      "loss": 4.7882,
      "step": 14842
    },
    {
      "epoch": 2.0327307586962475,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006259567906189177,
      "loss": 4.8044,
      "step": 14843
    },
    {
      "epoch": 2.0328677074774033,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000625919085960476,
      "loss": 4.8296,
      "step": 14844
    },
    {
      "epoch": 2.033004656258559,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000625881379844151,
      "loss": 4.8124,
      "step": 14845
    },
    {
      "epoch": 2.0331416050397153,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006258436722702551,
      "loss": 4.7634,
      "step": 14846
    },
    {
      "epoch": 2.033278553820871,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006258059632391009,
      "loss": 4.8169,
      "step": 14847
    },
    {
      "epoch": 2.033415502602027,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006257682527510008,
      "loss": 4.776,
      "step": 14848
    },
    {
      "epoch": 2.0335524513831826,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006257305408062671,
      "loss": 4.681,
      "step": 14849
    },
    {
      "epoch": 2.0336894001643384,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006256928274052127,
      "loss": 4.7768,
      "step": 14850
    },
    {
      "epoch": 2.033826348945494,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00062565511254815,
      "loss": 4.7515,
      "step": 14851
    },
    {
      "epoch": 2.0339632977266504,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006256173962353913,
      "loss": 4.7462,
      "step": 14852
    },
    {
      "epoch": 2.034100246507806,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006255796784672495,
      "loss": 4.8329,
      "step": 14853
    },
    {
      "epoch": 2.034237195288962,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006255419592440372,
      "loss": 4.8042,
      "step": 14854
    },
    {
      "epoch": 2.0343741440701177,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006255042385660667,
      "loss": 4.7316,
      "step": 14855
    },
    {
      "epoch": 2.0345110928512735,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006254665164336508,
      "loss": 4.7259,
      "step": 14856
    },
    {
      "epoch": 2.0346480416324293,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006254287928471021,
      "loss": 4.8189,
      "step": 14857
    },
    {
      "epoch": 2.0347849904135855,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006253910678067332,
      "loss": 4.7361,
      "step": 14858
    },
    {
      "epoch": 2.0349219391947413,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006253533413128566,
      "loss": 4.7645,
      "step": 14859
    },
    {
      "epoch": 2.035058887975897,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006253156133657852,
      "loss": 4.7552,
      "step": 14860
    },
    {
      "epoch": 2.035195836757053,
      "grad_norm": 0.375,
      "learning_rate": 0.0006252778839658316,
      "loss": 4.773,
      "step": 14861
    },
    {
      "epoch": 2.0353327855382086,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006252401531133082,
      "loss": 4.7375,
      "step": 14862
    },
    {
      "epoch": 2.0354697343193644,
      "grad_norm": 0.375,
      "learning_rate": 0.0006252024208085282,
      "loss": 4.7251,
      "step": 14863
    },
    {
      "epoch": 2.0356066831005206,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006251646870518037,
      "loss": 4.7436,
      "step": 14864
    },
    {
      "epoch": 2.0357436318816764,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006251269518434477,
      "loss": 4.7594,
      "step": 14865
    },
    {
      "epoch": 2.035880580662832,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000625089215183773,
      "loss": 4.6542,
      "step": 14866
    },
    {
      "epoch": 2.036017529443988,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006250514770730924,
      "loss": 4.8446,
      "step": 14867
    },
    {
      "epoch": 2.0361544782251437,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006250137375117184,
      "loss": 4.7254,
      "step": 14868
    },
    {
      "epoch": 2.0362914270062995,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006249759964999638,
      "loss": 4.7149,
      "step": 14869
    },
    {
      "epoch": 2.0364283757874553,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006249382540381413,
      "loss": 4.8051,
      "step": 14870
    },
    {
      "epoch": 2.0365653245686115,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000624900510126564,
      "loss": 4.7946,
      "step": 14871
    },
    {
      "epoch": 2.0367022733497673,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006248627647655444,
      "loss": 4.7336,
      "step": 14872
    },
    {
      "epoch": 2.036839222130923,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006248250179553955,
      "loss": 4.7595,
      "step": 14873
    },
    {
      "epoch": 2.036976170912079,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006247872696964299,
      "loss": 4.7406,
      "step": 14874
    },
    {
      "epoch": 2.0371131196932346,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006247495199889606,
      "loss": 4.7505,
      "step": 14875
    },
    {
      "epoch": 2.0372500684743904,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006247117688333004,
      "loss": 4.7168,
      "step": 14876
    },
    {
      "epoch": 2.0373870172555466,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006246740162297622,
      "loss": 4.7535,
      "step": 14877
    },
    {
      "epoch": 2.0375239660367024,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006246362621786588,
      "loss": 4.6864,
      "step": 14878
    },
    {
      "epoch": 2.037660914817858,
      "grad_norm": 0.375,
      "learning_rate": 0.0006245985066803029,
      "loss": 4.7313,
      "step": 14879
    },
    {
      "epoch": 2.037797863599014,
      "grad_norm": 0.375,
      "learning_rate": 0.0006245607497350077,
      "loss": 4.7045,
      "step": 14880
    },
    {
      "epoch": 2.0379348123801697,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006245229913430859,
      "loss": 4.7798,
      "step": 14881
    },
    {
      "epoch": 2.0380717611613255,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006244852315048507,
      "loss": 4.8366,
      "step": 14882
    },
    {
      "epoch": 2.0382087099424817,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006244474702206146,
      "loss": 4.7884,
      "step": 14883
    },
    {
      "epoch": 2.0383456587236375,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006244097074906908,
      "loss": 4.7989,
      "step": 14884
    },
    {
      "epoch": 2.0384826075047933,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006243719433153921,
      "loss": 4.7661,
      "step": 14885
    },
    {
      "epoch": 2.038619556285949,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0006243341776950318,
      "loss": 4.6959,
      "step": 14886
    },
    {
      "epoch": 2.038756505067105,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006242964106299224,
      "loss": 4.6995,
      "step": 14887
    },
    {
      "epoch": 2.0388934538482606,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006242586421203772,
      "loss": 4.7289,
      "step": 14888
    },
    {
      "epoch": 2.039030402629417,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006242208721667091,
      "loss": 4.7089,
      "step": 14889
    },
    {
      "epoch": 2.0391673514105726,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000624183100769231,
      "loss": 4.7878,
      "step": 14890
    },
    {
      "epoch": 2.0393043001917284,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006241453279282563,
      "loss": 4.7512,
      "step": 14891
    },
    {
      "epoch": 2.039441248972884,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006241075536440975,
      "loss": 4.6533,
      "step": 14892
    },
    {
      "epoch": 2.03957819775404,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006240697779170682,
      "loss": 4.7735,
      "step": 14893
    },
    {
      "epoch": 2.0397151465351957,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000624032000747481,
      "loss": 4.7334,
      "step": 14894
    },
    {
      "epoch": 2.0398520953163515,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006239942221356491,
      "loss": 4.6883,
      "step": 14895
    },
    {
      "epoch": 2.0399890440975077,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006239564420818857,
      "loss": 4.8109,
      "step": 14896
    },
    {
      "epoch": 2.0401259928786635,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006239186605865038,
      "loss": 4.752,
      "step": 14897
    },
    {
      "epoch": 2.0402629416598193,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006238808776498166,
      "loss": 4.761,
      "step": 14898
    },
    {
      "epoch": 2.040399890440975,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006238430932721369,
      "loss": 4.7419,
      "step": 14899
    },
    {
      "epoch": 2.040536839222131,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006238053074537782,
      "loss": 4.7524,
      "step": 14900
    },
    {
      "epoch": 2.0406737880032866,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006237675201950535,
      "loss": 4.7693,
      "step": 14901
    },
    {
      "epoch": 2.040810736784443,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006237297314962759,
      "loss": 4.7281,
      "step": 14902
    },
    {
      "epoch": 2.0409476855655986,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006236919413577587,
      "loss": 4.7464,
      "step": 14903
    },
    {
      "epoch": 2.0410846343467544,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000623654149779815,
      "loss": 4.7388,
      "step": 14904
    },
    {
      "epoch": 2.04122158312791,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000623616356762758,
      "loss": 4.7676,
      "step": 14905
    },
    {
      "epoch": 2.041358531909066,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006235785623069008,
      "loss": 4.759,
      "step": 14906
    },
    {
      "epoch": 2.0414954806902217,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006235407664125568,
      "loss": 4.7048,
      "step": 14907
    },
    {
      "epoch": 2.041632429471378,
      "grad_norm": 0.375,
      "learning_rate": 0.0006235029690800389,
      "loss": 4.8374,
      "step": 14908
    },
    {
      "epoch": 2.0417693782525337,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006234651703096607,
      "loss": 4.7754,
      "step": 14909
    },
    {
      "epoch": 2.0419063270336895,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006234273701017354,
      "loss": 4.7789,
      "step": 14910
    },
    {
      "epoch": 2.0420432758148452,
      "grad_norm": 0.375,
      "learning_rate": 0.000623389568456576,
      "loss": 4.7102,
      "step": 14911
    },
    {
      "epoch": 2.042180224596001,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006233517653744961,
      "loss": 4.7394,
      "step": 14912
    },
    {
      "epoch": 2.042317173377157,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006233139608558086,
      "loss": 4.6813,
      "step": 14913
    },
    {
      "epoch": 2.042454122158313,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006232761549008272,
      "loss": 4.7619,
      "step": 14914
    },
    {
      "epoch": 2.042591070939469,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006232383475098648,
      "loss": 4.8136,
      "step": 14915
    },
    {
      "epoch": 2.0427280197206246,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000623200538683235,
      "loss": 4.7592,
      "step": 14916
    },
    {
      "epoch": 2.0428649685017803,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006231627284212511,
      "loss": 4.7074,
      "step": 14917
    },
    {
      "epoch": 2.043001917282936,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006231249167242264,
      "loss": 4.7645,
      "step": 14918
    },
    {
      "epoch": 2.043138866064092,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006230871035924744,
      "loss": 4.7422,
      "step": 14919
    },
    {
      "epoch": 2.0432758148452477,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006230492890263081,
      "loss": 4.7804,
      "step": 14920
    },
    {
      "epoch": 2.043412763626404,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006230114730260412,
      "loss": 4.6878,
      "step": 14921
    },
    {
      "epoch": 2.0435497124075597,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000622973655591987,
      "loss": 4.7646,
      "step": 14922
    },
    {
      "epoch": 2.0436866611887154,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006229358367244591,
      "loss": 4.7572,
      "step": 14923
    },
    {
      "epoch": 2.0438236099698712,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006228980164237704,
      "loss": 4.7191,
      "step": 14924
    },
    {
      "epoch": 2.043960558751027,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006228601946902349,
      "loss": 4.7313,
      "step": 14925
    },
    {
      "epoch": 2.044097507532183,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006228223715241655,
      "loss": 4.8167,
      "step": 14926
    },
    {
      "epoch": 2.044234456313339,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000622784546925876,
      "loss": 4.741,
      "step": 14927
    },
    {
      "epoch": 2.0443714050944948,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006227467208956799,
      "loss": 4.6968,
      "step": 14928
    },
    {
      "epoch": 2.0445083538756506,
      "grad_norm": 0.375,
      "learning_rate": 0.0006227088934338905,
      "loss": 4.7933,
      "step": 14929
    },
    {
      "epoch": 2.0446453026568063,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006226710645408214,
      "loss": 4.7706,
      "step": 14930
    },
    {
      "epoch": 2.044782251437962,
      "grad_norm": 0.375,
      "learning_rate": 0.000622633234216786,
      "loss": 4.769,
      "step": 14931
    },
    {
      "epoch": 2.044919200219118,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006225954024620978,
      "loss": 4.8023,
      "step": 14932
    },
    {
      "epoch": 2.045056149000274,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006225575692770704,
      "loss": 4.7093,
      "step": 14933
    },
    {
      "epoch": 2.04519309778143,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006225197346620174,
      "loss": 4.7702,
      "step": 14934
    },
    {
      "epoch": 2.0453300465625857,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006224818986172521,
      "loss": 4.7765,
      "step": 14935
    },
    {
      "epoch": 2.0454669953437414,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006224440611430883,
      "loss": 4.7674,
      "step": 14936
    },
    {
      "epoch": 2.045603944124897,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006224062222398395,
      "loss": 4.8143,
      "step": 14937
    },
    {
      "epoch": 2.045740892906053,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006223683819078193,
      "loss": 4.7317,
      "step": 14938
    },
    {
      "epoch": 2.0458778416872088,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0006223305401473411,
      "loss": 4.6958,
      "step": 14939
    },
    {
      "epoch": 2.046014790468365,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006222926969587187,
      "loss": 4.7314,
      "step": 14940
    },
    {
      "epoch": 2.0461517392495208,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006222548523422657,
      "loss": 4.7964,
      "step": 14941
    },
    {
      "epoch": 2.0462886880306765,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006222170062982956,
      "loss": 4.7631,
      "step": 14942
    },
    {
      "epoch": 2.0464256368118323,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006221791588271223,
      "loss": 4.7745,
      "step": 14943
    },
    {
      "epoch": 2.046562585592988,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006221413099290591,
      "loss": 4.7359,
      "step": 14944
    },
    {
      "epoch": 2.046699534374144,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006221034596044199,
      "loss": 4.7882,
      "step": 14945
    },
    {
      "epoch": 2.0468364831553,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006220656078535185,
      "loss": 4.7614,
      "step": 14946
    },
    {
      "epoch": 2.046973431936456,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006220277546766681,
      "loss": 4.7488,
      "step": 14947
    },
    {
      "epoch": 2.0471103807176116,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000621989900074183,
      "loss": 4.7166,
      "step": 14948
    },
    {
      "epoch": 2.0472473294987674,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006219520440463764,
      "loss": 4.808,
      "step": 14949
    },
    {
      "epoch": 2.047384278279923,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0006219141865935623,
      "loss": 4.7749,
      "step": 14950
    },
    {
      "epoch": 2.047521227061079,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006218763277160543,
      "loss": 4.7145,
      "step": 14951
    },
    {
      "epoch": 2.047658175842235,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006218384674141664,
      "loss": 4.7772,
      "step": 14952
    },
    {
      "epoch": 2.047795124623391,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006218006056882121,
      "loss": 4.8523,
      "step": 14953
    },
    {
      "epoch": 2.0479320734045467,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006217627425385051,
      "loss": 4.6668,
      "step": 14954
    },
    {
      "epoch": 2.0480690221857025,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006217248779653595,
      "loss": 4.6456,
      "step": 14955
    },
    {
      "epoch": 2.0482059709668583,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006216870119690888,
      "loss": 4.8403,
      "step": 14956
    },
    {
      "epoch": 2.048342919748014,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006216491445500071,
      "loss": 4.7623,
      "step": 14957
    },
    {
      "epoch": 2.0484798685291703,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006216112757084279,
      "loss": 4.8,
      "step": 14958
    },
    {
      "epoch": 2.048616817310326,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006215734054446651,
      "loss": 4.7551,
      "step": 14959
    },
    {
      "epoch": 2.048753766091482,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006215355337590327,
      "loss": 4.71,
      "step": 14960
    },
    {
      "epoch": 2.0488907148726376,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006214976606518445,
      "loss": 4.8126,
      "step": 14961
    },
    {
      "epoch": 2.0490276636537934,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006214597861234142,
      "loss": 4.6729,
      "step": 14962
    },
    {
      "epoch": 2.049164612434949,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000621421910174056,
      "loss": 4.7485,
      "step": 14963
    },
    {
      "epoch": 2.049301561216105,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006213840328040833,
      "loss": 4.7302,
      "step": 14964
    },
    {
      "epoch": 2.049438509997261,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006213461540138104,
      "loss": 4.8395,
      "step": 14965
    },
    {
      "epoch": 2.049575458778417,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006213082738035512,
      "loss": 4.7552,
      "step": 14966
    },
    {
      "epoch": 2.0497124075595727,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006212703921736194,
      "loss": 4.7386,
      "step": 14967
    },
    {
      "epoch": 2.0498493563407285,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006212325091243292,
      "loss": 4.802,
      "step": 14968
    },
    {
      "epoch": 2.0499863051218843,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006211946246559942,
      "loss": 4.7348,
      "step": 14969
    },
    {
      "epoch": 2.05012325390304,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006211567387689286,
      "loss": 4.7925,
      "step": 14970
    },
    {
      "epoch": 2.0502602026841963,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006211188514634463,
      "loss": 4.6401,
      "step": 14971
    },
    {
      "epoch": 2.050397151465352,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006210809627398613,
      "loss": 4.7255,
      "step": 14972
    },
    {
      "epoch": 2.050534100246508,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006210430725984877,
      "loss": 4.7027,
      "step": 14973
    },
    {
      "epoch": 2.0506710490276636,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006210051810396393,
      "loss": 4.7213,
      "step": 14974
    },
    {
      "epoch": 2.0508079978088194,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006209672880636304,
      "loss": 4.6725,
      "step": 14975
    },
    {
      "epoch": 2.050944946589975,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006209293936707747,
      "loss": 4.7422,
      "step": 14976
    },
    {
      "epoch": 2.0510818953711314,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006208914978613864,
      "loss": 4.765,
      "step": 14977
    },
    {
      "epoch": 2.051218844152287,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006208536006357795,
      "loss": 4.7755,
      "step": 14978
    },
    {
      "epoch": 2.051355792933443,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006208157019942681,
      "loss": 4.785,
      "step": 14979
    },
    {
      "epoch": 2.0514927417145987,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006207778019371663,
      "loss": 4.7107,
      "step": 14980
    },
    {
      "epoch": 2.0516296904957545,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006207399004647882,
      "loss": 4.7528,
      "step": 14981
    },
    {
      "epoch": 2.0517666392769103,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006207019975774477,
      "loss": 4.854,
      "step": 14982
    },
    {
      "epoch": 2.0519035880580665,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006206640932754592,
      "loss": 4.7553,
      "step": 14983
    },
    {
      "epoch": 2.0520405368392223,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006206261875591367,
      "loss": 4.7331,
      "step": 14984
    },
    {
      "epoch": 2.052177485620378,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006205882804287943,
      "loss": 4.809,
      "step": 14985
    },
    {
      "epoch": 2.052314434401534,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006205503718847462,
      "loss": 4.8072,
      "step": 14986
    },
    {
      "epoch": 2.0524513831826896,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006205124619273066,
      "loss": 4.7953,
      "step": 14987
    },
    {
      "epoch": 2.0525883319638454,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006204745505567895,
      "loss": 4.7803,
      "step": 14988
    },
    {
      "epoch": 2.052725280745001,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006204366377735091,
      "loss": 4.7397,
      "step": 14989
    },
    {
      "epoch": 2.0528622295261574,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006203987235777798,
      "loss": 4.7934,
      "step": 14990
    },
    {
      "epoch": 2.052999178307313,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006203608079699156,
      "loss": 4.8663,
      "step": 14991
    },
    {
      "epoch": 2.053136127088469,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006203228909502308,
      "loss": 4.8211,
      "step": 14992
    },
    {
      "epoch": 2.0532730758696247,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006202849725190396,
      "loss": 4.8354,
      "step": 14993
    },
    {
      "epoch": 2.0534100246507805,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006202470526766562,
      "loss": 4.81,
      "step": 14994
    },
    {
      "epoch": 2.0535469734319363,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006202091314233951,
      "loss": 4.7407,
      "step": 14995
    },
    {
      "epoch": 2.0536839222130925,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006201712087595703,
      "loss": 4.6984,
      "step": 14996
    },
    {
      "epoch": 2.0538208709942483,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006201332846854961,
      "loss": 4.8323,
      "step": 14997
    },
    {
      "epoch": 2.053957819775404,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006200953592014868,
      "loss": 4.6582,
      "step": 14998
    },
    {
      "epoch": 2.05409476855656,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006200574323078569,
      "loss": 4.7437,
      "step": 14999
    },
    {
      "epoch": 2.0542317173377156,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006200195040049203,
      "loss": 4.6915,
      "step": 15000
    },
    {
      "epoch": 2.0543686661188714,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006199815742929917,
      "loss": 4.7062,
      "step": 15001
    },
    {
      "epoch": 2.0545056149000276,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006199436431723851,
      "loss": 4.7969,
      "step": 15002
    },
    {
      "epoch": 2.0546425636811834,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006199057106434152,
      "loss": 4.7935,
      "step": 15003
    },
    {
      "epoch": 2.054779512462339,
      "grad_norm": 0.375,
      "learning_rate": 0.0006198677767063963,
      "loss": 4.7119,
      "step": 15004
    },
    {
      "epoch": 2.054916461243495,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006198298413616424,
      "loss": 4.7852,
      "step": 15005
    },
    {
      "epoch": 2.0550534100246507,
      "grad_norm": 0.375,
      "learning_rate": 0.0006197919046094683,
      "loss": 4.6707,
      "step": 15006
    },
    {
      "epoch": 2.0551903588058065,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006197539664501882,
      "loss": 4.8141,
      "step": 15007
    },
    {
      "epoch": 2.0553273075869627,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006197160268841165,
      "loss": 4.7321,
      "step": 15008
    },
    {
      "epoch": 2.0554642563681185,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006196780859115677,
      "loss": 4.7062,
      "step": 15009
    },
    {
      "epoch": 2.0556012051492742,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000619640143532856,
      "loss": 4.7811,
      "step": 15010
    },
    {
      "epoch": 2.05573815393043,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006196021997482963,
      "loss": 4.7752,
      "step": 15011
    },
    {
      "epoch": 2.055875102711586,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006195642545582025,
      "loss": 4.7198,
      "step": 15012
    },
    {
      "epoch": 2.0560120514927416,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006195263079628895,
      "loss": 4.7601,
      "step": 15013
    },
    {
      "epoch": 2.0561490002738974,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006194883599626715,
      "loss": 4.8204,
      "step": 15014
    },
    {
      "epoch": 2.0562859490550536,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006194504105578631,
      "loss": 4.7704,
      "step": 15015
    },
    {
      "epoch": 2.0564228978362094,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006194124597487787,
      "loss": 4.8299,
      "step": 15016
    },
    {
      "epoch": 2.056559846617365,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000619374507535733,
      "loss": 4.7784,
      "step": 15017
    },
    {
      "epoch": 2.056696795398521,
      "grad_norm": 0.375,
      "learning_rate": 0.0006193365539190403,
      "loss": 4.7277,
      "step": 15018
    },
    {
      "epoch": 2.0568337441796767,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006192985988990152,
      "loss": 4.6721,
      "step": 15019
    },
    {
      "epoch": 2.0569706929608325,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006192606424759723,
      "loss": 4.846,
      "step": 15020
    },
    {
      "epoch": 2.0571076417419887,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006192226846502262,
      "loss": 4.8065,
      "step": 15021
    },
    {
      "epoch": 2.0572445905231445,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006191847254220914,
      "loss": 4.6701,
      "step": 15022
    },
    {
      "epoch": 2.0573815393043002,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006191467647918823,
      "loss": 4.686,
      "step": 15023
    },
    {
      "epoch": 2.057518488085456,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006191088027599138,
      "loss": 4.7978,
      "step": 15024
    },
    {
      "epoch": 2.057655436866612,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006190708393265002,
      "loss": 4.7501,
      "step": 15025
    },
    {
      "epoch": 2.0577923856477676,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006190328744919564,
      "loss": 4.7797,
      "step": 15026
    },
    {
      "epoch": 2.057929334428924,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006189949082565969,
      "loss": 4.7126,
      "step": 15027
    },
    {
      "epoch": 2.0580662832100796,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006189569406207363,
      "loss": 4.7724,
      "step": 15028
    },
    {
      "epoch": 2.0582032319912353,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006189189715846893,
      "loss": 4.7761,
      "step": 15029
    },
    {
      "epoch": 2.058340180772391,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006188810011487704,
      "loss": 4.7587,
      "step": 15030
    },
    {
      "epoch": 2.058477129553547,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006188430293132946,
      "loss": 4.8422,
      "step": 15031
    },
    {
      "epoch": 2.0586140783347027,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006188050560785762,
      "loss": 4.8055,
      "step": 15032
    },
    {
      "epoch": 2.058751027115859,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006187670814449302,
      "loss": 4.8007,
      "step": 15033
    },
    {
      "epoch": 2.0588879758970147,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000618729105412671,
      "loss": 4.7146,
      "step": 15034
    },
    {
      "epoch": 2.0590249246781704,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006186911279821138,
      "loss": 4.8141,
      "step": 15035
    },
    {
      "epoch": 2.059161873459326,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006186531491535729,
      "loss": 4.7652,
      "step": 15036
    },
    {
      "epoch": 2.059298822240482,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006186151689273631,
      "loss": 4.7478,
      "step": 15037
    },
    {
      "epoch": 2.0594357710216378,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006185771873037991,
      "loss": 4.8127,
      "step": 15038
    },
    {
      "epoch": 2.0595727198027936,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006185392042831959,
      "loss": 4.7465,
      "step": 15039
    },
    {
      "epoch": 2.0597096685839498,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006185012198658683,
      "loss": 4.7355,
      "step": 15040
    },
    {
      "epoch": 2.0598466173651055,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006184632340521307,
      "loss": 4.8286,
      "step": 15041
    },
    {
      "epoch": 2.0599835661462613,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006184252468422984,
      "loss": 4.8029,
      "step": 15042
    },
    {
      "epoch": 2.060120514927417,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006183872582366856,
      "loss": 4.7395,
      "step": 15043
    },
    {
      "epoch": 2.060257463708573,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006183492682356078,
      "loss": 4.8076,
      "step": 15044
    },
    {
      "epoch": 2.0603944124897287,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006183112768393793,
      "loss": 4.7396,
      "step": 15045
    },
    {
      "epoch": 2.060531361270885,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006182732840483153,
      "loss": 4.7753,
      "step": 15046
    },
    {
      "epoch": 2.0606683100520407,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006182352898627304,
      "loss": 4.754,
      "step": 15047
    },
    {
      "epoch": 2.0608052588331964,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006181972942829395,
      "loss": 4.8196,
      "step": 15048
    },
    {
      "epoch": 2.060942207614352,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006181592973092576,
      "loss": 4.7266,
      "step": 15049
    },
    {
      "epoch": 2.061079156395508,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006181212989419995,
      "loss": 4.7195,
      "step": 15050
    },
    {
      "epoch": 2.0612161051766638,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006180832991814802,
      "loss": 4.6519,
      "step": 15051
    },
    {
      "epoch": 2.06135305395782,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006180452980280145,
      "loss": 4.7308,
      "step": 15052
    },
    {
      "epoch": 2.0614900027389758,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006180072954819174,
      "loss": 4.8049,
      "step": 15053
    },
    {
      "epoch": 2.0616269515201315,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006179692915435038,
      "loss": 4.7204,
      "step": 15054
    },
    {
      "epoch": 2.0617639003012873,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006179312862130888,
      "loss": 4.7746,
      "step": 15055
    },
    {
      "epoch": 2.061900849082443,
      "grad_norm": 0.40625,
      "learning_rate": 0.000617893279490987,
      "loss": 4.7611,
      "step": 15056
    },
    {
      "epoch": 2.062037797863599,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006178552713775136,
      "loss": 4.6919,
      "step": 15057
    },
    {
      "epoch": 2.0621747466447546,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006178172618729837,
      "loss": 4.7672,
      "step": 15058
    },
    {
      "epoch": 2.062311695425911,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006177792509777121,
      "loss": 4.7831,
      "step": 15059
    },
    {
      "epoch": 2.0624486442070666,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006177412386920138,
      "loss": 4.727,
      "step": 15060
    },
    {
      "epoch": 2.0625855929882224,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000617703225016204,
      "loss": 4.7896,
      "step": 15061
    },
    {
      "epoch": 2.062722541769378,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006176652099505976,
      "loss": 4.7797,
      "step": 15062
    },
    {
      "epoch": 2.062859490550534,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006176271934955096,
      "loss": 4.7187,
      "step": 15063
    },
    {
      "epoch": 2.0629964393316897,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006175891756512551,
      "loss": 4.8193,
      "step": 15064
    },
    {
      "epoch": 2.063133388112846,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006175511564181492,
      "loss": 4.7236,
      "step": 15065
    },
    {
      "epoch": 2.0632703368940017,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006175131357965068,
      "loss": 4.7927,
      "step": 15066
    },
    {
      "epoch": 2.0634072856751575,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006174751137866433,
      "loss": 4.7931,
      "step": 15067
    },
    {
      "epoch": 2.0635442344563133,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006174370903888736,
      "loss": 4.6837,
      "step": 15068
    },
    {
      "epoch": 2.063681183237469,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006173990656035128,
      "loss": 4.7975,
      "step": 15069
    },
    {
      "epoch": 2.063818132018625,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006173610394308761,
      "loss": 4.7039,
      "step": 15070
    },
    {
      "epoch": 2.063955080799781,
      "grad_norm": 0.375,
      "learning_rate": 0.0006173230118712786,
      "loss": 4.7922,
      "step": 15071
    },
    {
      "epoch": 2.064092029580937,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006172849829250353,
      "loss": 4.7437,
      "step": 15072
    },
    {
      "epoch": 2.0642289783620926,
      "grad_norm": 0.328125,
      "learning_rate": 0.0006172469525924616,
      "loss": 4.8078,
      "step": 15073
    },
    {
      "epoch": 2.0643659271432484,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006172089208738724,
      "loss": 4.7911,
      "step": 15074
    },
    {
      "epoch": 2.064502875924404,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006171708877695832,
      "loss": 4.7528,
      "step": 15075
    },
    {
      "epoch": 2.06463982470556,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000617132853279909,
      "loss": 4.7658,
      "step": 15076
    },
    {
      "epoch": 2.064776773486716,
      "grad_norm": 0.375,
      "learning_rate": 0.0006170948174051649,
      "loss": 4.8134,
      "step": 15077
    },
    {
      "epoch": 2.064913722267872,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006170567801456663,
      "loss": 4.7236,
      "step": 15078
    },
    {
      "epoch": 2.0650506710490277,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006170187415017284,
      "loss": 4.8161,
      "step": 15079
    },
    {
      "epoch": 2.0651876198301835,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006169807014736664,
      "loss": 4.7396,
      "step": 15080
    },
    {
      "epoch": 2.0653245686113393,
      "grad_norm": 0.375,
      "learning_rate": 0.0006169426600617955,
      "loss": 4.7461,
      "step": 15081
    },
    {
      "epoch": 2.065461517392495,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000616904617266431,
      "loss": 4.8093,
      "step": 15082
    },
    {
      "epoch": 2.0655984661736513,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006168665730878882,
      "loss": 4.6712,
      "step": 15083
    },
    {
      "epoch": 2.065735414954807,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006168285275264823,
      "loss": 4.7617,
      "step": 15084
    },
    {
      "epoch": 2.065872363735963,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006167904805825286,
      "loss": 4.7616,
      "step": 15085
    },
    {
      "epoch": 2.0660093125171186,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006167524322563424,
      "loss": 4.824,
      "step": 15086
    },
    {
      "epoch": 2.0661462612982744,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006167143825482393,
      "loss": 4.7445,
      "step": 15087
    },
    {
      "epoch": 2.06628321007943,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006166763314585342,
      "loss": 4.7523,
      "step": 15088
    },
    {
      "epoch": 2.066420158860586,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006166382789875426,
      "loss": 4.7944,
      "step": 15089
    },
    {
      "epoch": 2.066557107641742,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006166002251355799,
      "loss": 4.7752,
      "step": 15090
    },
    {
      "epoch": 2.066694056422898,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006165621699029616,
      "loss": 4.8019,
      "step": 15091
    },
    {
      "epoch": 2.0668310052040537,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006165241132900026,
      "loss": 4.7593,
      "step": 15092
    },
    {
      "epoch": 2.0669679539852095,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006164860552970189,
      "loss": 4.7234,
      "step": 15093
    },
    {
      "epoch": 2.0671049027663653,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006164479959243254,
      "loss": 4.7815,
      "step": 15094
    },
    {
      "epoch": 2.067241851547521,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006164099351722377,
      "loss": 4.7937,
      "step": 15095
    },
    {
      "epoch": 2.0673788003286773,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006163718730410713,
      "loss": 4.745,
      "step": 15096
    },
    {
      "epoch": 2.067515749109833,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006163338095311415,
      "loss": 4.8152,
      "step": 15097
    },
    {
      "epoch": 2.067652697890989,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006162957446427638,
      "loss": 4.7223,
      "step": 15098
    },
    {
      "epoch": 2.0677896466721446,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006162576783762537,
      "loss": 4.7731,
      "step": 15099
    },
    {
      "epoch": 2.0679265954533004,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006162196107319265,
      "loss": 4.7206,
      "step": 15100
    },
    {
      "epoch": 2.068063544234456,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006161815417100978,
      "loss": 4.7486,
      "step": 15101
    },
    {
      "epoch": 2.0682004930156124,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000616143471311083,
      "loss": 4.7776,
      "step": 15102
    },
    {
      "epoch": 2.068337441796768,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006161053995351977,
      "loss": 4.6971,
      "step": 15103
    },
    {
      "epoch": 2.068474390577924,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006160673263827575,
      "loss": 4.7095,
      "step": 15104
    },
    {
      "epoch": 2.0686113393590797,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006160292518540778,
      "loss": 4.8652,
      "step": 15105
    },
    {
      "epoch": 2.0687482881402355,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006159911759494739,
      "loss": 4.7869,
      "step": 15106
    },
    {
      "epoch": 2.0688852369213913,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006159530986692618,
      "loss": 4.6792,
      "step": 15107
    },
    {
      "epoch": 2.069022185702547,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006159150200137567,
      "loss": 4.7149,
      "step": 15108
    },
    {
      "epoch": 2.0691591344837033,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006158769399832743,
      "loss": 4.735,
      "step": 15109
    },
    {
      "epoch": 2.069296083264859,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006158388585781303,
      "loss": 4.7825,
      "step": 15110
    },
    {
      "epoch": 2.069433032046015,
      "grad_norm": 0.390625,
      "learning_rate": 0.00061580077579864,
      "loss": 4.7118,
      "step": 15111
    },
    {
      "epoch": 2.0695699808271706,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006157626916451192,
      "loss": 4.6732,
      "step": 15112
    },
    {
      "epoch": 2.0697069296083264,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006157246061178834,
      "loss": 4.7538,
      "step": 15113
    },
    {
      "epoch": 2.069843878389482,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006156865192172485,
      "loss": 4.7526,
      "step": 15114
    },
    {
      "epoch": 2.0699808271706384,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006156484309435298,
      "loss": 4.7197,
      "step": 15115
    },
    {
      "epoch": 2.070117775951794,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006156103412970431,
      "loss": 4.7905,
      "step": 15116
    },
    {
      "epoch": 2.07025472473295,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006155722502781041,
      "loss": 4.8606,
      "step": 15117
    },
    {
      "epoch": 2.0703916735141057,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006155341578870284,
      "loss": 4.7699,
      "step": 15118
    },
    {
      "epoch": 2.0705286222952615,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006154960641241317,
      "loss": 4.7789,
      "step": 15119
    },
    {
      "epoch": 2.0706655710764172,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006154579689897294,
      "loss": 4.7165,
      "step": 15120
    },
    {
      "epoch": 2.0708025198575735,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006154198724841379,
      "loss": 4.7007,
      "step": 15121
    },
    {
      "epoch": 2.0709394686387292,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006153817746076723,
      "loss": 4.7453,
      "step": 15122
    },
    {
      "epoch": 2.071076417419885,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006153436753606484,
      "loss": 4.8082,
      "step": 15123
    },
    {
      "epoch": 2.071213366201041,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006153055747433822,
      "loss": 4.7526,
      "step": 15124
    },
    {
      "epoch": 2.0713503149821966,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006152674727561893,
      "loss": 4.78,
      "step": 15125
    },
    {
      "epoch": 2.0714872637633523,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006152293693993856,
      "loss": 4.6478,
      "step": 15126
    },
    {
      "epoch": 2.071624212544508,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006151912646732866,
      "loss": 4.8032,
      "step": 15127
    },
    {
      "epoch": 2.0717611613256643,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006151531585782084,
      "loss": 4.71,
      "step": 15128
    },
    {
      "epoch": 2.07189811010682,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006151150511144663,
      "loss": 4.7901,
      "step": 15129
    },
    {
      "epoch": 2.072035058887976,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006150769422823766,
      "loss": 4.7911,
      "step": 15130
    },
    {
      "epoch": 2.0721720076691317,
      "grad_norm": 0.375,
      "learning_rate": 0.000615038832082255,
      "loss": 4.7225,
      "step": 15131
    },
    {
      "epoch": 2.0723089564502875,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006150007205144172,
      "loss": 4.7326,
      "step": 15132
    },
    {
      "epoch": 2.0724459052314432,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006149626075791791,
      "loss": 4.726,
      "step": 15133
    },
    {
      "epoch": 2.0725828540125995,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006149244932768566,
      "loss": 4.7794,
      "step": 15134
    },
    {
      "epoch": 2.0727198027937552,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006148863776077655,
      "loss": 4.6166,
      "step": 15135
    },
    {
      "epoch": 2.072856751574911,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006148482605722218,
      "loss": 4.8493,
      "step": 15136
    },
    {
      "epoch": 2.072993700356067,
      "grad_norm": 0.375,
      "learning_rate": 0.0006148101421705411,
      "loss": 4.7386,
      "step": 15137
    },
    {
      "epoch": 2.0731306491372226,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006147720224030397,
      "loss": 4.7515,
      "step": 15138
    },
    {
      "epoch": 2.0732675979183783,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000614733901270033,
      "loss": 4.7667,
      "step": 15139
    },
    {
      "epoch": 2.0734045466995346,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006146957787718376,
      "loss": 4.7475,
      "step": 15140
    },
    {
      "epoch": 2.0735414954806903,
      "grad_norm": 0.375,
      "learning_rate": 0.0006146576549087687,
      "loss": 4.7363,
      "step": 15141
    },
    {
      "epoch": 2.073678444261846,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006146195296811426,
      "loss": 4.7575,
      "step": 15142
    },
    {
      "epoch": 2.073815393043002,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006145814030892754,
      "loss": 4.7831,
      "step": 15143
    },
    {
      "epoch": 2.0739523418241577,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006145432751334828,
      "loss": 4.7668,
      "step": 15144
    },
    {
      "epoch": 2.0740892906053134,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000614505145814081,
      "loss": 4.6918,
      "step": 15145
    },
    {
      "epoch": 2.0742262393864697,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006144670151313857,
      "loss": 4.7551,
      "step": 15146
    },
    {
      "epoch": 2.0743631881676254,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006144288830857131,
      "loss": 4.7901,
      "step": 15147
    },
    {
      "epoch": 2.074500136948781,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006143907496773791,
      "loss": 4.7287,
      "step": 15148
    },
    {
      "epoch": 2.074637085729937,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006143526149066999,
      "loss": 4.6835,
      "step": 15149
    },
    {
      "epoch": 2.0747740345110928,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006143144787739914,
      "loss": 4.7102,
      "step": 15150
    },
    {
      "epoch": 2.0749109832922485,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006142763412795695,
      "loss": 4.8351,
      "step": 15151
    },
    {
      "epoch": 2.0750479320734048,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006142382024237505,
      "loss": 4.722,
      "step": 15152
    },
    {
      "epoch": 2.0751848808545605,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006142000622068504,
      "loss": 4.7698,
      "step": 15153
    },
    {
      "epoch": 2.0753218296357163,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006141619206291854,
      "loss": 4.6998,
      "step": 15154
    },
    {
      "epoch": 2.075458778416872,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006141237776910713,
      "loss": 4.7434,
      "step": 15155
    },
    {
      "epoch": 2.075595727198028,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006140856333928244,
      "loss": 4.771,
      "step": 15156
    },
    {
      "epoch": 2.0757326759791837,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006140474877347607,
      "loss": 4.686,
      "step": 15157
    },
    {
      "epoch": 2.0758696247603394,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006140093407171964,
      "loss": 4.7631,
      "step": 15158
    },
    {
      "epoch": 2.0760065735414956,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006139711923404477,
      "loss": 4.7835,
      "step": 15159
    },
    {
      "epoch": 2.0761435223226514,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006139330426048305,
      "loss": 4.7357,
      "step": 15160
    },
    {
      "epoch": 2.076280471103807,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006138948915106612,
      "loss": 4.8102,
      "step": 15161
    },
    {
      "epoch": 2.076417419884963,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006138567390582559,
      "loss": 4.736,
      "step": 15162
    },
    {
      "epoch": 2.0765543686661188,
      "grad_norm": 0.375,
      "learning_rate": 0.0006138185852479308,
      "loss": 4.7215,
      "step": 15163
    },
    {
      "epoch": 2.0766913174472745,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006137804300800019,
      "loss": 4.6972,
      "step": 15164
    },
    {
      "epoch": 2.0768282662284308,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006137422735547856,
      "loss": 4.7735,
      "step": 15165
    },
    {
      "epoch": 2.0769652150095865,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000613704115672598,
      "loss": 4.7869,
      "step": 15166
    },
    {
      "epoch": 2.0771021637907423,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006136659564337557,
      "loss": 4.7906,
      "step": 15167
    },
    {
      "epoch": 2.077239112571898,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006136277958385742,
      "loss": 4.7827,
      "step": 15168
    },
    {
      "epoch": 2.077376061353054,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006135896338873703,
      "loss": 4.6655,
      "step": 15169
    },
    {
      "epoch": 2.0775130101342096,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006135514705804603,
      "loss": 4.7336,
      "step": 15170
    },
    {
      "epoch": 2.077649958915366,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00061351330591816,
      "loss": 4.8596,
      "step": 15171
    },
    {
      "epoch": 2.0777869076965216,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006134751399007862,
      "loss": 4.7235,
      "step": 15172
    },
    {
      "epoch": 2.0779238564776774,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006134369725286549,
      "loss": 4.8116,
      "step": 15173
    },
    {
      "epoch": 2.078060805258833,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006133988038020825,
      "loss": 4.7477,
      "step": 15174
    },
    {
      "epoch": 2.078197754039989,
      "grad_norm": 0.375,
      "learning_rate": 0.0006133606337213852,
      "loss": 4.7807,
      "step": 15175
    },
    {
      "epoch": 2.0783347028211447,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006133224622868796,
      "loss": 4.7224,
      "step": 15176
    },
    {
      "epoch": 2.0784716516023005,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006132842894988816,
      "loss": 4.7585,
      "step": 15177
    },
    {
      "epoch": 2.0786086003834567,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006132461153577079,
      "loss": 4.7278,
      "step": 15178
    },
    {
      "epoch": 2.0787455491646125,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006132079398636747,
      "loss": 4.7663,
      "step": 15179
    },
    {
      "epoch": 2.0788824979457683,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006131697630170984,
      "loss": 4.782,
      "step": 15180
    },
    {
      "epoch": 2.079019446726924,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006131315848182954,
      "loss": 4.6827,
      "step": 15181
    },
    {
      "epoch": 2.07915639550808,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006130934052675821,
      "loss": 4.7901,
      "step": 15182
    },
    {
      "epoch": 2.0792933442892356,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006130552243652749,
      "loss": 4.8856,
      "step": 15183
    },
    {
      "epoch": 2.079430293070392,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006130170421116901,
      "loss": 4.7826,
      "step": 15184
    },
    {
      "epoch": 2.0795672418515476,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006129788585071444,
      "loss": 4.7638,
      "step": 15185
    },
    {
      "epoch": 2.0797041906327034,
      "grad_norm": 0.359375,
      "learning_rate": 0.000612940673551954,
      "loss": 4.7686,
      "step": 15186
    },
    {
      "epoch": 2.079841139413859,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006129024872464354,
      "loss": 4.6625,
      "step": 15187
    },
    {
      "epoch": 2.079978088195015,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006128642995909051,
      "loss": 4.7694,
      "step": 15188
    },
    {
      "epoch": 2.0801150369761707,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006128261105856793,
      "loss": 4.7914,
      "step": 15189
    },
    {
      "epoch": 2.080251985757327,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000612787920231075,
      "loss": 4.7537,
      "step": 15190
    },
    {
      "epoch": 2.0803889345384827,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006127497285274083,
      "loss": 4.7622,
      "step": 15191
    },
    {
      "epoch": 2.0805258833196385,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006127115354749959,
      "loss": 4.7367,
      "step": 15192
    },
    {
      "epoch": 2.0806628321007943,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0006126733410741543,
      "loss": 4.7166,
      "step": 15193
    },
    {
      "epoch": 2.08079978088195,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006126351453251997,
      "loss": 4.747,
      "step": 15194
    },
    {
      "epoch": 2.080936729663106,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006125969482284491,
      "loss": 4.8157,
      "step": 15195
    },
    {
      "epoch": 2.081073678444262,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006125587497842189,
      "loss": 4.7145,
      "step": 15196
    },
    {
      "epoch": 2.081210627225418,
      "grad_norm": 0.375,
      "learning_rate": 0.0006125205499928254,
      "loss": 4.742,
      "step": 15197
    },
    {
      "epoch": 2.0813475760065736,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006124823488545854,
      "loss": 4.8301,
      "step": 15198
    },
    {
      "epoch": 2.0814845247877294,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006124441463698156,
      "loss": 4.743,
      "step": 15199
    },
    {
      "epoch": 2.081621473568885,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006124059425388324,
      "loss": 4.7444,
      "step": 15200
    },
    {
      "epoch": 2.081758422350041,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006123677373619525,
      "loss": 4.7703,
      "step": 15201
    },
    {
      "epoch": 2.081895371131197,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006123295308394924,
      "loss": 4.8203,
      "step": 15202
    },
    {
      "epoch": 2.082032319912353,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006122913229717687,
      "loss": 4.8054,
      "step": 15203
    },
    {
      "epoch": 2.0821692686935087,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006122531137590983,
      "loss": 4.7333,
      "step": 15204
    },
    {
      "epoch": 2.0823062174746645,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006122149032017977,
      "loss": 4.7983,
      "step": 15205
    },
    {
      "epoch": 2.0824431662558203,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006121766913001834,
      "loss": 4.7117,
      "step": 15206
    },
    {
      "epoch": 2.082580115036976,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006121384780545722,
      "loss": 4.7456,
      "step": 15207
    },
    {
      "epoch": 2.082717063818132,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006121002634652809,
      "loss": 4.8481,
      "step": 15208
    },
    {
      "epoch": 2.082854012599288,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006120620475326261,
      "loss": 4.7642,
      "step": 15209
    },
    {
      "epoch": 2.082990961380444,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006120238302569245,
      "loss": 4.7774,
      "step": 15210
    },
    {
      "epoch": 2.0831279101615996,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006119856116384927,
      "loss": 4.6561,
      "step": 15211
    },
    {
      "epoch": 2.0832648589427554,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006119473916776477,
      "loss": 4.82,
      "step": 15212
    },
    {
      "epoch": 2.083401807723911,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006119091703747059,
      "loss": 4.8181,
      "step": 15213
    },
    {
      "epoch": 2.083538756505067,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006118709477299844,
      "loss": 4.6892,
      "step": 15214
    },
    {
      "epoch": 2.083675705286223,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006118327237437996,
      "loss": 4.7612,
      "step": 15215
    },
    {
      "epoch": 2.083812654067379,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006117944984164685,
      "loss": 4.7519,
      "step": 15216
    },
    {
      "epoch": 2.0839496028485347,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000611756271748308,
      "loss": 4.7336,
      "step": 15217
    },
    {
      "epoch": 2.0840865516296905,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006117180437396344,
      "loss": 4.8061,
      "step": 15218
    },
    {
      "epoch": 2.0842235004108463,
      "grad_norm": 0.375,
      "learning_rate": 0.0006116798143907653,
      "loss": 4.7526,
      "step": 15219
    },
    {
      "epoch": 2.084360449192002,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006116415837020167,
      "loss": 4.7575,
      "step": 15220
    },
    {
      "epoch": 2.0844973979731583,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006116033516737058,
      "loss": 4.7612,
      "step": 15221
    },
    {
      "epoch": 2.084634346754314,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006115651183061496,
      "loss": 4.8325,
      "step": 15222
    },
    {
      "epoch": 2.08477129553547,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006115268835996646,
      "loss": 4.7821,
      "step": 15223
    },
    {
      "epoch": 2.0849082443166256,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006114886475545679,
      "loss": 4.7953,
      "step": 15224
    },
    {
      "epoch": 2.0850451930977814,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006114504101711764,
      "loss": 4.7818,
      "step": 15225
    },
    {
      "epoch": 2.085182141878937,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006114121714498067,
      "loss": 4.7729,
      "step": 15226
    },
    {
      "epoch": 2.085319090660093,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006113739313907759,
      "loss": 4.6802,
      "step": 15227
    },
    {
      "epoch": 2.085456039441249,
      "grad_norm": 0.359375,
      "learning_rate": 0.000611335689994401,
      "loss": 4.7153,
      "step": 15228
    },
    {
      "epoch": 2.085592988222405,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006112974472609986,
      "loss": 4.7308,
      "step": 15229
    },
    {
      "epoch": 2.0857299370035607,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000611259203190886,
      "loss": 4.6957,
      "step": 15230
    },
    {
      "epoch": 2.0858668857847165,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006112209577843799,
      "loss": 4.6644,
      "step": 15231
    },
    {
      "epoch": 2.0860038345658722,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006111827110417973,
      "loss": 4.723,
      "step": 15232
    },
    {
      "epoch": 2.086140783347028,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006111444629634552,
      "loss": 4.7464,
      "step": 15233
    },
    {
      "epoch": 2.0862777321281842,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006111062135496704,
      "loss": 4.7789,
      "step": 15234
    },
    {
      "epoch": 2.08641468090934,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006110679628007601,
      "loss": 4.7634,
      "step": 15235
    },
    {
      "epoch": 2.086551629690496,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006110297107170412,
      "loss": 4.6904,
      "step": 15236
    },
    {
      "epoch": 2.0866885784716516,
      "grad_norm": 0.375,
      "learning_rate": 0.0006109914572988309,
      "loss": 4.6669,
      "step": 15237
    },
    {
      "epoch": 2.0868255272528073,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006109532025464458,
      "loss": 4.7709,
      "step": 15238
    },
    {
      "epoch": 2.086962476033963,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006109149464602033,
      "loss": 4.7724,
      "step": 15239
    },
    {
      "epoch": 2.0870994248151193,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006108766890404202,
      "loss": 4.756,
      "step": 15240
    },
    {
      "epoch": 2.087236373596275,
      "grad_norm": 0.375,
      "learning_rate": 0.0006108384302874137,
      "loss": 4.7119,
      "step": 15241
    },
    {
      "epoch": 2.087373322377431,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006108001702015008,
      "loss": 4.6958,
      "step": 15242
    },
    {
      "epoch": 2.0875102711585867,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006107619087829984,
      "loss": 4.7837,
      "step": 15243
    },
    {
      "epoch": 2.0876472199397424,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000610723646032224,
      "loss": 4.7219,
      "step": 15244
    },
    {
      "epoch": 2.0877841687208982,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006106853819494943,
      "loss": 4.7887,
      "step": 15245
    },
    {
      "epoch": 2.0879211175020544,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006106471165351266,
      "loss": 4.8635,
      "step": 15246
    },
    {
      "epoch": 2.0880580662832102,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006106088497894379,
      "loss": 4.7692,
      "step": 15247
    },
    {
      "epoch": 2.088195015064366,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006105705817127455,
      "loss": 4.7538,
      "step": 15248
    },
    {
      "epoch": 2.0883319638455218,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006105323123053664,
      "loss": 4.7788,
      "step": 15249
    },
    {
      "epoch": 2.0884689126266776,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006104940415676178,
      "loss": 4.7098,
      "step": 15250
    },
    {
      "epoch": 2.0886058614078333,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006104557694998167,
      "loss": 4.7368,
      "step": 15251
    },
    {
      "epoch": 2.088742810188989,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006104174961022806,
      "loss": 4.7816,
      "step": 15252
    },
    {
      "epoch": 2.0888797589701453,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006103792213753264,
      "loss": 4.7314,
      "step": 15253
    },
    {
      "epoch": 2.089016707751301,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006103409453192713,
      "loss": 4.7694,
      "step": 15254
    },
    {
      "epoch": 2.089153656532457,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006103026679344327,
      "loss": 4.7912,
      "step": 15255
    },
    {
      "epoch": 2.0892906053136127,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006102643892211275,
      "loss": 4.7257,
      "step": 15256
    },
    {
      "epoch": 2.0894275540947684,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006102261091796733,
      "loss": 4.6788,
      "step": 15257
    },
    {
      "epoch": 2.089564502875924,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006101878278103871,
      "loss": 4.7192,
      "step": 15258
    },
    {
      "epoch": 2.0897014516570804,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006101495451135863,
      "loss": 4.7194,
      "step": 15259
    },
    {
      "epoch": 2.089838400438236,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006101112610895879,
      "loss": 4.7594,
      "step": 15260
    },
    {
      "epoch": 2.089975349219392,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006100729757387094,
      "loss": 4.689,
      "step": 15261
    },
    {
      "epoch": 2.0901122980005478,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006100346890612679,
      "loss": 4.6979,
      "step": 15262
    },
    {
      "epoch": 2.0902492467817035,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006099964010575808,
      "loss": 4.6381,
      "step": 15263
    },
    {
      "epoch": 2.0903861955628593,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006099581117279656,
      "loss": 4.7803,
      "step": 15264
    },
    {
      "epoch": 2.0905231443440155,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006099198210727393,
      "loss": 4.7341,
      "step": 15265
    },
    {
      "epoch": 2.0906600931251713,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006098815290922194,
      "loss": 4.7842,
      "step": 15266
    },
    {
      "epoch": 2.090797041906327,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000609843235786723,
      "loss": 4.7833,
      "step": 15267
    },
    {
      "epoch": 2.090933990687483,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006098049411565679,
      "loss": 4.6695,
      "step": 15268
    },
    {
      "epoch": 2.0910709394686386,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006097666452020709,
      "loss": 4.6939,
      "step": 15269
    },
    {
      "epoch": 2.0912078882497944,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006097283479235497,
      "loss": 4.7541,
      "step": 15270
    },
    {
      "epoch": 2.0913448370309506,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006096900493213216,
      "loss": 4.6705,
      "step": 15271
    },
    {
      "epoch": 2.0914817858121064,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000609651749395704,
      "loss": 4.8144,
      "step": 15272
    },
    {
      "epoch": 2.091618734593262,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006096134481470145,
      "loss": 4.7454,
      "step": 15273
    },
    {
      "epoch": 2.091755683374418,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006095751455755702,
      "loss": 4.784,
      "step": 15274
    },
    {
      "epoch": 2.0918926321555738,
      "grad_norm": 0.375,
      "learning_rate": 0.0006095368416816886,
      "loss": 4.8972,
      "step": 15275
    },
    {
      "epoch": 2.0920295809367295,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006094985364656872,
      "loss": 4.8164,
      "step": 15276
    },
    {
      "epoch": 2.0921665297178853,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006094602299278835,
      "loss": 4.7774,
      "step": 15277
    },
    {
      "epoch": 2.0923034784990415,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006094219220685948,
      "loss": 4.8212,
      "step": 15278
    },
    {
      "epoch": 2.0924404272801973,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006093836128881387,
      "loss": 4.781,
      "step": 15279
    },
    {
      "epoch": 2.092577376061353,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006093453023868324,
      "loss": 4.6607,
      "step": 15280
    },
    {
      "epoch": 2.092714324842509,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006093069905649939,
      "loss": 4.7414,
      "step": 15281
    },
    {
      "epoch": 2.0928512736236646,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006092686774229404,
      "loss": 4.7297,
      "step": 15282
    },
    {
      "epoch": 2.0929882224048204,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006092303629609893,
      "loss": 4.6951,
      "step": 15283
    },
    {
      "epoch": 2.0931251711859766,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006091920471794582,
      "loss": 4.8118,
      "step": 15284
    },
    {
      "epoch": 2.0932621199671324,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0006091537300786647,
      "loss": 4.7364,
      "step": 15285
    },
    {
      "epoch": 2.093399068748288,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006091154116589264,
      "loss": 4.7603,
      "step": 15286
    },
    {
      "epoch": 2.093536017529444,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006090770919205607,
      "loss": 4.8559,
      "step": 15287
    },
    {
      "epoch": 2.0936729663105997,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006090387708638852,
      "loss": 4.7251,
      "step": 15288
    },
    {
      "epoch": 2.0938099150917555,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006090004484892175,
      "loss": 4.7226,
      "step": 15289
    },
    {
      "epoch": 2.0939468638729117,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006089621247968752,
      "loss": 4.7821,
      "step": 15290
    },
    {
      "epoch": 2.0940838126540675,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006089237997871761,
      "loss": 4.7891,
      "step": 15291
    },
    {
      "epoch": 2.0942207614352233,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006088854734604374,
      "loss": 4.7455,
      "step": 15292
    },
    {
      "epoch": 2.094357710216379,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006088471458169769,
      "loss": 4.7405,
      "step": 15293
    },
    {
      "epoch": 2.094494658997535,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006088088168571124,
      "loss": 4.6618,
      "step": 15294
    },
    {
      "epoch": 2.0946316077786906,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006087704865811613,
      "loss": 4.7815,
      "step": 15295
    },
    {
      "epoch": 2.0947685565598464,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006087321549894413,
      "loss": 4.7304,
      "step": 15296
    },
    {
      "epoch": 2.0949055053410026,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006086938220822702,
      "loss": 4.6363,
      "step": 15297
    },
    {
      "epoch": 2.0950424541221584,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0006086554878599655,
      "loss": 4.7567,
      "step": 15298
    },
    {
      "epoch": 2.095179402903314,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006086171523228451,
      "loss": 4.847,
      "step": 15299
    },
    {
      "epoch": 2.09531635168447,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006085788154712265,
      "loss": 4.7512,
      "step": 15300
    },
    {
      "epoch": 2.0954533004656257,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006085404773054274,
      "loss": 4.8343,
      "step": 15301
    },
    {
      "epoch": 2.0955902492467815,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006085021378257656,
      "loss": 4.7583,
      "step": 15302
    },
    {
      "epoch": 2.0957271980279377,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006084637970325588,
      "loss": 4.7649,
      "step": 15303
    },
    {
      "epoch": 2.0958641468090935,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006084254549261249,
      "loss": 4.6904,
      "step": 15304
    },
    {
      "epoch": 2.0960010955902493,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006083871115067813,
      "loss": 4.7799,
      "step": 15305
    },
    {
      "epoch": 2.096138044371405,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006083487667748461,
      "loss": 4.7633,
      "step": 15306
    },
    {
      "epoch": 2.096274993152561,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006083104207306369,
      "loss": 4.7538,
      "step": 15307
    },
    {
      "epoch": 2.0964119419337166,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006082720733744716,
      "loss": 4.7806,
      "step": 15308
    },
    {
      "epoch": 2.096548890714873,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006082337247066677,
      "loss": 4.789,
      "step": 15309
    },
    {
      "epoch": 2.0966858394960286,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006081953747275433,
      "loss": 4.8494,
      "step": 15310
    },
    {
      "epoch": 2.0968227882771844,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006081570234374163,
      "loss": 4.7823,
      "step": 15311
    },
    {
      "epoch": 2.09695973705834,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006081186708366042,
      "loss": 4.7138,
      "step": 15312
    },
    {
      "epoch": 2.097096685839496,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006080803169254251,
      "loss": 4.7777,
      "step": 15313
    },
    {
      "epoch": 2.0972336346206517,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006080419617041967,
      "loss": 4.679,
      "step": 15314
    },
    {
      "epoch": 2.097370583401808,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006080036051732369,
      "loss": 4.7224,
      "step": 15315
    },
    {
      "epoch": 2.0975075321829637,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006079652473328635,
      "loss": 4.7229,
      "step": 15316
    },
    {
      "epoch": 2.0976444809641195,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006079268881833946,
      "loss": 4.6955,
      "step": 15317
    },
    {
      "epoch": 2.0977814297452753,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006078885277251479,
      "loss": 4.7156,
      "step": 15318
    },
    {
      "epoch": 2.097918378526431,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006078501659584412,
      "loss": 4.6971,
      "step": 15319
    },
    {
      "epoch": 2.098055327307587,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006078118028835928,
      "loss": 4.796,
      "step": 15320
    },
    {
      "epoch": 2.098192276088743,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006077734385009202,
      "loss": 4.8248,
      "step": 15321
    },
    {
      "epoch": 2.098329224869899,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006077350728107416,
      "loss": 4.8096,
      "step": 15322
    },
    {
      "epoch": 2.0984661736510546,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006076967058133749,
      "loss": 4.8215,
      "step": 15323
    },
    {
      "epoch": 2.0986031224322104,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006076583375091379,
      "loss": 4.6986,
      "step": 15324
    },
    {
      "epoch": 2.098740071213366,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006076199678983488,
      "loss": 4.7966,
      "step": 15325
    },
    {
      "epoch": 2.098877019994522,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006075815969813254,
      "loss": 4.7068,
      "step": 15326
    },
    {
      "epoch": 2.0990139687756777,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006075432247583858,
      "loss": 4.7139,
      "step": 15327
    },
    {
      "epoch": 2.099150917556834,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006075048512298478,
      "loss": 4.7733,
      "step": 15328
    },
    {
      "epoch": 2.0992878663379897,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006074664763960298,
      "loss": 4.7551,
      "step": 15329
    },
    {
      "epoch": 2.0994248151191455,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006074281002572495,
      "loss": 4.6908,
      "step": 15330
    },
    {
      "epoch": 2.0995617639003012,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006073897228138251,
      "loss": 4.7903,
      "step": 15331
    },
    {
      "epoch": 2.099698712681457,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006073513440660744,
      "loss": 4.7613,
      "step": 15332
    },
    {
      "epoch": 2.099835661462613,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006073129640143158,
      "loss": 4.7867,
      "step": 15333
    },
    {
      "epoch": 2.099972610243769,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006072745826588669,
      "loss": 4.7896,
      "step": 15334
    },
    {
      "epoch": 2.100109559024925,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006072362000000462,
      "loss": 4.7058,
      "step": 15335
    },
    {
      "epoch": 2.1002465078060806,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006071978160381718,
      "loss": 4.7893,
      "step": 15336
    },
    {
      "epoch": 2.1003834565872364,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0006071594307735614,
      "loss": 4.6871,
      "step": 15337
    },
    {
      "epoch": 2.100520405368392,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006071210442065336,
      "loss": 4.7901,
      "step": 15338
    },
    {
      "epoch": 2.100657354149548,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006070826563374061,
      "loss": 4.7398,
      "step": 15339
    },
    {
      "epoch": 2.100794302930704,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006070442671664973,
      "loss": 4.7775,
      "step": 15340
    },
    {
      "epoch": 2.10093125171186,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000607005876694125,
      "loss": 4.7398,
      "step": 15341
    },
    {
      "epoch": 2.1010682004930157,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006069674849206078,
      "loss": 4.6985,
      "step": 15342
    },
    {
      "epoch": 2.1012051492741715,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006069290918462635,
      "loss": 4.8025,
      "step": 15343
    },
    {
      "epoch": 2.1013420980553272,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006068906974714105,
      "loss": 4.835,
      "step": 15344
    },
    {
      "epoch": 2.101479046836483,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000606852301796367,
      "loss": 4.7341,
      "step": 15345
    },
    {
      "epoch": 2.101615995617639,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000606813904821451,
      "loss": 4.7614,
      "step": 15346
    },
    {
      "epoch": 2.101752944398795,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000606775506546981,
      "loss": 4.8122,
      "step": 15347
    },
    {
      "epoch": 2.101889893179951,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006067371069732747,
      "loss": 4.8206,
      "step": 15348
    },
    {
      "epoch": 2.1020268419611066,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006066987061006508,
      "loss": 4.7979,
      "step": 15349
    },
    {
      "epoch": 2.1021637907422623,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006066603039294275,
      "loss": 4.6628,
      "step": 15350
    },
    {
      "epoch": 2.102300739523418,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006066219004599227,
      "loss": 4.7401,
      "step": 15351
    },
    {
      "epoch": 2.102437688304574,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006065834956924551,
      "loss": 4.7982,
      "step": 15352
    },
    {
      "epoch": 2.10257463708573,
      "grad_norm": 0.375,
      "learning_rate": 0.0006065450896273425,
      "loss": 4.7263,
      "step": 15353
    },
    {
      "epoch": 2.102711585866886,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006065066822649037,
      "loss": 4.7609,
      "step": 15354
    },
    {
      "epoch": 2.1028485346480417,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006064682736054565,
      "loss": 4.7536,
      "step": 15355
    },
    {
      "epoch": 2.1029854834291974,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006064298636493196,
      "loss": 4.824,
      "step": 15356
    },
    {
      "epoch": 2.103122432210353,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006063914523968112,
      "loss": 4.7714,
      "step": 15357
    },
    {
      "epoch": 2.103259380991509,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006063530398482494,
      "loss": 4.7123,
      "step": 15358
    },
    {
      "epoch": 2.103396329772665,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0006063146260039528,
      "loss": 4.7574,
      "step": 15359
    },
    {
      "epoch": 2.103533278553821,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006062762108642397,
      "loss": 4.7097,
      "step": 15360
    },
    {
      "epoch": 2.1036702273349768,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0006062377944294283,
      "loss": 4.7488,
      "step": 15361
    },
    {
      "epoch": 2.1038071761161325,
      "grad_norm": 0.4375,
      "learning_rate": 0.000606199376699837,
      "loss": 4.826,
      "step": 15362
    },
    {
      "epoch": 2.1039441248972883,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0006061609576757843,
      "loss": 4.7124,
      "step": 15363
    },
    {
      "epoch": 2.104081073678444,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006061225373575886,
      "loss": 4.6954,
      "step": 15364
    },
    {
      "epoch": 2.1042180224596003,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006060841157455683,
      "loss": 4.6868,
      "step": 15365
    },
    {
      "epoch": 2.104354971240756,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006060456928400417,
      "loss": 4.732,
      "step": 15366
    },
    {
      "epoch": 2.104491920021912,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000606007268641327,
      "loss": 4.7383,
      "step": 15367
    },
    {
      "epoch": 2.1046288688030677,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006059688431497432,
      "loss": 4.6867,
      "step": 15368
    },
    {
      "epoch": 2.1047658175842234,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006059304163656083,
      "loss": 4.7725,
      "step": 15369
    },
    {
      "epoch": 2.104902766365379,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0006058919882892408,
      "loss": 4.7672,
      "step": 15370
    },
    {
      "epoch": 2.105039715146535,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006058535589209595,
      "loss": 4.7345,
      "step": 15371
    },
    {
      "epoch": 2.105176663927691,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006058151282610823,
      "loss": 4.8019,
      "step": 15372
    },
    {
      "epoch": 2.105313612708847,
      "grad_norm": 0.375,
      "learning_rate": 0.0006057766963099284,
      "loss": 4.7236,
      "step": 15373
    },
    {
      "epoch": 2.1054505614900028,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0006057382630678154,
      "loss": 4.7409,
      "step": 15374
    },
    {
      "epoch": 2.1055875102711585,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006056998285350627,
      "loss": 4.8039,
      "step": 15375
    },
    {
      "epoch": 2.1057244590523143,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0006056613927119883,
      "loss": 4.7293,
      "step": 15376
    },
    {
      "epoch": 2.10586140783347,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006056229555989108,
      "loss": 4.7491,
      "step": 15377
    },
    {
      "epoch": 2.1059983566146263,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0006055845171961489,
      "loss": 4.7172,
      "step": 15378
    },
    {
      "epoch": 2.106135305395782,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006055460775040209,
      "loss": 4.7054,
      "step": 15379
    },
    {
      "epoch": 2.106272254176938,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0006055076365228456,
      "loss": 4.8517,
      "step": 15380
    },
    {
      "epoch": 2.1064092029580936,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006054691942529414,
      "loss": 4.6306,
      "step": 15381
    },
    {
      "epoch": 2.1065461517392494,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000605430750694627,
      "loss": 4.7608,
      "step": 15382
    },
    {
      "epoch": 2.106683100520405,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006053923058482208,
      "loss": 4.6974,
      "step": 15383
    },
    {
      "epoch": 2.1068200493015614,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006053538597140416,
      "loss": 4.7679,
      "step": 15384
    },
    {
      "epoch": 2.106956998082717,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000605315412292408,
      "loss": 4.7529,
      "step": 15385
    },
    {
      "epoch": 2.107093946863873,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006052769635836385,
      "loss": 4.7711,
      "step": 15386
    },
    {
      "epoch": 2.1072308956450287,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006052385135880519,
      "loss": 4.7412,
      "step": 15387
    },
    {
      "epoch": 2.1073678444261845,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006052000623059667,
      "loss": 4.7406,
      "step": 15388
    },
    {
      "epoch": 2.1075047932073403,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006051616097377016,
      "loss": 4.8006,
      "step": 15389
    },
    {
      "epoch": 2.1076417419884965,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006051231558835752,
      "loss": 4.7296,
      "step": 15390
    },
    {
      "epoch": 2.1077786907696523,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006050847007439062,
      "loss": 4.7581,
      "step": 15391
    },
    {
      "epoch": 2.107915639550808,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006050462443190135,
      "loss": 4.8589,
      "step": 15392
    },
    {
      "epoch": 2.108052588331964,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006050077866092154,
      "loss": 4.7809,
      "step": 15393
    },
    {
      "epoch": 2.1081895371131196,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006049693276148311,
      "loss": 4.8211,
      "step": 15394
    },
    {
      "epoch": 2.1083264858942754,
      "grad_norm": 0.34375,
      "learning_rate": 0.000604930867336179,
      "loss": 4.7145,
      "step": 15395
    },
    {
      "epoch": 2.108463434675431,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006048924057735778,
      "loss": 4.6965,
      "step": 15396
    },
    {
      "epoch": 2.1086003834565874,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0006048539429273463,
      "loss": 4.7131,
      "step": 15397
    },
    {
      "epoch": 2.108737332237743,
      "grad_norm": 0.375,
      "learning_rate": 0.0006048154787978035,
      "loss": 4.6825,
      "step": 15398
    },
    {
      "epoch": 2.108874281018899,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0006047770133852676,
      "loss": 4.686,
      "step": 15399
    },
    {
      "epoch": 2.1090112298000547,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000604738546690058,
      "loss": 4.7217,
      "step": 15400
    },
    {
      "epoch": 2.1091481785812105,
      "grad_norm": 0.42578125,
      "learning_rate": 0.000604700078712493,
      "loss": 4.741,
      "step": 15401
    },
    {
      "epoch": 2.1092851273623663,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006046616094528915,
      "loss": 4.7735,
      "step": 15402
    },
    {
      "epoch": 2.1094220761435225,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006046231389115726,
      "loss": 4.6605,
      "step": 15403
    },
    {
      "epoch": 2.1095590249246783,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006045846670888549,
      "loss": 4.7585,
      "step": 15404
    },
    {
      "epoch": 2.109695973705834,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006045461939850572,
      "loss": 4.7593,
      "step": 15405
    },
    {
      "epoch": 2.10983292248699,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006045077196004983,
      "loss": 4.8025,
      "step": 15406
    },
    {
      "epoch": 2.1099698712681456,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006044692439354972,
      "loss": 4.7183,
      "step": 15407
    },
    {
      "epoch": 2.1101068200493014,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006044307669903725,
      "loss": 4.7555,
      "step": 15408
    },
    {
      "epoch": 2.1102437688304576,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0006043922887654435,
      "loss": 4.7575,
      "step": 15409
    },
    {
      "epoch": 2.1103807176116134,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006043538092610286,
      "loss": 4.7446,
      "step": 15410
    },
    {
      "epoch": 2.110517666392769,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000604315328477447,
      "loss": 4.7811,
      "step": 15411
    },
    {
      "epoch": 2.110654615173925,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006042768464150173,
      "loss": 4.6775,
      "step": 15412
    },
    {
      "epoch": 2.1107915639550807,
      "grad_norm": 0.375,
      "learning_rate": 0.0006042383630740588,
      "loss": 4.8034,
      "step": 15413
    },
    {
      "epoch": 2.1109285127362365,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006041998784548902,
      "loss": 4.7867,
      "step": 15414
    },
    {
      "epoch": 2.1110654615173923,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006041613925578305,
      "loss": 4.7796,
      "step": 15415
    },
    {
      "epoch": 2.1112024102985485,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006041229053831985,
      "loss": 4.8792,
      "step": 15416
    },
    {
      "epoch": 2.1113393590797043,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006040844169313134,
      "loss": 4.7549,
      "step": 15417
    },
    {
      "epoch": 2.11147630786086,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006040459272024939,
      "loss": 4.7894,
      "step": 15418
    },
    {
      "epoch": 2.111613256642016,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0006040074361970592,
      "loss": 4.655,
      "step": 15419
    },
    {
      "epoch": 2.1117502054231716,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006039689439153281,
      "loss": 4.8019,
      "step": 15420
    },
    {
      "epoch": 2.1118871542043274,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006039304503576197,
      "loss": 4.7635,
      "step": 15421
    },
    {
      "epoch": 2.1120241029854836,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000603891955524253,
      "loss": 4.7524,
      "step": 15422
    },
    {
      "epoch": 2.1121610517666394,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000603853459415547,
      "loss": 4.7622,
      "step": 15423
    },
    {
      "epoch": 2.112298000547795,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006038149620318207,
      "loss": 4.7357,
      "step": 15424
    },
    {
      "epoch": 2.112434949328951,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000603776463373393,
      "loss": 4.7695,
      "step": 15425
    },
    {
      "epoch": 2.1125718981101067,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006037379634405834,
      "loss": 4.8495,
      "step": 15426
    },
    {
      "epoch": 2.1127088468912625,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0006036994622337105,
      "loss": 4.7848,
      "step": 15427
    },
    {
      "epoch": 2.1128457956724187,
      "grad_norm": 0.375,
      "learning_rate": 0.0006036609597530935,
      "loss": 4.6732,
      "step": 15428
    },
    {
      "epoch": 2.1129827444535745,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006036224559990516,
      "loss": 4.7883,
      "step": 15429
    },
    {
      "epoch": 2.1131196932347303,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006035839509719037,
      "loss": 4.7941,
      "step": 15430
    },
    {
      "epoch": 2.113256642015886,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000603545444671969,
      "loss": 4.7098,
      "step": 15431
    },
    {
      "epoch": 2.113393590797042,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006035069370995667,
      "loss": 4.7111,
      "step": 15432
    },
    {
      "epoch": 2.1135305395781976,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006034684282550159,
      "loss": 4.785,
      "step": 15433
    },
    {
      "epoch": 2.113667488359354,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006034299181386356,
      "loss": 4.7726,
      "step": 15434
    },
    {
      "epoch": 2.1138044371405096,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0006033914067507448,
      "loss": 4.7981,
      "step": 15435
    },
    {
      "epoch": 2.1139413859216654,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006033528940916633,
      "loss": 4.7849,
      "step": 15436
    },
    {
      "epoch": 2.114078334702821,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006033143801617094,
      "loss": 4.7656,
      "step": 15437
    },
    {
      "epoch": 2.114215283483977,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0006032758649612029,
      "loss": 4.7457,
      "step": 15438
    },
    {
      "epoch": 2.1143522322651327,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006032373484904628,
      "loss": 4.7214,
      "step": 15439
    },
    {
      "epoch": 2.114489181046289,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006031988307498081,
      "loss": 4.7529,
      "step": 15440
    },
    {
      "epoch": 2.1146261298274447,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006031603117395584,
      "loss": 4.8223,
      "step": 15441
    },
    {
      "epoch": 2.1147630786086005,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006031217914600326,
      "loss": 4.6475,
      "step": 15442
    },
    {
      "epoch": 2.1149000273897562,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0006030832699115501,
      "loss": 4.7361,
      "step": 15443
    },
    {
      "epoch": 2.115036976170912,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006030447470944299,
      "loss": 4.7909,
      "step": 15444
    },
    {
      "epoch": 2.115173924952068,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006030062230089916,
      "loss": 4.7706,
      "step": 15445
    },
    {
      "epoch": 2.1153108737332236,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006029676976555541,
      "loss": 4.7405,
      "step": 15446
    },
    {
      "epoch": 2.11544782251438,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006029291710344369,
      "loss": 4.8069,
      "step": 15447
    },
    {
      "epoch": 2.1155847712955356,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0006028906431459593,
      "loss": 4.7958,
      "step": 15448
    },
    {
      "epoch": 2.1157217200766913,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006028521139904404,
      "loss": 4.7266,
      "step": 15449
    },
    {
      "epoch": 2.115858668857847,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006028135835681997,
      "loss": 4.7669,
      "step": 15450
    },
    {
      "epoch": 2.115995617639003,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006027750518795564,
      "loss": 4.6965,
      "step": 15451
    },
    {
      "epoch": 2.1161325664201587,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006027365189248298,
      "loss": 4.7459,
      "step": 15452
    },
    {
      "epoch": 2.116269515201315,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006026979847043392,
      "loss": 4.7311,
      "step": 15453
    },
    {
      "epoch": 2.1164064639824707,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006026594492184043,
      "loss": 4.7505,
      "step": 15454
    },
    {
      "epoch": 2.1165434127636265,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006026209124673439,
      "loss": 4.7605,
      "step": 15455
    },
    {
      "epoch": 2.1166803615447822,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006025823744514777,
      "loss": 4.8175,
      "step": 15456
    },
    {
      "epoch": 2.116817310325938,
      "grad_norm": 0.39453125,
      "learning_rate": 0.000602543835171125,
      "loss": 4.724,
      "step": 15457
    },
    {
      "epoch": 2.116954259107094,
      "grad_norm": 0.34375,
      "learning_rate": 0.0006025052946266052,
      "loss": 4.7616,
      "step": 15458
    },
    {
      "epoch": 2.11709120788825,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006024667528182376,
      "loss": 4.7396,
      "step": 15459
    },
    {
      "epoch": 2.117228156669406,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006024282097463418,
      "loss": 4.682,
      "step": 15460
    },
    {
      "epoch": 2.1173651054505616,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006023896654112371,
      "loss": 4.7163,
      "step": 15461
    },
    {
      "epoch": 2.1175020542317173,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006023511198132429,
      "loss": 4.8331,
      "step": 15462
    },
    {
      "epoch": 2.117639003012873,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0006023125729526787,
      "loss": 4.7341,
      "step": 15463
    },
    {
      "epoch": 2.117775951794029,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006022740248298639,
      "loss": 4.6946,
      "step": 15464
    },
    {
      "epoch": 2.1179129005751847,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000602235475445118,
      "loss": 4.7989,
      "step": 15465
    },
    {
      "epoch": 2.118049849356341,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006021969247987604,
      "loss": 4.8021,
      "step": 15466
    },
    {
      "epoch": 2.1181867981374967,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0006021583728911106,
      "loss": 4.6784,
      "step": 15467
    },
    {
      "epoch": 2.1183237469186524,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0006021198197224882,
      "loss": 4.7827,
      "step": 15468
    },
    {
      "epoch": 2.118460695699808,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0006020812652932124,
      "loss": 4.7178,
      "step": 15469
    },
    {
      "epoch": 2.118597644480964,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006020427096036031,
      "loss": 4.7382,
      "step": 15470
    },
    {
      "epoch": 2.1187345932621198,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006020041526539796,
      "loss": 4.7243,
      "step": 15471
    },
    {
      "epoch": 2.118871542043276,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0006019655944446615,
      "loss": 4.7303,
      "step": 15472
    },
    {
      "epoch": 2.1190084908244318,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0006019270349759681,
      "loss": 4.7487,
      "step": 15473
    },
    {
      "epoch": 2.1191454396055875,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0006018884742482194,
      "loss": 4.7823,
      "step": 15474
    },
    {
      "epoch": 2.1192823883867433,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006018499122617345,
      "loss": 4.7294,
      "step": 15475
    },
    {
      "epoch": 2.119419337167899,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006018113490168335,
      "loss": 4.8056,
      "step": 15476
    },
    {
      "epoch": 2.119556285949055,
      "grad_norm": 0.40625,
      "learning_rate": 0.0006017727845138354,
      "loss": 4.6166,
      "step": 15477
    },
    {
      "epoch": 2.119693234730211,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006017342187530601,
      "loss": 4.7331,
      "step": 15478
    },
    {
      "epoch": 2.119830183511367,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0006016956517348274,
      "loss": 4.6969,
      "step": 15479
    },
    {
      "epoch": 2.1199671322925226,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006016570834594565,
      "loss": 4.7343,
      "step": 15480
    },
    {
      "epoch": 2.1201040810736784,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006016185139272672,
      "loss": 4.7616,
      "step": 15481
    },
    {
      "epoch": 2.120241029854834,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006015799431385792,
      "loss": 4.7808,
      "step": 15482
    },
    {
      "epoch": 2.12037797863599,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0006015413710937121,
      "loss": 4.7661,
      "step": 15483
    },
    {
      "epoch": 2.120514927417146,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006015027977929855,
      "loss": 4.7483,
      "step": 15484
    },
    {
      "epoch": 2.120651876198302,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006014642232367192,
      "loss": 4.7777,
      "step": 15485
    },
    {
      "epoch": 2.1207888249794578,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0006014256474252326,
      "loss": 4.7311,
      "step": 15486
    },
    {
      "epoch": 2.1209257737606135,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0006013870703588457,
      "loss": 4.8097,
      "step": 15487
    },
    {
      "epoch": 2.1210627225417693,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006013484920378781,
      "loss": 4.7721,
      "step": 15488
    },
    {
      "epoch": 2.121199671322925,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0006013099124626495,
      "loss": 4.7114,
      "step": 15489
    },
    {
      "epoch": 2.121336620104081,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006012713316334797,
      "loss": 4.6718,
      "step": 15490
    },
    {
      "epoch": 2.121473568885237,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006012327495506881,
      "loss": 4.7849,
      "step": 15491
    },
    {
      "epoch": 2.121610517666393,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006011941662145948,
      "loss": 4.6926,
      "step": 15492
    },
    {
      "epoch": 2.1217474664475486,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0006011555816255194,
      "loss": 4.7962,
      "step": 15493
    },
    {
      "epoch": 2.1218844152287044,
      "grad_norm": 0.359375,
      "learning_rate": 0.0006011169957837816,
      "loss": 4.7209,
      "step": 15494
    },
    {
      "epoch": 2.12202136400986,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006010784086897013,
      "loss": 4.781,
      "step": 15495
    },
    {
      "epoch": 2.122158312791016,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006010398203435981,
      "loss": 4.7725,
      "step": 15496
    },
    {
      "epoch": 2.122295261572172,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0006010012307457921,
      "loss": 4.7879,
      "step": 15497
    },
    {
      "epoch": 2.122432210353328,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0006009626398966029,
      "loss": 4.7355,
      "step": 15498
    },
    {
      "epoch": 2.1225691591344837,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006009240477963504,
      "loss": 4.6633,
      "step": 15499
    },
    {
      "epoch": 2.1227061079156395,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006008854544453542,
      "loss": 4.7391,
      "step": 15500
    },
    {
      "epoch": 2.1228430566967953,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006008468598439344,
      "loss": 4.8233,
      "step": 15501
    },
    {
      "epoch": 2.122980005477951,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006008082639924106,
      "loss": 4.7721,
      "step": 15502
    },
    {
      "epoch": 2.1231169542591073,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006007696668911028,
      "loss": 4.7248,
      "step": 15503
    },
    {
      "epoch": 2.123253903040263,
      "grad_norm": 0.375,
      "learning_rate": 0.0006007310685403308,
      "loss": 4.7803,
      "step": 15504
    },
    {
      "epoch": 2.123390851821419,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006006924689404147,
      "loss": 4.7296,
      "step": 15505
    },
    {
      "epoch": 2.1235278006025746,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006006538680916741,
      "loss": 4.778,
      "step": 15506
    },
    {
      "epoch": 2.1236647493837304,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0006006152659944289,
      "loss": 4.7871,
      "step": 15507
    },
    {
      "epoch": 2.123801698164886,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0006005766626489993,
      "loss": 4.7319,
      "step": 15508
    },
    {
      "epoch": 2.1239386469460424,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000600538058055705,
      "loss": 4.704,
      "step": 15509
    },
    {
      "epoch": 2.124075595727198,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0006004994522148658,
      "loss": 4.7874,
      "step": 15510
    },
    {
      "epoch": 2.124212544508354,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0006004608451268018,
      "loss": 4.8359,
      "step": 15511
    },
    {
      "epoch": 2.1243494932895097,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000600422236791833,
      "loss": 4.7552,
      "step": 15512
    },
    {
      "epoch": 2.1244864420706655,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0006003836272102793,
      "loss": 4.7412,
      "step": 15513
    },
    {
      "epoch": 2.1246233908518213,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0006003450163824605,
      "loss": 4.7366,
      "step": 15514
    },
    {
      "epoch": 2.124760339632977,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0006003064043086968,
      "loss": 4.7316,
      "step": 15515
    },
    {
      "epoch": 2.1248972884141333,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000600267790989308,
      "loss": 4.7573,
      "step": 15516
    },
    {
      "epoch": 2.125034237195289,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0006002291764246145,
      "loss": 4.7222,
      "step": 15517
    },
    {
      "epoch": 2.125171185976445,
      "grad_norm": 0.390625,
      "learning_rate": 0.0006001905606149358,
      "loss": 4.6755,
      "step": 15518
    },
    {
      "epoch": 2.1253081347576006,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0006001519435605923,
      "loss": 4.6733,
      "step": 15519
    },
    {
      "epoch": 2.1254450835387564,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0006001133252619036,
      "loss": 4.7635,
      "step": 15520
    },
    {
      "epoch": 2.125582032319912,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0006000747057191903,
      "loss": 4.6249,
      "step": 15521
    },
    {
      "epoch": 2.1257189811010684,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0006000360849327719,
      "loss": 4.6767,
      "step": 15522
    },
    {
      "epoch": 2.125855929882224,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005999974629029689,
      "loss": 4.7769,
      "step": 15523
    },
    {
      "epoch": 2.12599287866338,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005999588396301011,
      "loss": 4.8165,
      "step": 15524
    },
    {
      "epoch": 2.1261298274445357,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005999202151144885,
      "loss": 4.8379,
      "step": 15525
    },
    {
      "epoch": 2.1262667762256915,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005998815893564517,
      "loss": 4.7956,
      "step": 15526
    },
    {
      "epoch": 2.1264037250068473,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005998429623563102,
      "loss": 4.7112,
      "step": 15527
    },
    {
      "epoch": 2.1265406737880035,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005998043341143845,
      "loss": 4.7664,
      "step": 15528
    },
    {
      "epoch": 2.1266776225691593,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005997657046309944,
      "loss": 4.8046,
      "step": 15529
    },
    {
      "epoch": 2.126814571350315,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005997270739064604,
      "loss": 4.7008,
      "step": 15530
    },
    {
      "epoch": 2.126951520131471,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005996884419411024,
      "loss": 4.7394,
      "step": 15531
    },
    {
      "epoch": 2.1270884689126266,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005996498087352407,
      "loss": 4.8137,
      "step": 15532
    },
    {
      "epoch": 2.1272254176937824,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005996111742891952,
      "loss": 4.761,
      "step": 15533
    },
    {
      "epoch": 2.127362366474938,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005995725386032864,
      "loss": 4.7618,
      "step": 15534
    },
    {
      "epoch": 2.1274993152560944,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005995339016778343,
      "loss": 4.7462,
      "step": 15535
    },
    {
      "epoch": 2.12763626403725,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000599495263513159,
      "loss": 4.7083,
      "step": 15536
    },
    {
      "epoch": 2.127773212818406,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000599456624109581,
      "loss": 4.7227,
      "step": 15537
    },
    {
      "epoch": 2.1279101615995617,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005994179834674202,
      "loss": 4.7157,
      "step": 15538
    },
    {
      "epoch": 2.1280471103807175,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005993793415869971,
      "loss": 4.6617,
      "step": 15539
    },
    {
      "epoch": 2.1281840591618733,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005993406984686316,
      "loss": 4.8187,
      "step": 15540
    },
    {
      "epoch": 2.1283210079430295,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005993020541126444,
      "loss": 4.7851,
      "step": 15541
    },
    {
      "epoch": 2.1284579567241853,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005992634085193553,
      "loss": 4.77,
      "step": 15542
    },
    {
      "epoch": 2.128594905505341,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005992247616890848,
      "loss": 4.7198,
      "step": 15543
    },
    {
      "epoch": 2.128731854286497,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005991861136221533,
      "loss": 4.787,
      "step": 15544
    },
    {
      "epoch": 2.1288688030676526,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005991474643188808,
      "loss": 4.71,
      "step": 15545
    },
    {
      "epoch": 2.1290057518488084,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005991088137795878,
      "loss": 4.7445,
      "step": 15546
    },
    {
      "epoch": 2.1291427006299646,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005990701620045944,
      "loss": 4.7833,
      "step": 15547
    },
    {
      "epoch": 2.1292796494111204,
      "grad_norm": 0.375,
      "learning_rate": 0.0005990315089942212,
      "loss": 4.6958,
      "step": 15548
    },
    {
      "epoch": 2.129416598192276,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005989928547487881,
      "loss": 4.7588,
      "step": 15549
    },
    {
      "epoch": 2.129553546973432,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005989541992686161,
      "loss": 4.7111,
      "step": 15550
    },
    {
      "epoch": 2.1296904957545877,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005989155425540248,
      "loss": 4.7872,
      "step": 15551
    },
    {
      "epoch": 2.1298274445357435,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005988768846053351,
      "loss": 4.8235,
      "step": 15552
    },
    {
      "epoch": 2.1299643933168997,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005988382254228672,
      "loss": 4.7563,
      "step": 15553
    },
    {
      "epoch": 2.1301013420980555,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005987995650069414,
      "loss": 4.7526,
      "step": 15554
    },
    {
      "epoch": 2.1302382908792112,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005987609033578782,
      "loss": 4.7206,
      "step": 15555
    },
    {
      "epoch": 2.130375239660367,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005987222404759978,
      "loss": 4.7492,
      "step": 15556
    },
    {
      "epoch": 2.130512188441523,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005986835763616209,
      "loss": 4.7801,
      "step": 15557
    },
    {
      "epoch": 2.1306491372226786,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005986449110150677,
      "loss": 4.6778,
      "step": 15558
    },
    {
      "epoch": 2.130786086003835,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005986062444366589,
      "loss": 4.7163,
      "step": 15559
    },
    {
      "epoch": 2.1309230347849906,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005985675766267144,
      "loss": 4.7476,
      "step": 15560
    },
    {
      "epoch": 2.1310599835661463,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005985289075855552,
      "loss": 4.766,
      "step": 15561
    },
    {
      "epoch": 2.131196932347302,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005984902373135016,
      "loss": 4.8115,
      "step": 15562
    },
    {
      "epoch": 2.131333881128458,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005984515658108738,
      "loss": 4.7525,
      "step": 15563
    },
    {
      "epoch": 2.1314708299096137,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005984128930779928,
      "loss": 4.7079,
      "step": 15564
    },
    {
      "epoch": 2.1316077786907695,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005983742191151786,
      "loss": 4.6873,
      "step": 15565
    },
    {
      "epoch": 2.1317447274719257,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005983355439227519,
      "loss": 4.7678,
      "step": 15566
    },
    {
      "epoch": 2.1318816762530814,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005982968675010331,
      "loss": 4.7576,
      "step": 15567
    },
    {
      "epoch": 2.1320186250342372,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005982581898503431,
      "loss": 4.7153,
      "step": 15568
    },
    {
      "epoch": 2.132155573815393,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005982195109710019,
      "loss": 4.7378,
      "step": 15569
    },
    {
      "epoch": 2.132292522596549,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005981808308633303,
      "loss": 4.75,
      "step": 15570
    },
    {
      "epoch": 2.1324294713777046,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000598142149527649,
      "loss": 4.7109,
      "step": 15571
    },
    {
      "epoch": 2.1325664201588608,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005981034669642783,
      "loss": 4.8071,
      "step": 15572
    },
    {
      "epoch": 2.1327033689400166,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000598064783173539,
      "loss": 4.7841,
      "step": 15573
    },
    {
      "epoch": 2.1328403177211723,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005980260981557514,
      "loss": 4.6987,
      "step": 15574
    },
    {
      "epoch": 2.132977266502328,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005979874119112364,
      "loss": 4.7336,
      "step": 15575
    },
    {
      "epoch": 2.133114215283484,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005979487244403143,
      "loss": 4.7459,
      "step": 15576
    },
    {
      "epoch": 2.1332511640646397,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000597910035743306,
      "loss": 4.729,
      "step": 15577
    },
    {
      "epoch": 2.133388112845796,
      "grad_norm": 0.375,
      "learning_rate": 0.000597871345820532,
      "loss": 4.7624,
      "step": 15578
    },
    {
      "epoch": 2.1335250616269517,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0005978326546723127,
      "loss": 4.7502,
      "step": 15579
    },
    {
      "epoch": 2.1336620104081074,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005977939622989692,
      "loss": 4.7246,
      "step": 15580
    },
    {
      "epoch": 2.133798959189263,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005977552687008218,
      "loss": 4.8635,
      "step": 15581
    },
    {
      "epoch": 2.133935907970419,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005977165738781915,
      "loss": 4.7741,
      "step": 15582
    },
    {
      "epoch": 2.1340728567515748,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005976778778313985,
      "loss": 4.8345,
      "step": 15583
    },
    {
      "epoch": 2.1342098055327305,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005976391805607637,
      "loss": 4.7273,
      "step": 15584
    },
    {
      "epoch": 2.1343467543138868,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005976004820666081,
      "loss": 4.7343,
      "step": 15585
    },
    {
      "epoch": 2.1344837030950425,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005975617823492518,
      "loss": 4.7969,
      "step": 15586
    },
    {
      "epoch": 2.1346206518761983,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005975230814090161,
      "loss": 4.7754,
      "step": 15587
    },
    {
      "epoch": 2.134757600657354,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005974843792462214,
      "loss": 4.7873,
      "step": 15588
    },
    {
      "epoch": 2.13489454943851,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005974456758611885,
      "loss": 4.7619,
      "step": 15589
    },
    {
      "epoch": 2.1350314982196656,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000597406971254238,
      "loss": 4.7411,
      "step": 15590
    },
    {
      "epoch": 2.135168447000822,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000597368265425691,
      "loss": 4.6489,
      "step": 15591
    },
    {
      "epoch": 2.1353053957819776,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005973295583758681,
      "loss": 4.7426,
      "step": 15592
    },
    {
      "epoch": 2.1354423445631334,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00059729085010509,
      "loss": 4.7547,
      "step": 15593
    },
    {
      "epoch": 2.135579293344289,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005972521406136775,
      "loss": 4.8172,
      "step": 15594
    },
    {
      "epoch": 2.135716242125445,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005972134299019514,
      "loss": 4.7656,
      "step": 15595
    },
    {
      "epoch": 2.1358531909066008,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005971747179702325,
      "loss": 4.6995,
      "step": 15596
    },
    {
      "epoch": 2.135990139687757,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005971360048188417,
      "loss": 4.6625,
      "step": 15597
    },
    {
      "epoch": 2.1361270884689127,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005970972904480997,
      "loss": 4.7682,
      "step": 15598
    },
    {
      "epoch": 2.1362640372500685,
      "grad_norm": 0.375,
      "learning_rate": 0.0005970585748583275,
      "loss": 4.7405,
      "step": 15599
    },
    {
      "epoch": 2.1364009860312243,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005970198580498458,
      "loss": 4.6976,
      "step": 15600
    },
    {
      "epoch": 2.13653793481238,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005969811400229754,
      "loss": 4.774,
      "step": 15601
    },
    {
      "epoch": 2.136674883593536,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005969424207780375,
      "loss": 4.8185,
      "step": 15602
    },
    {
      "epoch": 2.1368118323746916,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005969037003153527,
      "loss": 4.8346,
      "step": 15603
    },
    {
      "epoch": 2.136948781155848,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005968649786352418,
      "loss": 4.7206,
      "step": 15604
    },
    {
      "epoch": 2.1370857299370036,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005968262557380258,
      "loss": 4.7327,
      "step": 15605
    },
    {
      "epoch": 2.1372226787181594,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005967875316240258,
      "loss": 4.7904,
      "step": 15606
    },
    {
      "epoch": 2.137359627499315,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005967488062935623,
      "loss": 4.7736,
      "step": 15607
    },
    {
      "epoch": 2.137496576280471,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005967100797469567,
      "loss": 4.7615,
      "step": 15608
    },
    {
      "epoch": 2.137633525061627,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005966713519845296,
      "loss": 4.7446,
      "step": 15609
    },
    {
      "epoch": 2.137770473842783,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005966326230066021,
      "loss": 4.7291,
      "step": 15610
    },
    {
      "epoch": 2.1379074226239387,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005965938928134952,
      "loss": 4.6787,
      "step": 15611
    },
    {
      "epoch": 2.1380443714050945,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005965551614055296,
      "loss": 4.7356,
      "step": 15612
    },
    {
      "epoch": 2.1381813201862503,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005965164287830266,
      "loss": 4.7818,
      "step": 15613
    },
    {
      "epoch": 2.138318268967406,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000596477694946307,
      "loss": 4.6684,
      "step": 15614
    },
    {
      "epoch": 2.138455217748562,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005964389598956918,
      "loss": 4.7406,
      "step": 15615
    },
    {
      "epoch": 2.138592166529718,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000596400223631502,
      "loss": 4.7108,
      "step": 15616
    },
    {
      "epoch": 2.138729115310874,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005963614861540587,
      "loss": 4.6658,
      "step": 15617
    },
    {
      "epoch": 2.1388660640920296,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000596322747463683,
      "loss": 4.7749,
      "step": 15618
    },
    {
      "epoch": 2.1390030128731854,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005962840075606957,
      "loss": 4.7973,
      "step": 15619
    },
    {
      "epoch": 2.139139961654341,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000596245266445418,
      "loss": 4.7077,
      "step": 15620
    },
    {
      "epoch": 2.139276910435497,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005962065241181709,
      "loss": 4.757,
      "step": 15621
    },
    {
      "epoch": 2.139413859216653,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005961677805792757,
      "loss": 4.7527,
      "step": 15622
    },
    {
      "epoch": 2.139550807997809,
      "grad_norm": 0.359375,
      "learning_rate": 0.000596129035829053,
      "loss": 4.7482,
      "step": 15623
    },
    {
      "epoch": 2.1396877567789647,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005960902898678244,
      "loss": 4.8145,
      "step": 15624
    },
    {
      "epoch": 2.1398247055601205,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005960515426959105,
      "loss": 4.764,
      "step": 15625
    },
    {
      "epoch": 2.1399616543412763,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005960127943136328,
      "loss": 4.722,
      "step": 15626
    },
    {
      "epoch": 2.140098603122432,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005959740447213124,
      "loss": 4.714,
      "step": 15627
    },
    {
      "epoch": 2.1402355519035883,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00059593529391927,
      "loss": 4.7608,
      "step": 15628
    },
    {
      "epoch": 2.140372500684744,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005958965419078273,
      "loss": 4.7652,
      "step": 15629
    },
    {
      "epoch": 2.1405094494659,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000595857788687305,
      "loss": 4.7338,
      "step": 15630
    },
    {
      "epoch": 2.1406463982470556,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005958190342580246,
      "loss": 4.7794,
      "step": 15631
    },
    {
      "epoch": 2.1407833470282114,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005957802786203071,
      "loss": 4.7088,
      "step": 15632
    },
    {
      "epoch": 2.140920295809367,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005957415217744738,
      "loss": 4.7375,
      "step": 15633
    },
    {
      "epoch": 2.141057244590523,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005957027637208455,
      "loss": 4.7747,
      "step": 15634
    },
    {
      "epoch": 2.141194193371679,
      "grad_norm": 0.375,
      "learning_rate": 0.0005956640044597439,
      "loss": 4.7056,
      "step": 15635
    },
    {
      "epoch": 2.141331142152835,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005956252439914899,
      "loss": 4.7163,
      "step": 15636
    },
    {
      "epoch": 2.1414680909339907,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005955864823164048,
      "loss": 4.6343,
      "step": 15637
    },
    {
      "epoch": 2.1416050397151465,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005955477194348099,
      "loss": 4.7532,
      "step": 15638
    },
    {
      "epoch": 2.1417419884963023,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005955089553470263,
      "loss": 4.7105,
      "step": 15639
    },
    {
      "epoch": 2.141878937277458,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005954701900533753,
      "loss": 4.6967,
      "step": 15640
    },
    {
      "epoch": 2.1420158860586143,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005954314235541783,
      "loss": 4.7005,
      "step": 15641
    },
    {
      "epoch": 2.14215283483977,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005953926558497564,
      "loss": 4.7596,
      "step": 15642
    },
    {
      "epoch": 2.142289783620926,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005953538869404307,
      "loss": 4.7117,
      "step": 15643
    },
    {
      "epoch": 2.1424267324020816,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005953151168265229,
      "loss": 4.6981,
      "step": 15644
    },
    {
      "epoch": 2.1425636811832374,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000595276345508354,
      "loss": 4.6918,
      "step": 15645
    },
    {
      "epoch": 2.142700629964393,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005952375729862455,
      "loss": 4.806,
      "step": 15646
    },
    {
      "epoch": 2.1428375787455494,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005951987992605186,
      "loss": 4.8353,
      "step": 15647
    },
    {
      "epoch": 2.142974527526705,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005951600243314946,
      "loss": 4.7042,
      "step": 15648
    },
    {
      "epoch": 2.143111476307861,
      "grad_norm": 0.375,
      "learning_rate": 0.000595121248199495,
      "loss": 4.7564,
      "step": 15649
    },
    {
      "epoch": 2.1432484250890167,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000595082470864841,
      "loss": 4.8516,
      "step": 15650
    },
    {
      "epoch": 2.1433853738701725,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005950436923278539,
      "loss": 4.7132,
      "step": 15651
    },
    {
      "epoch": 2.1435223226513282,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005950049125888552,
      "loss": 4.7495,
      "step": 15652
    },
    {
      "epoch": 2.143659271432484,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005949661316481663,
      "loss": 4.7134,
      "step": 15653
    },
    {
      "epoch": 2.1437962202136402,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005949273495061085,
      "loss": 4.6959,
      "step": 15654
    },
    {
      "epoch": 2.143933168994796,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005948885661630031,
      "loss": 4.6926,
      "step": 15655
    },
    {
      "epoch": 2.144070117775952,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005948497816191718,
      "loss": 4.7257,
      "step": 15656
    },
    {
      "epoch": 2.1442070665571076,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005948109958749358,
      "loss": 4.6996,
      "step": 15657
    },
    {
      "epoch": 2.1443440153382634,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005947722089306164,
      "loss": 4.7628,
      "step": 15658
    },
    {
      "epoch": 2.144480964119419,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005947334207865355,
      "loss": 4.7288,
      "step": 15659
    },
    {
      "epoch": 2.1446179129005754,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005946946314430141,
      "loss": 4.769,
      "step": 15660
    },
    {
      "epoch": 2.144754861681731,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005946558409003738,
      "loss": 4.7435,
      "step": 15661
    },
    {
      "epoch": 2.144891810462887,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005946170491589361,
      "loss": 4.7337,
      "step": 15662
    },
    {
      "epoch": 2.1450287592440427,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005945782562190224,
      "loss": 4.7868,
      "step": 15663
    },
    {
      "epoch": 2.1451657080251985,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005945394620809542,
      "loss": 4.6742,
      "step": 15664
    },
    {
      "epoch": 2.1453026568063542,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005945006667450531,
      "loss": 4.7603,
      "step": 15665
    },
    {
      "epoch": 2.1454396055875105,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005944618702116406,
      "loss": 4.6749,
      "step": 15666
    },
    {
      "epoch": 2.1455765543686662,
      "grad_norm": 0.375,
      "learning_rate": 0.0005944230724810382,
      "loss": 4.7239,
      "step": 15667
    },
    {
      "epoch": 2.145713503149822,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005943842735535673,
      "loss": 4.731,
      "step": 15668
    },
    {
      "epoch": 2.145850451930978,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005943454734295495,
      "loss": 4.8121,
      "step": 15669
    },
    {
      "epoch": 2.1459874007121336,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005943066721093063,
      "loss": 4.6492,
      "step": 15670
    },
    {
      "epoch": 2.1461243494932893,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005942678695931595,
      "loss": 4.6392,
      "step": 15671
    },
    {
      "epoch": 2.1462612982744456,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005942290658814303,
      "loss": 4.6704,
      "step": 15672
    },
    {
      "epoch": 2.1463982470556013,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005941902609744406,
      "loss": 4.8394,
      "step": 15673
    },
    {
      "epoch": 2.146535195836757,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005941514548725118,
      "loss": 4.8448,
      "step": 15674
    },
    {
      "epoch": 2.146672144617913,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005941126475759654,
      "loss": 4.676,
      "step": 15675
    },
    {
      "epoch": 2.1468090933990687,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005940738390851236,
      "loss": 4.7451,
      "step": 15676
    },
    {
      "epoch": 2.1469460421802244,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005940350294003071,
      "loss": 4.7572,
      "step": 15677
    },
    {
      "epoch": 2.1470829909613807,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005939962185218382,
      "loss": 4.7129,
      "step": 15678
    },
    {
      "epoch": 2.1472199397425364,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005939574064500382,
      "loss": 4.7776,
      "step": 15679
    },
    {
      "epoch": 2.147356888523692,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005939185931852292,
      "loss": 4.7013,
      "step": 15680
    },
    {
      "epoch": 2.147493837304848,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005938797787277321,
      "loss": 4.7922,
      "step": 15681
    },
    {
      "epoch": 2.1476307860860038,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005938409630778691,
      "loss": 4.7994,
      "step": 15682
    },
    {
      "epoch": 2.1477677348671596,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000593802146235962,
      "loss": 4.7315,
      "step": 15683
    },
    {
      "epoch": 2.1479046836483153,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005937633282023319,
      "loss": 4.7618,
      "step": 15684
    },
    {
      "epoch": 2.1480416324294715,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005937245089773011,
      "loss": 4.7961,
      "step": 15685
    },
    {
      "epoch": 2.1481785812106273,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005936856885611909,
      "loss": 4.7437,
      "step": 15686
    },
    {
      "epoch": 2.148315529991783,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005936468669543233,
      "loss": 4.7236,
      "step": 15687
    },
    {
      "epoch": 2.148452478772939,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005936080441570198,
      "loss": 4.6869,
      "step": 15688
    },
    {
      "epoch": 2.1485894275540947,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005935692201696022,
      "loss": 4.6983,
      "step": 15689
    },
    {
      "epoch": 2.1487263763352504,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005935303949923923,
      "loss": 4.8139,
      "step": 15690
    },
    {
      "epoch": 2.1488633251164067,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005934915686257119,
      "loss": 4.7067,
      "step": 15691
    },
    {
      "epoch": 2.1490002738975624,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005934527410698826,
      "loss": 4.7149,
      "step": 15692
    },
    {
      "epoch": 2.149137222678718,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005934139123252261,
      "loss": 4.7296,
      "step": 15693
    },
    {
      "epoch": 2.149274171459874,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005933750823920646,
      "loss": 4.71,
      "step": 15694
    },
    {
      "epoch": 2.1494111202410298,
      "grad_norm": 0.375,
      "learning_rate": 0.0005933362512707194,
      "loss": 4.7015,
      "step": 15695
    },
    {
      "epoch": 2.1495480690221855,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005932974189615127,
      "loss": 4.7148,
      "step": 15696
    },
    {
      "epoch": 2.1496850178033418,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000593258585464766,
      "loss": 4.7459,
      "step": 15697
    },
    {
      "epoch": 2.1498219665844975,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005932197507808014,
      "loss": 4.7193,
      "step": 15698
    },
    {
      "epoch": 2.1499589153656533,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005931809149099406,
      "loss": 4.756,
      "step": 15699
    },
    {
      "epoch": 2.150095864146809,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005931420778525053,
      "loss": 4.699,
      "step": 15700
    },
    {
      "epoch": 2.150232812927965,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005931032396088175,
      "loss": 4.7936,
      "step": 15701
    },
    {
      "epoch": 2.1503697617091206,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005930644001791992,
      "loss": 4.7394,
      "step": 15702
    },
    {
      "epoch": 2.1505067104902764,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000593025559563972,
      "loss": 4.69,
      "step": 15703
    },
    {
      "epoch": 2.1506436592714326,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005929867177634579,
      "loss": 4.8264,
      "step": 15704
    },
    {
      "epoch": 2.1507806080525884,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005929478747779789,
      "loss": 4.7338,
      "step": 15705
    },
    {
      "epoch": 2.150917556833744,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005929090306078568,
      "loss": 4.7564,
      "step": 15706
    },
    {
      "epoch": 2.1510545056149,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005928701852534135,
      "loss": 4.6984,
      "step": 15707
    },
    {
      "epoch": 2.1511914543960557,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005928313387149708,
      "loss": 4.7365,
      "step": 15708
    },
    {
      "epoch": 2.1513284031772115,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000592792490992851,
      "loss": 4.6895,
      "step": 15709
    },
    {
      "epoch": 2.1514653519583677,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005927536420873755,
      "loss": 4.6331,
      "step": 15710
    },
    {
      "epoch": 2.1516023007395235,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005927147919988667,
      "loss": 4.7465,
      "step": 15711
    },
    {
      "epoch": 2.1517392495206793,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005926759407276464,
      "loss": 4.7107,
      "step": 15712
    },
    {
      "epoch": 2.151876198301835,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005926370882740367,
      "loss": 4.7283,
      "step": 15713
    },
    {
      "epoch": 2.152013147082991,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005925982346383594,
      "loss": 4.7054,
      "step": 15714
    },
    {
      "epoch": 2.1521500958641466,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005925593798209364,
      "loss": 4.7438,
      "step": 15715
    },
    {
      "epoch": 2.152287044645303,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00059252052382209,
      "loss": 4.7487,
      "step": 15716
    },
    {
      "epoch": 2.1524239934264586,
      "grad_norm": 0.34375,
      "learning_rate": 0.000592481666642142,
      "loss": 4.7599,
      "step": 15717
    },
    {
      "epoch": 2.1525609422076144,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005924428082814144,
      "loss": 4.7342,
      "step": 15718
    },
    {
      "epoch": 2.15269789098877,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005924039487402295,
      "loss": 4.7821,
      "step": 15719
    },
    {
      "epoch": 2.152834839769926,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000592365088018909,
      "loss": 4.6826,
      "step": 15720
    },
    {
      "epoch": 2.1529717885510817,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005923262261177751,
      "loss": 4.7103,
      "step": 15721
    },
    {
      "epoch": 2.1531087373322375,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00059228736303715,
      "loss": 4.7555,
      "step": 15722
    },
    {
      "epoch": 2.1532456861133937,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005922484987773554,
      "loss": 4.7674,
      "step": 15723
    },
    {
      "epoch": 2.1533826348945495,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005922096333387137,
      "loss": 4.6795,
      "step": 15724
    },
    {
      "epoch": 2.1535195836757053,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000592170766721547,
      "loss": 4.7484,
      "step": 15725
    },
    {
      "epoch": 2.153656532456861,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005921318989261771,
      "loss": 4.7958,
      "step": 15726
    },
    {
      "epoch": 2.153793481238017,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005920930299529263,
      "loss": 4.8326,
      "step": 15727
    },
    {
      "epoch": 2.153930430019173,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005920541598021168,
      "loss": 4.7379,
      "step": 15728
    },
    {
      "epoch": 2.154067378800329,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005920152884740706,
      "loss": 4.7436,
      "step": 15729
    },
    {
      "epoch": 2.1542043275814846,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005919764159691098,
      "loss": 4.771,
      "step": 15730
    },
    {
      "epoch": 2.1543412763626404,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005919375422875566,
      "loss": 4.81,
      "step": 15731
    },
    {
      "epoch": 2.154478225143796,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005918986674297332,
      "loss": 4.7244,
      "step": 15732
    },
    {
      "epoch": 2.154615173924952,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005918597913959617,
      "loss": 4.7604,
      "step": 15733
    },
    {
      "epoch": 2.1547521227061077,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005918209141865645,
      "loss": 4.7732,
      "step": 15734
    },
    {
      "epoch": 2.154889071487264,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005917820358018633,
      "loss": 4.7589,
      "step": 15735
    },
    {
      "epoch": 2.1550260202684197,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005917431562421808,
      "loss": 4.7227,
      "step": 15736
    },
    {
      "epoch": 2.1551629690495755,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000591704275507839,
      "loss": 4.726,
      "step": 15737
    },
    {
      "epoch": 2.1552999178307313,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005916653935991599,
      "loss": 4.7551,
      "step": 15738
    },
    {
      "epoch": 2.155436866611887,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005916265105164661,
      "loss": 4.7791,
      "step": 15739
    },
    {
      "epoch": 2.155573815393043,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005915876262600795,
      "loss": 4.7343,
      "step": 15740
    },
    {
      "epoch": 2.155710764174199,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005915487408303228,
      "loss": 4.7382,
      "step": 15741
    },
    {
      "epoch": 2.155847712955355,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005915098542275177,
      "loss": 4.6974,
      "step": 15742
    },
    {
      "epoch": 2.1559846617365106,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005914709664519868,
      "loss": 4.729,
      "step": 15743
    },
    {
      "epoch": 2.1561216105176664,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005914320775040521,
      "loss": 4.6837,
      "step": 15744
    },
    {
      "epoch": 2.156258559298822,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005913931873840363,
      "loss": 4.7767,
      "step": 15745
    },
    {
      "epoch": 2.156395508079978,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005913542960922614,
      "loss": 4.6368,
      "step": 15746
    },
    {
      "epoch": 2.156532456861134,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005913154036290498,
      "loss": 4.7374,
      "step": 15747
    },
    {
      "epoch": 2.15666940564229,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005912765099947237,
      "loss": 4.7379,
      "step": 15748
    },
    {
      "epoch": 2.1568063544234457,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005912376151896054,
      "loss": 4.7359,
      "step": 15749
    },
    {
      "epoch": 2.1569433032046015,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005911987192140177,
      "loss": 4.7278,
      "step": 15750
    },
    {
      "epoch": 2.1570802519857573,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005911598220682822,
      "loss": 4.6389,
      "step": 15751
    },
    {
      "epoch": 2.157217200766913,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005911209237527219,
      "loss": 4.6737,
      "step": 15752
    },
    {
      "epoch": 2.157354149548069,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005910820242676587,
      "loss": 4.7742,
      "step": 15753
    },
    {
      "epoch": 2.157491098329225,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005910431236134152,
      "loss": 4.7487,
      "step": 15754
    },
    {
      "epoch": 2.157628047110381,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005910042217903136,
      "loss": 4.7662,
      "step": 15755
    },
    {
      "epoch": 2.1577649958915366,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005909653187986765,
      "loss": 4.7188,
      "step": 15756
    },
    {
      "epoch": 2.1579019446726924,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005909264146388263,
      "loss": 4.7653,
      "step": 15757
    },
    {
      "epoch": 2.158038893453848,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005908875093110852,
      "loss": 4.7647,
      "step": 15758
    },
    {
      "epoch": 2.158175842235004,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005908486028157759,
      "loss": 4.6862,
      "step": 15759
    },
    {
      "epoch": 2.15831279101616,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005908096951532204,
      "loss": 4.6338,
      "step": 15760
    },
    {
      "epoch": 2.158449739797316,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005907707863237417,
      "loss": 4.7321,
      "step": 15761
    },
    {
      "epoch": 2.1585866885784717,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005907318763276617,
      "loss": 4.8247,
      "step": 15762
    },
    {
      "epoch": 2.1587236373596275,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005906929651653033,
      "loss": 4.6186,
      "step": 15763
    },
    {
      "epoch": 2.1588605861407832,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005906540528369886,
      "loss": 4.6941,
      "step": 15764
    },
    {
      "epoch": 2.158997534921939,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005906151393430404,
      "loss": 4.7823,
      "step": 15765
    },
    {
      "epoch": 2.1591344837030952,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005905762246837809,
      "loss": 4.7075,
      "step": 15766
    },
    {
      "epoch": 2.159271432484251,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005905373088595328,
      "loss": 4.7242,
      "step": 15767
    },
    {
      "epoch": 2.159408381265407,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005904983918706186,
      "loss": 4.6644,
      "step": 15768
    },
    {
      "epoch": 2.1595453300465626,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005904594737173607,
      "loss": 4.677,
      "step": 15769
    },
    {
      "epoch": 2.1596822788277183,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005904205544000816,
      "loss": 4.7746,
      "step": 15770
    },
    {
      "epoch": 2.159819227608874,
      "grad_norm": 0.451171875,
      "learning_rate": 0.000590381633919104,
      "loss": 4.837,
      "step": 15771
    },
    {
      "epoch": 2.15995617639003,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005903427122747504,
      "loss": 4.795,
      "step": 15772
    },
    {
      "epoch": 2.160093125171186,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005903037894673431,
      "loss": 4.8,
      "step": 15773
    },
    {
      "epoch": 2.160230073952342,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000590264865497205,
      "loss": 4.8067,
      "step": 15774
    },
    {
      "epoch": 2.1603670227334977,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005902259403646585,
      "loss": 4.6781,
      "step": 15775
    },
    {
      "epoch": 2.1605039715146535,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005901870140700261,
      "loss": 4.7766,
      "step": 15776
    },
    {
      "epoch": 2.1606409202958092,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005901480866136307,
      "loss": 4.6722,
      "step": 15777
    },
    {
      "epoch": 2.160777869076965,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005901091579957946,
      "loss": 4.8299,
      "step": 15778
    },
    {
      "epoch": 2.1609148178581212,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005900702282168407,
      "loss": 4.7614,
      "step": 15779
    },
    {
      "epoch": 2.161051766639277,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005900312972770913,
      "loss": 4.7267,
      "step": 15780
    },
    {
      "epoch": 2.161188715420433,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005899923651768692,
      "loss": 4.6929,
      "step": 15781
    },
    {
      "epoch": 2.1613256642015886,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000589953431916497,
      "loss": 4.6823,
      "step": 15782
    },
    {
      "epoch": 2.1614626129827443,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005899144974962974,
      "loss": 4.7099,
      "step": 15783
    },
    {
      "epoch": 2.1615995617639,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005898755619165929,
      "loss": 4.7492,
      "step": 15784
    },
    {
      "epoch": 2.1617365105450563,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005898366251777063,
      "loss": 4.7099,
      "step": 15785
    },
    {
      "epoch": 2.161873459326212,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005897976872799604,
      "loss": 4.6926,
      "step": 15786
    },
    {
      "epoch": 2.162010408107368,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005897587482236776,
      "loss": 4.6996,
      "step": 15787
    },
    {
      "epoch": 2.1621473568885237,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005897198080091809,
      "loss": 4.7592,
      "step": 15788
    },
    {
      "epoch": 2.1622843056696794,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005896808666367927,
      "loss": 4.7654,
      "step": 15789
    },
    {
      "epoch": 2.162421254450835,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005896419241068361,
      "loss": 4.6599,
      "step": 15790
    },
    {
      "epoch": 2.1625582032319914,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005896029804196333,
      "loss": 4.7671,
      "step": 15791
    },
    {
      "epoch": 2.162695152013147,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005895640355755077,
      "loss": 4.6747,
      "step": 15792
    },
    {
      "epoch": 2.162832100794303,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005895250895747813,
      "loss": 4.6815,
      "step": 15793
    },
    {
      "epoch": 2.1629690495754588,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005894861424177775,
      "loss": 4.7829,
      "step": 15794
    },
    {
      "epoch": 2.1631059983566145,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005894471941048187,
      "loss": 4.7781,
      "step": 15795
    },
    {
      "epoch": 2.1632429471377703,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005894082446362278,
      "loss": 4.6824,
      "step": 15796
    },
    {
      "epoch": 2.1633798959189265,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005893692940123276,
      "loss": 4.7617,
      "step": 15797
    },
    {
      "epoch": 2.1635168447000823,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005893303422334407,
      "loss": 4.7201,
      "step": 15798
    },
    {
      "epoch": 2.163653793481238,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005892913892998902,
      "loss": 4.7174,
      "step": 15799
    },
    {
      "epoch": 2.163790742262394,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005892524352119987,
      "loss": 4.7931,
      "step": 15800
    },
    {
      "epoch": 2.1639276910435497,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005892134799700892,
      "loss": 4.7273,
      "step": 15801
    },
    {
      "epoch": 2.1640646398247054,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005891745235744843,
      "loss": 4.797,
      "step": 15802
    },
    {
      "epoch": 2.164201588605861,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000589135566025507,
      "loss": 4.6999,
      "step": 15803
    },
    {
      "epoch": 2.1643385373870174,
      "grad_norm": 0.359375,
      "learning_rate": 0.00058909660732348,
      "loss": 4.7087,
      "step": 15804
    },
    {
      "epoch": 2.164475486168173,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005890576474687263,
      "loss": 4.6502,
      "step": 15805
    },
    {
      "epoch": 2.164612434949329,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000589018686461569,
      "loss": 4.7747,
      "step": 15806
    },
    {
      "epoch": 2.1647493837304848,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005889797243023304,
      "loss": 4.7933,
      "step": 15807
    },
    {
      "epoch": 2.1648863325116405,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005889407609913339,
      "loss": 4.8078,
      "step": 15808
    },
    {
      "epoch": 2.1650232812927963,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005889017965289022,
      "loss": 4.7923,
      "step": 15809
    },
    {
      "epoch": 2.1651602300739525,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005888628309153579,
      "loss": 4.7327,
      "step": 15810
    },
    {
      "epoch": 2.1652971788551083,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005888238641510246,
      "loss": 4.6531,
      "step": 15811
    },
    {
      "epoch": 2.165434127636264,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005887848962362247,
      "loss": 4.7735,
      "step": 15812
    },
    {
      "epoch": 2.16557107641742,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005887459271712814,
      "loss": 4.7708,
      "step": 15813
    },
    {
      "epoch": 2.1657080251985756,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005887069569565174,
      "loss": 4.7674,
      "step": 15814
    },
    {
      "epoch": 2.1658449739797314,
      "grad_norm": 0.359375,
      "learning_rate": 0.000588667985592256,
      "loss": 4.6892,
      "step": 15815
    },
    {
      "epoch": 2.1659819227608876,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005886290130788197,
      "loss": 4.7398,
      "step": 15816
    },
    {
      "epoch": 2.1661188715420434,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005885900394165318,
      "loss": 4.7261,
      "step": 15817
    },
    {
      "epoch": 2.166255820323199,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005885510646057154,
      "loss": 4.7649,
      "step": 15818
    },
    {
      "epoch": 2.166392769104355,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005885120886466931,
      "loss": 4.7261,
      "step": 15819
    },
    {
      "epoch": 2.1665297178855107,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005884731115397883,
      "loss": 4.8032,
      "step": 15820
    },
    {
      "epoch": 2.1666666666666665,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005884341332853237,
      "loss": 4.7351,
      "step": 15821
    },
    {
      "epoch": 2.1668036154478223,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005883951538836226,
      "loss": 4.7504,
      "step": 15822
    },
    {
      "epoch": 2.1669405642289785,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005883561733350076,
      "loss": 4.7259,
      "step": 15823
    },
    {
      "epoch": 2.1670775130101343,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005883171916398022,
      "loss": 4.7822,
      "step": 15824
    },
    {
      "epoch": 2.16721446179129,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005882782087983293,
      "loss": 4.7795,
      "step": 15825
    },
    {
      "epoch": 2.167351410572446,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005882392248109118,
      "loss": 4.6927,
      "step": 15826
    },
    {
      "epoch": 2.1674883593536016,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005882002396778731,
      "loss": 4.7063,
      "step": 15827
    },
    {
      "epoch": 2.1676253081347574,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000588161253399536,
      "loss": 4.7621,
      "step": 15828
    },
    {
      "epoch": 2.1677622569159136,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005881222659762237,
      "loss": 4.8244,
      "step": 15829
    },
    {
      "epoch": 2.1678992056970694,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005880832774082592,
      "loss": 4.7818,
      "step": 15830
    },
    {
      "epoch": 2.168036154478225,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005880442876959657,
      "loss": 4.6973,
      "step": 15831
    },
    {
      "epoch": 2.168173103259381,
      "grad_norm": 0.3125,
      "learning_rate": 0.0005880052968396663,
      "loss": 4.7019,
      "step": 15832
    },
    {
      "epoch": 2.1683100520405367,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005879663048396841,
      "loss": 4.8038,
      "step": 15833
    },
    {
      "epoch": 2.1684470008216925,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005879273116963424,
      "loss": 4.8575,
      "step": 15834
    },
    {
      "epoch": 2.1685839496028487,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005878883174099639,
      "loss": 4.7819,
      "step": 15835
    },
    {
      "epoch": 2.1687208983840045,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005878493219808724,
      "loss": 4.7846,
      "step": 15836
    },
    {
      "epoch": 2.1688578471651603,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005878103254093904,
      "loss": 4.7796,
      "step": 15837
    },
    {
      "epoch": 2.168994795946316,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005877713276958416,
      "loss": 4.7447,
      "step": 15838
    },
    {
      "epoch": 2.169131744727472,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005877323288405489,
      "loss": 4.7949,
      "step": 15839
    },
    {
      "epoch": 2.1692686935086276,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005876933288438355,
      "loss": 4.7516,
      "step": 15840
    },
    {
      "epoch": 2.1694056422897834,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0005876543277060249,
      "loss": 4.8275,
      "step": 15841
    },
    {
      "epoch": 2.1695425910709396,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00058761532542744,
      "loss": 4.6689,
      "step": 15842
    },
    {
      "epoch": 2.1696795398520954,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000587576322008404,
      "loss": 4.7679,
      "step": 15843
    },
    {
      "epoch": 2.169816488633251,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005875373174492404,
      "loss": 4.6529,
      "step": 15844
    },
    {
      "epoch": 2.169953437414407,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005874983117502721,
      "loss": 4.8145,
      "step": 15845
    },
    {
      "epoch": 2.1700903861955627,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005874593049118226,
      "loss": 4.757,
      "step": 15846
    },
    {
      "epoch": 2.170227334976719,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000587420296934215,
      "loss": 4.7877,
      "step": 15847
    },
    {
      "epoch": 2.1703642837578747,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005873812878177729,
      "loss": 4.8083,
      "step": 15848
    },
    {
      "epoch": 2.1705012325390305,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005873422775628191,
      "loss": 4.7975,
      "step": 15849
    },
    {
      "epoch": 2.1706381813201863,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005873032661696772,
      "loss": 4.6673,
      "step": 15850
    },
    {
      "epoch": 2.170775130101342,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005872642536386704,
      "loss": 4.6455,
      "step": 15851
    },
    {
      "epoch": 2.170912078882498,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005872252399701219,
      "loss": 4.6071,
      "step": 15852
    },
    {
      "epoch": 2.1710490276636536,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005871862251643554,
      "loss": 4.7075,
      "step": 15853
    },
    {
      "epoch": 2.17118597644481,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005871472092216938,
      "loss": 4.736,
      "step": 15854
    },
    {
      "epoch": 2.1713229252259656,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005871081921424607,
      "loss": 4.736,
      "step": 15855
    },
    {
      "epoch": 2.1714598740071214,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005870691739269792,
      "loss": 4.796,
      "step": 15856
    },
    {
      "epoch": 2.171596822788277,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005870301545755729,
      "loss": 4.7427,
      "step": 15857
    },
    {
      "epoch": 2.171733771569433,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000586991134088565,
      "loss": 4.7169,
      "step": 15858
    },
    {
      "epoch": 2.1718707203505887,
      "grad_norm": 0.34375,
      "learning_rate": 0.000586952112466279,
      "loss": 4.8183,
      "step": 15859
    },
    {
      "epoch": 2.172007669131745,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000586913089709038,
      "loss": 4.7153,
      "step": 15860
    },
    {
      "epoch": 2.1721446179129007,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005868740658171658,
      "loss": 4.7921,
      "step": 15861
    },
    {
      "epoch": 2.1722815666940565,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005868350407909857,
      "loss": 4.7415,
      "step": 15862
    },
    {
      "epoch": 2.1724185154752123,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005867960146308207,
      "loss": 4.7571,
      "step": 15863
    },
    {
      "epoch": 2.172555464256368,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005867569873369947,
      "loss": 4.748,
      "step": 15864
    },
    {
      "epoch": 2.172692413037524,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000586717958909831,
      "loss": 4.6565,
      "step": 15865
    },
    {
      "epoch": 2.17282936181868,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000586678929349653,
      "loss": 4.7559,
      "step": 15866
    },
    {
      "epoch": 2.172966310599836,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000586639898656784,
      "loss": 4.8042,
      "step": 15867
    },
    {
      "epoch": 2.1731032593809916,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005866008668315478,
      "loss": 4.8082,
      "step": 15868
    },
    {
      "epoch": 2.1732402081621474,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005865618338742676,
      "loss": 4.7437,
      "step": 15869
    },
    {
      "epoch": 2.173377156943303,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005865227997852668,
      "loss": 4.674,
      "step": 15870
    },
    {
      "epoch": 2.173514105724459,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005864837645648693,
      "loss": 4.6915,
      "step": 15871
    },
    {
      "epoch": 2.1736510545056147,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005864447282133981,
      "loss": 4.699,
      "step": 15872
    },
    {
      "epoch": 2.173788003286771,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005864056907311769,
      "loss": 4.7605,
      "step": 15873
    },
    {
      "epoch": 2.1739249520679267,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005863666521185294,
      "loss": 4.7479,
      "step": 15874
    },
    {
      "epoch": 2.1740619008490825,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005863276123757789,
      "loss": 4.7738,
      "step": 15875
    },
    {
      "epoch": 2.1741988496302382,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005862885715032488,
      "loss": 4.6684,
      "step": 15876
    },
    {
      "epoch": 2.174335798411394,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005862495295012631,
      "loss": 4.6949,
      "step": 15877
    },
    {
      "epoch": 2.17447274719255,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000586210486370145,
      "loss": 4.7476,
      "step": 15878
    },
    {
      "epoch": 2.174609695973706,
      "grad_norm": 0.375,
      "learning_rate": 0.000586171442110218,
      "loss": 4.8025,
      "step": 15879
    },
    {
      "epoch": 2.174746644754862,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0005861323967218061,
      "loss": 4.8005,
      "step": 15880
    },
    {
      "epoch": 2.1748835935360176,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005860933502052323,
      "loss": 4.7242,
      "step": 15881
    },
    {
      "epoch": 2.1750205423171733,
      "grad_norm": 0.5,
      "learning_rate": 0.0005860543025608207,
      "loss": 4.7596,
      "step": 15882
    },
    {
      "epoch": 2.175157491098329,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005860152537888945,
      "loss": 4.7398,
      "step": 15883
    },
    {
      "epoch": 2.175294439879485,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0005859762038897776,
      "loss": 4.7109,
      "step": 15884
    },
    {
      "epoch": 2.175431388660641,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005859371528637935,
      "loss": 4.7731,
      "step": 15885
    },
    {
      "epoch": 2.175568337441797,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005858981007112657,
      "loss": 4.8159,
      "step": 15886
    },
    {
      "epoch": 2.1757052862229527,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0005858590474325181,
      "loss": 4.7635,
      "step": 15887
    },
    {
      "epoch": 2.1758422350041084,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000585819993027874,
      "loss": 4.6896,
      "step": 15888
    },
    {
      "epoch": 2.1759791837852642,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005857809374976576,
      "loss": 4.7133,
      "step": 15889
    },
    {
      "epoch": 2.17611613256642,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005857418808421918,
      "loss": 4.7989,
      "step": 15890
    },
    {
      "epoch": 2.176253081347576,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000585702823061801,
      "loss": 4.7394,
      "step": 15891
    },
    {
      "epoch": 2.176390030128732,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005856637641568084,
      "loss": 4.7611,
      "step": 15892
    },
    {
      "epoch": 2.1765269789098878,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005856247041275379,
      "loss": 4.8127,
      "step": 15893
    },
    {
      "epoch": 2.1766639276910436,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005855856429743132,
      "loss": 4.7978,
      "step": 15894
    },
    {
      "epoch": 2.1768008764721993,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000585546580697458,
      "loss": 4.7317,
      "step": 15895
    },
    {
      "epoch": 2.176937825253355,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005855075172972959,
      "loss": 4.7437,
      "step": 15896
    },
    {
      "epoch": 2.1770747740345113,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005854684527741508,
      "loss": 4.7041,
      "step": 15897
    },
    {
      "epoch": 2.177211722815667,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005854293871283462,
      "loss": 4.6977,
      "step": 15898
    },
    {
      "epoch": 2.177348671596823,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005853903203602061,
      "loss": 4.7524,
      "step": 15899
    },
    {
      "epoch": 2.1774856203779787,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005853512524700543,
      "loss": 4.7938,
      "step": 15900
    },
    {
      "epoch": 2.1776225691591344,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005853121834582142,
      "loss": 4.6981,
      "step": 15901
    },
    {
      "epoch": 2.17775951794029,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005852731133250098,
      "loss": 4.7553,
      "step": 15902
    },
    {
      "epoch": 2.177896466721446,
      "grad_norm": 0.359375,
      "learning_rate": 0.000585234042070765,
      "loss": 4.7762,
      "step": 15903
    },
    {
      "epoch": 2.178033415502602,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005851949696958034,
      "loss": 4.7796,
      "step": 15904
    },
    {
      "epoch": 2.178170364283758,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005851558962004489,
      "loss": 4.7516,
      "step": 15905
    },
    {
      "epoch": 2.1783073130649138,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005851168215850252,
      "loss": 4.7775,
      "step": 15906
    },
    {
      "epoch": 2.1784442618460695,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005850777458498563,
      "loss": 4.742,
      "step": 15907
    },
    {
      "epoch": 2.1785812106272253,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005850386689952659,
      "loss": 4.7173,
      "step": 15908
    },
    {
      "epoch": 2.178718159408381,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005849995910215779,
      "loss": 4.7345,
      "step": 15909
    },
    {
      "epoch": 2.1788551081895373,
      "grad_norm": 0.375,
      "learning_rate": 0.000584960511929116,
      "loss": 4.6856,
      "step": 15910
    },
    {
      "epoch": 2.178992056970693,
      "grad_norm": 0.421875,
      "learning_rate": 0.0005849214317182043,
      "loss": 4.6944,
      "step": 15911
    },
    {
      "epoch": 2.179129005751849,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005848823503891665,
      "loss": 4.8094,
      "step": 15912
    },
    {
      "epoch": 2.1792659545330046,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005848432679423266,
      "loss": 4.7275,
      "step": 15913
    },
    {
      "epoch": 2.1794029033141604,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0005848041843780083,
      "loss": 4.7289,
      "step": 15914
    },
    {
      "epoch": 2.179539852095316,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005847650996965356,
      "loss": 4.752,
      "step": 15915
    },
    {
      "epoch": 2.1796768008764724,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005847260138982324,
      "loss": 4.8245,
      "step": 15916
    },
    {
      "epoch": 2.179813749657628,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005846869269834228,
      "loss": 4.8753,
      "step": 15917
    },
    {
      "epoch": 2.179950698438784,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005846478389524304,
      "loss": 4.7779,
      "step": 15918
    },
    {
      "epoch": 2.1800876472199398,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005846087498055793,
      "loss": 4.6963,
      "step": 15919
    },
    {
      "epoch": 2.1802245960010955,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005845696595431934,
      "loss": 4.7638,
      "step": 15920
    },
    {
      "epoch": 2.1803615447822513,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0005845305681655966,
      "loss": 4.7467,
      "step": 15921
    },
    {
      "epoch": 2.180498493563407,
      "grad_norm": 0.375,
      "learning_rate": 0.000584491475673113,
      "loss": 4.7015,
      "step": 15922
    },
    {
      "epoch": 2.1806354423445633,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005844523820660665,
      "loss": 4.6238,
      "step": 15923
    },
    {
      "epoch": 2.180772391125719,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005844132873447811,
      "loss": 4.6688,
      "step": 15924
    },
    {
      "epoch": 2.180909339906875,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005843741915095807,
      "loss": 4.8026,
      "step": 15925
    },
    {
      "epoch": 2.1810462886880306,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005843350945607894,
      "loss": 4.6432,
      "step": 15926
    },
    {
      "epoch": 2.1811832374691864,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005842959964987312,
      "loss": 4.6262,
      "step": 15927
    },
    {
      "epoch": 2.181320186250342,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005842568973237299,
      "loss": 4.7024,
      "step": 15928
    },
    {
      "epoch": 2.1814571350314984,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005842177970361099,
      "loss": 4.7534,
      "step": 15929
    },
    {
      "epoch": 2.181594083812654,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005841786956361949,
      "loss": 4.7666,
      "step": 15930
    },
    {
      "epoch": 2.18173103259381,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005841395931243091,
      "loss": 4.7856,
      "step": 15931
    },
    {
      "epoch": 2.1818679813749657,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005841004895007766,
      "loss": 4.8139,
      "step": 15932
    },
    {
      "epoch": 2.1820049301561215,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005840613847659212,
      "loss": 4.6826,
      "step": 15933
    },
    {
      "epoch": 2.1821418789372773,
      "grad_norm": 0.375,
      "learning_rate": 0.0005840222789200673,
      "loss": 4.8081,
      "step": 15934
    },
    {
      "epoch": 2.1822788277184335,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005839831719635388,
      "loss": 4.7467,
      "step": 15935
    },
    {
      "epoch": 2.1824157764995893,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00058394406389666,
      "loss": 4.715,
      "step": 15936
    },
    {
      "epoch": 2.182552725280745,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005839049547197545,
      "loss": 4.7829,
      "step": 15937
    },
    {
      "epoch": 2.182689674061901,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000583865844433147,
      "loss": 4.8069,
      "step": 15938
    },
    {
      "epoch": 2.1828266228430566,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005838267330371613,
      "loss": 4.7495,
      "step": 15939
    },
    {
      "epoch": 2.1829635716242124,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005837876205321217,
      "loss": 4.7699,
      "step": 15940
    },
    {
      "epoch": 2.183100520405368,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005837485069183518,
      "loss": 4.7829,
      "step": 15941
    },
    {
      "epoch": 2.1832374691865244,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005837093921961765,
      "loss": 4.7863,
      "step": 15942
    },
    {
      "epoch": 2.18337441796768,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005836702763659195,
      "loss": 4.757,
      "step": 15943
    },
    {
      "epoch": 2.183511366748836,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000583631159427905,
      "loss": 4.7788,
      "step": 15944
    },
    {
      "epoch": 2.1836483155299917,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005835920413824574,
      "loss": 4.7434,
      "step": 15945
    },
    {
      "epoch": 2.1837852643111475,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005835529222299006,
      "loss": 4.8287,
      "step": 15946
    },
    {
      "epoch": 2.1839222130923033,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000583513801970559,
      "loss": 4.7877,
      "step": 15947
    },
    {
      "epoch": 2.1840591618734595,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005834746806047566,
      "loss": 4.6821,
      "step": 15948
    },
    {
      "epoch": 2.1841961106546153,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005834355581328178,
      "loss": 4.7419,
      "step": 15949
    },
    {
      "epoch": 2.184333059435771,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005833964345550668,
      "loss": 4.7488,
      "step": 15950
    },
    {
      "epoch": 2.184470008216927,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005833573098718276,
      "loss": 4.7425,
      "step": 15951
    },
    {
      "epoch": 2.1846069569980826,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005833181840834246,
      "loss": 4.5889,
      "step": 15952
    },
    {
      "epoch": 2.1847439057792384,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000583279057190182,
      "loss": 4.7583,
      "step": 15953
    },
    {
      "epoch": 2.1848808545603946,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005832399291924243,
      "loss": 4.7083,
      "step": 15954
    },
    {
      "epoch": 2.1850178033415504,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005832008000904754,
      "loss": 4.6999,
      "step": 15955
    },
    {
      "epoch": 2.185154752122706,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005831616698846599,
      "loss": 4.6663,
      "step": 15956
    },
    {
      "epoch": 2.185291700903862,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005831225385753017,
      "loss": 4.6997,
      "step": 15957
    },
    {
      "epoch": 2.1854286496850177,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005830834061627255,
      "loss": 4.7191,
      "step": 15958
    },
    {
      "epoch": 2.1855655984661735,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005830442726472553,
      "loss": 4.7676,
      "step": 15959
    },
    {
      "epoch": 2.1857025472473293,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005830051380292155,
      "loss": 4.7675,
      "step": 15960
    },
    {
      "epoch": 2.1858394960284855,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005829660023089304,
      "loss": 4.8608,
      "step": 15961
    },
    {
      "epoch": 2.1859764448096413,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005829268654867243,
      "loss": 4.6447,
      "step": 15962
    },
    {
      "epoch": 2.186113393590797,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005828877275629217,
      "loss": 4.7488,
      "step": 15963
    },
    {
      "epoch": 2.186250342371953,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0005828485885378469,
      "loss": 4.7559,
      "step": 15964
    },
    {
      "epoch": 2.1863872911531086,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000582809448411824,
      "loss": 4.6243,
      "step": 15965
    },
    {
      "epoch": 2.186524239934265,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005827703071851777,
      "loss": 4.7201,
      "step": 15966
    },
    {
      "epoch": 2.1866611887154206,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005827311648582322,
      "loss": 4.765,
      "step": 15967
    },
    {
      "epoch": 2.1867981374965764,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005826920214313119,
      "loss": 4.7404,
      "step": 15968
    },
    {
      "epoch": 2.186935086277732,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005826528769047412,
      "loss": 4.7508,
      "step": 15969
    },
    {
      "epoch": 2.187072035058888,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005826137312788444,
      "loss": 4.7516,
      "step": 15970
    },
    {
      "epoch": 2.1872089838400437,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005825745845539462,
      "loss": 4.6762,
      "step": 15971
    },
    {
      "epoch": 2.1873459326211995,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005825354367303707,
      "loss": 4.7645,
      "step": 15972
    },
    {
      "epoch": 2.1874828814023557,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005824962878084425,
      "loss": 4.7317,
      "step": 15973
    },
    {
      "epoch": 2.1876198301835115,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005824571377884859,
      "loss": 4.7661,
      "step": 15974
    },
    {
      "epoch": 2.1877567789646672,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005824179866708256,
      "loss": 4.7203,
      "step": 15975
    },
    {
      "epoch": 2.187893727745823,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005823788344557858,
      "loss": 4.7818,
      "step": 15976
    },
    {
      "epoch": 2.188030676526979,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000582339681143691,
      "loss": 4.783,
      "step": 15977
    },
    {
      "epoch": 2.1881676253081346,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005823005267348658,
      "loss": 4.7362,
      "step": 15978
    },
    {
      "epoch": 2.188304574089291,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0005822613712296345,
      "loss": 4.7629,
      "step": 15979
    },
    {
      "epoch": 2.1884415228704466,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005822222146283219,
      "loss": 4.8209,
      "step": 15980
    },
    {
      "epoch": 2.1885784716516024,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0005821830569312521,
      "loss": 4.7291,
      "step": 15981
    },
    {
      "epoch": 2.188715420432758,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005821438981387499,
      "loss": 4.736,
      "step": 15982
    },
    {
      "epoch": 2.188852369213914,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005821047382511399,
      "loss": 4.751,
      "step": 15983
    },
    {
      "epoch": 2.1889893179950697,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005820655772687461,
      "loss": 4.8143,
      "step": 15984
    },
    {
      "epoch": 2.189126266776226,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005820264151918937,
      "loss": 4.7007,
      "step": 15985
    },
    {
      "epoch": 2.1892632155573817,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005819872520209068,
      "loss": 4.6668,
      "step": 15986
    },
    {
      "epoch": 2.1894001643385375,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005819480877561102,
      "loss": 4.7001,
      "step": 15987
    },
    {
      "epoch": 2.1895371131196932,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005819089223978282,
      "loss": 4.7908,
      "step": 15988
    },
    {
      "epoch": 2.189674061900849,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005818697559463856,
      "loss": 4.7555,
      "step": 15989
    },
    {
      "epoch": 2.189811010682005,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005818305884021069,
      "loss": 4.7187,
      "step": 15990
    },
    {
      "epoch": 2.1899479594631606,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005817914197653166,
      "loss": 4.7174,
      "step": 15991
    },
    {
      "epoch": 2.190084908244317,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005817522500363396,
      "loss": 4.7909,
      "step": 15992
    },
    {
      "epoch": 2.1902218570254726,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005817130792155001,
      "loss": 4.7437,
      "step": 15993
    },
    {
      "epoch": 2.1903588058066283,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005816739073031231,
      "loss": 4.7106,
      "step": 15994
    },
    {
      "epoch": 2.190495754587784,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000581634734299533,
      "loss": 4.7943,
      "step": 15995
    },
    {
      "epoch": 2.19063270336894,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005815955602050545,
      "loss": 4.7634,
      "step": 15996
    },
    {
      "epoch": 2.1907696521500957,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005815563850200122,
      "loss": 4.7429,
      "step": 15997
    },
    {
      "epoch": 2.190906600931252,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005815172087447308,
      "loss": 4.7101,
      "step": 15998
    },
    {
      "epoch": 2.1910435497124077,
      "grad_norm": 0.328125,
      "learning_rate": 0.000581478031379535,
      "loss": 4.8057,
      "step": 15999
    },
    {
      "epoch": 2.1911804984935634,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005814388529247493,
      "loss": 4.7872,
      "step": 16000
    },
    {
      "epoch": 2.191317447274719,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005813996733806987,
      "loss": 4.7538,
      "step": 16001
    },
    {
      "epoch": 2.191454396055875,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005813604927477075,
      "loss": 4.7818,
      "step": 16002
    },
    {
      "epoch": 2.1915913448370308,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005813213110261007,
      "loss": 4.765,
      "step": 16003
    },
    {
      "epoch": 2.191728293618187,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005812821282162029,
      "loss": 4.7377,
      "step": 16004
    },
    {
      "epoch": 2.1918652423993428,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005812429443183387,
      "loss": 4.7953,
      "step": 16005
    },
    {
      "epoch": 2.1920021911804985,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005812037593328331,
      "loss": 4.7183,
      "step": 16006
    },
    {
      "epoch": 2.1921391399616543,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005811645732600108,
      "loss": 4.7747,
      "step": 16007
    },
    {
      "epoch": 2.19227608874281,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005811253861001961,
      "loss": 4.7305,
      "step": 16008
    },
    {
      "epoch": 2.192413037523966,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005810861978537143,
      "loss": 4.6895,
      "step": 16009
    },
    {
      "epoch": 2.1925499863051217,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005810470085208898,
      "loss": 4.7761,
      "step": 16010
    },
    {
      "epoch": 2.192686935086278,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005810078181020476,
      "loss": 4.7455,
      "step": 16011
    },
    {
      "epoch": 2.1928238838674337,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005809686265975124,
      "loss": 4.7288,
      "step": 16012
    },
    {
      "epoch": 2.1929608326485894,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005809294340076089,
      "loss": 4.7213,
      "step": 16013
    },
    {
      "epoch": 2.193097781429745,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005808902403326621,
      "loss": 4.6599,
      "step": 16014
    },
    {
      "epoch": 2.193234730210901,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005808510455729967,
      "loss": 4.7868,
      "step": 16015
    },
    {
      "epoch": 2.193371678992057,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005808118497289375,
      "loss": 4.8306,
      "step": 16016
    },
    {
      "epoch": 2.193508627773213,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005807726528008092,
      "loss": 4.6896,
      "step": 16017
    },
    {
      "epoch": 2.1936455765543688,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005807334547889368,
      "loss": 4.7544,
      "step": 16018
    },
    {
      "epoch": 2.1937825253355245,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005806942556936453,
      "loss": 4.6926,
      "step": 16019
    },
    {
      "epoch": 2.1939194741166803,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005806550555152591,
      "loss": 4.6828,
      "step": 16020
    },
    {
      "epoch": 2.194056422897836,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005806158542541034,
      "loss": 4.7717,
      "step": 16021
    },
    {
      "epoch": 2.194193371678992,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005805766519105031,
      "loss": 4.6485,
      "step": 16022
    },
    {
      "epoch": 2.194330320460148,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000580537448484783,
      "loss": 4.7313,
      "step": 16023
    },
    {
      "epoch": 2.194467269241304,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005804982439772677,
      "loss": 4.7754,
      "step": 16024
    },
    {
      "epoch": 2.1946042180224596,
      "grad_norm": 0.375,
      "learning_rate": 0.0005804590383882825,
      "loss": 4.7775,
      "step": 16025
    },
    {
      "epoch": 2.1947411668036154,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005804198317181522,
      "loss": 4.8161,
      "step": 16026
    },
    {
      "epoch": 2.194878115584771,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005803806239672017,
      "loss": 4.7861,
      "step": 16027
    },
    {
      "epoch": 2.195015064365927,
      "grad_norm": 0.375,
      "learning_rate": 0.0005803414151357558,
      "loss": 4.7287,
      "step": 16028
    },
    {
      "epoch": 2.195152013147083,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005803022052241395,
      "loss": 4.6784,
      "step": 16029
    },
    {
      "epoch": 2.195288961928239,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000580262994232678,
      "loss": 4.7377,
      "step": 16030
    },
    {
      "epoch": 2.1954259107093947,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005802237821616958,
      "loss": 4.7723,
      "step": 16031
    },
    {
      "epoch": 2.1955628594905505,
      "grad_norm": 0.375,
      "learning_rate": 0.0005801845690115182,
      "loss": 4.7861,
      "step": 16032
    },
    {
      "epoch": 2.1956998082717063,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005801453547824701,
      "loss": 4.8273,
      "step": 16033
    },
    {
      "epoch": 2.195836757052862,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005801061394748764,
      "loss": 4.7263,
      "step": 16034
    },
    {
      "epoch": 2.1959737058340183,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000580066923089062,
      "loss": 4.7916,
      "step": 16035
    },
    {
      "epoch": 2.196110654615174,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005800277056253522,
      "loss": 4.7273,
      "step": 16036
    },
    {
      "epoch": 2.19624760339633,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005799884870840716,
      "loss": 4.7179,
      "step": 16037
    },
    {
      "epoch": 2.1963845521774856,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005799492674655455,
      "loss": 4.7503,
      "step": 16038
    },
    {
      "epoch": 2.1965215009586414,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000579910046770099,
      "loss": 4.7248,
      "step": 16039
    },
    {
      "epoch": 2.196658449739797,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005798708249980569,
      "loss": 4.7977,
      "step": 16040
    },
    {
      "epoch": 2.196795398520953,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005798316021497443,
      "loss": 4.7908,
      "step": 16041
    },
    {
      "epoch": 2.196932347302109,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005797923782254863,
      "loss": 4.7311,
      "step": 16042
    },
    {
      "epoch": 2.197069296083265,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005797531532256079,
      "loss": 4.7748,
      "step": 16043
    },
    {
      "epoch": 2.1972062448644207,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005797139271504342,
      "loss": 4.8589,
      "step": 16044
    },
    {
      "epoch": 2.1973431936455765,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005796747000002903,
      "loss": 4.7327,
      "step": 16045
    },
    {
      "epoch": 2.1974801424267323,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005796354717755013,
      "loss": 4.7388,
      "step": 16046
    },
    {
      "epoch": 2.197617091207888,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005795962424763921,
      "loss": 4.778,
      "step": 16047
    },
    {
      "epoch": 2.1977540399890443,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005795570121032882,
      "loss": 4.5864,
      "step": 16048
    },
    {
      "epoch": 2.1978909887702,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005795177806565142,
      "loss": 4.7793,
      "step": 16049
    },
    {
      "epoch": 2.198027937551356,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005794785481363958,
      "loss": 4.7335,
      "step": 16050
    },
    {
      "epoch": 2.1981648863325116,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005794393145432576,
      "loss": 4.7913,
      "step": 16051
    },
    {
      "epoch": 2.1983018351136674,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000579400079877425,
      "loss": 4.7266,
      "step": 16052
    },
    {
      "epoch": 2.198438783894823,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005793608441392232,
      "loss": 4.8114,
      "step": 16053
    },
    {
      "epoch": 2.1985757326759794,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005793216073289771,
      "loss": 4.7099,
      "step": 16054
    },
    {
      "epoch": 2.198712681457135,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005792823694470121,
      "loss": 4.7695,
      "step": 16055
    },
    {
      "epoch": 2.198849630238291,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005792431304936532,
      "loss": 4.817,
      "step": 16056
    },
    {
      "epoch": 2.1989865790194467,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005792038904692258,
      "loss": 4.677,
      "step": 16057
    },
    {
      "epoch": 2.1991235278006025,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005791646493740549,
      "loss": 4.7905,
      "step": 16058
    },
    {
      "epoch": 2.1992604765817583,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005791254072084659,
      "loss": 4.7198,
      "step": 16059
    },
    {
      "epoch": 2.199397425362914,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005790861639727838,
      "loss": 4.7054,
      "step": 16060
    },
    {
      "epoch": 2.1995343741440703,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000579046919667334,
      "loss": 4.6883,
      "step": 16061
    },
    {
      "epoch": 2.199671322925226,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005790076742924415,
      "loss": 4.7406,
      "step": 16062
    },
    {
      "epoch": 2.199808271706382,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005789684278484316,
      "loss": 4.6926,
      "step": 16063
    },
    {
      "epoch": 2.1999452204875376,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005789291803356298,
      "loss": 4.6898,
      "step": 16064
    },
    {
      "epoch": 2.2000821692686934,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000578889931754361,
      "loss": 4.7294,
      "step": 16065
    },
    {
      "epoch": 2.200219118049849,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005788506821049508,
      "loss": 4.6701,
      "step": 16066
    },
    {
      "epoch": 2.2003560668310054,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005788114313877241,
      "loss": 4.7135,
      "step": 16067
    },
    {
      "epoch": 2.200493015612161,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005787721796030065,
      "loss": 4.6879,
      "step": 16068
    },
    {
      "epoch": 2.200629964393317,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005787329267511231,
      "loss": 4.7823,
      "step": 16069
    },
    {
      "epoch": 2.2007669131744727,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005786936728323992,
      "loss": 4.7087,
      "step": 16070
    },
    {
      "epoch": 2.2009038619556285,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005786544178471603,
      "loss": 4.7449,
      "step": 16071
    },
    {
      "epoch": 2.2010408107367843,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005786151617957314,
      "loss": 4.7463,
      "step": 16072
    },
    {
      "epoch": 2.2011777595179405,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005785759046784382,
      "loss": 4.8034,
      "step": 16073
    },
    {
      "epoch": 2.2013147082990963,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005785366464956058,
      "loss": 4.701,
      "step": 16074
    },
    {
      "epoch": 2.201451657080252,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005784973872475595,
      "loss": 4.6483,
      "step": 16075
    },
    {
      "epoch": 2.201588605861408,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005784581269346248,
      "loss": 4.6513,
      "step": 16076
    },
    {
      "epoch": 2.2017255546425636,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005784188655571269,
      "loss": 4.7823,
      "step": 16077
    },
    {
      "epoch": 2.2018625034237194,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005783796031153916,
      "loss": 4.7291,
      "step": 16078
    },
    {
      "epoch": 2.2019994522048756,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005783403396097435,
      "loss": 4.755,
      "step": 16079
    },
    {
      "epoch": 2.2021364009860314,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005783010750405086,
      "loss": 4.7359,
      "step": 16080
    },
    {
      "epoch": 2.202273349767187,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005782618094080121,
      "loss": 4.781,
      "step": 16081
    },
    {
      "epoch": 2.202410298548343,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005782225427125794,
      "loss": 4.7763,
      "step": 16082
    },
    {
      "epoch": 2.2025472473294987,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000578183274954536,
      "loss": 4.6893,
      "step": 16083
    },
    {
      "epoch": 2.2026841961106545,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005781440061342073,
      "loss": 4.6841,
      "step": 16084
    },
    {
      "epoch": 2.2028211448918107,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005781047362519187,
      "loss": 4.7834,
      "step": 16085
    },
    {
      "epoch": 2.2029580936729665,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005780654653079954,
      "loss": 4.6939,
      "step": 16086
    },
    {
      "epoch": 2.2030950424541222,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005780261933027632,
      "loss": 4.7896,
      "step": 16087
    },
    {
      "epoch": 2.203231991235278,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005779869202365475,
      "loss": 4.7287,
      "step": 16088
    },
    {
      "epoch": 2.203368940016434,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005779476461096737,
      "loss": 4.6915,
      "step": 16089
    },
    {
      "epoch": 2.2035058887975896,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005779083709224671,
      "loss": 4.8049,
      "step": 16090
    },
    {
      "epoch": 2.2036428375787454,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005778690946752535,
      "loss": 4.777,
      "step": 16091
    },
    {
      "epoch": 2.2037797863599016,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005778298173683582,
      "loss": 4.8051,
      "step": 16092
    },
    {
      "epoch": 2.2039167351410573,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005777905390021067,
      "loss": 4.8034,
      "step": 16093
    },
    {
      "epoch": 2.204053683922213,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005777512595768246,
      "loss": 4.6866,
      "step": 16094
    },
    {
      "epoch": 2.204190632703369,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005777119790928375,
      "loss": 4.7249,
      "step": 16095
    },
    {
      "epoch": 2.2043275814845247,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005776726975504705,
      "loss": 4.7227,
      "step": 16096
    },
    {
      "epoch": 2.2044645302656805,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005776334149500497,
      "loss": 4.7524,
      "step": 16097
    },
    {
      "epoch": 2.2046014790468367,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005775941312919002,
      "loss": 4.7862,
      "step": 16098
    },
    {
      "epoch": 2.2047384278279925,
      "grad_norm": 0.326171875,
      "learning_rate": 0.000577554846576348,
      "loss": 4.7562,
      "step": 16099
    },
    {
      "epoch": 2.2048753766091482,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000577515560803718,
      "loss": 4.7759,
      "step": 16100
    },
    {
      "epoch": 2.205012325390304,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005774762739743364,
      "loss": 4.7609,
      "step": 16101
    },
    {
      "epoch": 2.20514927417146,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005774369860885285,
      "loss": 4.6118,
      "step": 16102
    },
    {
      "epoch": 2.2052862229526156,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005773976971466198,
      "loss": 4.7485,
      "step": 16103
    },
    {
      "epoch": 2.205423171733772,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005773584071489362,
      "loss": 4.7603,
      "step": 16104
    },
    {
      "epoch": 2.2055601205149276,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005773191160958031,
      "loss": 4.6802,
      "step": 16105
    },
    {
      "epoch": 2.2056970692960833,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005772798239875461,
      "loss": 4.7832,
      "step": 16106
    },
    {
      "epoch": 2.205834018077239,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005772405308244909,
      "loss": 4.7845,
      "step": 16107
    },
    {
      "epoch": 2.205970966858395,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005772012366069631,
      "loss": 4.7444,
      "step": 16108
    },
    {
      "epoch": 2.2061079156395507,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005771619413352884,
      "loss": 4.685,
      "step": 16109
    },
    {
      "epoch": 2.2062448644207064,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005771226450097924,
      "loss": 4.6406,
      "step": 16110
    },
    {
      "epoch": 2.2063818132018627,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005770833476308006,
      "loss": 4.7351,
      "step": 16111
    },
    {
      "epoch": 2.2065187619830184,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000577044049198639,
      "loss": 4.7607,
      "step": 16112
    },
    {
      "epoch": 2.206655710764174,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000577004749713633,
      "loss": 4.7448,
      "step": 16113
    },
    {
      "epoch": 2.20679265954533,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005769654491761083,
      "loss": 4.6986,
      "step": 16114
    },
    {
      "epoch": 2.2069296083264858,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005769261475863908,
      "loss": 4.6625,
      "step": 16115
    },
    {
      "epoch": 2.2070665571076415,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000576886844944806,
      "loss": 4.7127,
      "step": 16116
    },
    {
      "epoch": 2.2072035058887978,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005768475412516798,
      "loss": 4.7355,
      "step": 16117
    },
    {
      "epoch": 2.2073404546699535,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005768082365073375,
      "loss": 4.7902,
      "step": 16118
    },
    {
      "epoch": 2.2074774034511093,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005767689307121054,
      "loss": 4.8044,
      "step": 16119
    },
    {
      "epoch": 2.207614352232265,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005767296238663089,
      "loss": 4.7405,
      "step": 16120
    },
    {
      "epoch": 2.207751301013421,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005766903159702737,
      "loss": 4.7333,
      "step": 16121
    },
    {
      "epoch": 2.2078882497945767,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005766510070243257,
      "loss": 4.7234,
      "step": 16122
    },
    {
      "epoch": 2.208025198575733,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005766116970287908,
      "loss": 4.7071,
      "step": 16123
    },
    {
      "epoch": 2.2081621473568886,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005765723859839945,
      "loss": 4.7019,
      "step": 16124
    },
    {
      "epoch": 2.2082990961380444,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005765330738902626,
      "loss": 4.7096,
      "step": 16125
    },
    {
      "epoch": 2.2084360449192,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005764937607479211,
      "loss": 4.7303,
      "step": 16126
    },
    {
      "epoch": 2.208572993700356,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005764544465572955,
      "loss": 4.7601,
      "step": 16127
    },
    {
      "epoch": 2.2087099424815118,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005764151313187119,
      "loss": 4.6565,
      "step": 16128
    },
    {
      "epoch": 2.2088468912626675,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005763758150324959,
      "loss": 4.7439,
      "step": 16129
    },
    {
      "epoch": 2.2089838400438238,
      "grad_norm": 0.375,
      "learning_rate": 0.0005763364976989733,
      "loss": 4.7859,
      "step": 16130
    },
    {
      "epoch": 2.2091207888249795,
      "grad_norm": 0.375,
      "learning_rate": 0.0005762971793184704,
      "loss": 4.6721,
      "step": 16131
    },
    {
      "epoch": 2.2092577376061353,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005762578598913123,
      "loss": 4.7724,
      "step": 16132
    },
    {
      "epoch": 2.209394686387291,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005762185394178255,
      "loss": 4.8573,
      "step": 16133
    },
    {
      "epoch": 2.209531635168447,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005761792178983355,
      "loss": 4.7434,
      "step": 16134
    },
    {
      "epoch": 2.209668583949603,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005761398953331682,
      "loss": 4.8099,
      "step": 16135
    },
    {
      "epoch": 2.209805532730759,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005761005717226494,
      "loss": 4.7611,
      "step": 16136
    },
    {
      "epoch": 2.2099424815119146,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005760612470671055,
      "loss": 4.735,
      "step": 16137
    },
    {
      "epoch": 2.2100794302930704,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005760219213668616,
      "loss": 4.7099,
      "step": 16138
    },
    {
      "epoch": 2.210216379074226,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005759825946222443,
      "loss": 4.6374,
      "step": 16139
    },
    {
      "epoch": 2.210353327855382,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005759432668335791,
      "loss": 4.7281,
      "step": 16140
    },
    {
      "epoch": 2.2104902766365377,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000575903938001192,
      "loss": 4.7055,
      "step": 16141
    },
    {
      "epoch": 2.210627225417694,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000575864608125409,
      "loss": 4.7707,
      "step": 16142
    },
    {
      "epoch": 2.2107641741988497,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005758252772065561,
      "loss": 4.6732,
      "step": 16143
    },
    {
      "epoch": 2.2109011229800055,
      "grad_norm": 0.40625,
      "learning_rate": 0.000575785945244959,
      "loss": 4.755,
      "step": 16144
    },
    {
      "epoch": 2.2110380717611613,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005757466122409437,
      "loss": 4.7018,
      "step": 16145
    },
    {
      "epoch": 2.211175020542317,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005757072781948366,
      "loss": 4.749,
      "step": 16146
    },
    {
      "epoch": 2.211311969323473,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000575667943106963,
      "loss": 4.7322,
      "step": 16147
    },
    {
      "epoch": 2.211448918104629,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005756286069776495,
      "loss": 4.7104,
      "step": 16148
    },
    {
      "epoch": 2.211585866885785,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005755892698072215,
      "loss": 4.7489,
      "step": 16149
    },
    {
      "epoch": 2.2117228156669406,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005755499315960055,
      "loss": 4.8415,
      "step": 16150
    },
    {
      "epoch": 2.2118597644480964,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005755105923443272,
      "loss": 4.7295,
      "step": 16151
    },
    {
      "epoch": 2.211996713229252,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005754712520525129,
      "loss": 4.7377,
      "step": 16152
    },
    {
      "epoch": 2.212133662010408,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005754319107208883,
      "loss": 4.8169,
      "step": 16153
    },
    {
      "epoch": 2.212270610791564,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005753925683497795,
      "loss": 4.6764,
      "step": 16154
    },
    {
      "epoch": 2.21240755957272,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005753532249395128,
      "loss": 4.829,
      "step": 16155
    },
    {
      "epoch": 2.2125445083538757,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005753138804904138,
      "loss": 4.7458,
      "step": 16156
    },
    {
      "epoch": 2.2126814571350315,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000575274535002809,
      "loss": 4.7416,
      "step": 16157
    },
    {
      "epoch": 2.2128184059161873,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005752351884770241,
      "loss": 4.7423,
      "step": 16158
    },
    {
      "epoch": 2.212955354697343,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005751958409133855,
      "loss": 4.6809,
      "step": 16159
    },
    {
      "epoch": 2.213092303478499,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005751564923122192,
      "loss": 4.657,
      "step": 16160
    },
    {
      "epoch": 2.213229252259655,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000575117142673851,
      "loss": 4.8055,
      "step": 16161
    },
    {
      "epoch": 2.213366201040811,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005750777919986074,
      "loss": 4.6985,
      "step": 16162
    },
    {
      "epoch": 2.2135031498219666,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005750384402868143,
      "loss": 4.7434,
      "step": 16163
    },
    {
      "epoch": 2.2136400986031224,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005749990875387979,
      "loss": 4.7421,
      "step": 16164
    },
    {
      "epoch": 2.213777047384278,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005749597337548841,
      "loss": 4.8572,
      "step": 16165
    },
    {
      "epoch": 2.213913996165434,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005749203789353995,
      "loss": 4.762,
      "step": 16166
    },
    {
      "epoch": 2.21405094494659,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005748810230806696,
      "loss": 4.7316,
      "step": 16167
    },
    {
      "epoch": 2.214187893727746,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005748416661910211,
      "loss": 4.7791,
      "step": 16168
    },
    {
      "epoch": 2.2143248425089017,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00057480230826678,
      "loss": 4.7546,
      "step": 16169
    },
    {
      "epoch": 2.2144617912900575,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005747629493082723,
      "loss": 4.8225,
      "step": 16170
    },
    {
      "epoch": 2.2145987400712133,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005747235893158245,
      "loss": 4.8243,
      "step": 16171
    },
    {
      "epoch": 2.214735688852369,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005746842282897623,
      "loss": 4.7239,
      "step": 16172
    },
    {
      "epoch": 2.2148726376335253,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005746448662304125,
      "loss": 4.7547,
      "step": 16173
    },
    {
      "epoch": 2.215009586414681,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0005746055031381008,
      "loss": 4.7878,
      "step": 16174
    },
    {
      "epoch": 2.215146535195837,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005745661390131537,
      "loss": 4.713,
      "step": 16175
    },
    {
      "epoch": 2.2152834839769926,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005745267738558973,
      "loss": 4.7664,
      "step": 16176
    },
    {
      "epoch": 2.2154204327581484,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005744874076666579,
      "loss": 4.55,
      "step": 16177
    },
    {
      "epoch": 2.215557381539304,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005744480404457615,
      "loss": 4.7576,
      "step": 16178
    },
    {
      "epoch": 2.21569433032046,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005744086721935347,
      "loss": 4.7553,
      "step": 16179
    },
    {
      "epoch": 2.215831279101616,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005743693029103036,
      "loss": 4.7604,
      "step": 16180
    },
    {
      "epoch": 2.215968227882772,
      "grad_norm": 0.375,
      "learning_rate": 0.0005743299325963944,
      "loss": 4.7225,
      "step": 16181
    },
    {
      "epoch": 2.2161051766639277,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005742905612521335,
      "loss": 4.7018,
      "step": 16182
    },
    {
      "epoch": 2.2162421254450835,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005742511888778469,
      "loss": 4.7095,
      "step": 16183
    },
    {
      "epoch": 2.2163790742262393,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005742118154738613,
      "loss": 4.7989,
      "step": 16184
    },
    {
      "epoch": 2.216516023007395,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005741724410405026,
      "loss": 4.7561,
      "step": 16185
    },
    {
      "epoch": 2.2166529717885513,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005741330655780974,
      "loss": 4.7439,
      "step": 16186
    },
    {
      "epoch": 2.216789920569707,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000574093689086972,
      "loss": 4.7546,
      "step": 16187
    },
    {
      "epoch": 2.216926869350863,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005740543115674525,
      "loss": 4.7448,
      "step": 16188
    },
    {
      "epoch": 2.2170638181320186,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005740149330198654,
      "loss": 4.7429,
      "step": 16189
    },
    {
      "epoch": 2.2172007669131744,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000573975553444537,
      "loss": 4.6084,
      "step": 16190
    },
    {
      "epoch": 2.21733771569433,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005739361728417937,
      "loss": 4.7371,
      "step": 16191
    },
    {
      "epoch": 2.2174746644754864,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005738967912119617,
      "loss": 4.7006,
      "step": 16192
    },
    {
      "epoch": 2.217611613256642,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005738574085553675,
      "loss": 4.7592,
      "step": 16193
    },
    {
      "epoch": 2.217748562037798,
      "grad_norm": 0.375,
      "learning_rate": 0.0005738180248723374,
      "loss": 4.8313,
      "step": 16194
    },
    {
      "epoch": 2.2178855108189537,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000573778640163198,
      "loss": 4.6661,
      "step": 16195
    },
    {
      "epoch": 2.2180224596001095,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005737392544282753,
      "loss": 4.7104,
      "step": 16196
    },
    {
      "epoch": 2.2181594083812652,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000573699867667896,
      "loss": 4.7297,
      "step": 16197
    },
    {
      "epoch": 2.2182963571624215,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005736604798823864,
      "loss": 4.6797,
      "step": 16198
    },
    {
      "epoch": 2.2184333059435772,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000573621091072073,
      "loss": 4.7488,
      "step": 16199
    },
    {
      "epoch": 2.218570254724733,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005735817012372821,
      "loss": 4.6923,
      "step": 16200
    },
    {
      "epoch": 2.218707203505889,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005735423103783402,
      "loss": 4.6885,
      "step": 16201
    },
    {
      "epoch": 2.2188441522870446,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005735029184955737,
      "loss": 4.7794,
      "step": 16202
    },
    {
      "epoch": 2.2189811010682003,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005734635255893091,
      "loss": 4.803,
      "step": 16203
    },
    {
      "epoch": 2.2191180498493566,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000573424131659873,
      "loss": 4.6906,
      "step": 16204
    },
    {
      "epoch": 2.2192549986305123,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005733847367075915,
      "loss": 4.7221,
      "step": 16205
    },
    {
      "epoch": 2.219391947411668,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005733453407327913,
      "loss": 4.7538,
      "step": 16206
    },
    {
      "epoch": 2.219528896192824,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000573305943735799,
      "loss": 4.6313,
      "step": 16207
    },
    {
      "epoch": 2.2196658449739797,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005732665457169409,
      "loss": 4.7739,
      "step": 16208
    },
    {
      "epoch": 2.2198027937551355,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005732271466765437,
      "loss": 4.7753,
      "step": 16209
    },
    {
      "epoch": 2.2199397425362912,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005731877466149336,
      "loss": 4.7667,
      "step": 16210
    },
    {
      "epoch": 2.2200766913174474,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005731483455324374,
      "loss": 4.6491,
      "step": 16211
    },
    {
      "epoch": 2.2202136400986032,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005731089434293814,
      "loss": 4.7135,
      "step": 16212
    },
    {
      "epoch": 2.220350588879759,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005730695403060923,
      "loss": 4.7839,
      "step": 16213
    },
    {
      "epoch": 2.220487537660915,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005730301361628966,
      "loss": 4.7919,
      "step": 16214
    },
    {
      "epoch": 2.2206244864420706,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005729907310001208,
      "loss": 4.6972,
      "step": 16215
    },
    {
      "epoch": 2.2207614352232263,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005729513248180915,
      "loss": 4.7151,
      "step": 16216
    },
    {
      "epoch": 2.2208983840043826,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005729119176171353,
      "loss": 4.744,
      "step": 16217
    },
    {
      "epoch": 2.2210353327855383,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005728725093975788,
      "loss": 4.7322,
      "step": 16218
    },
    {
      "epoch": 2.221172281566694,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005728331001597485,
      "loss": 4.7044,
      "step": 16219
    },
    {
      "epoch": 2.22130923034785,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000572793689903971,
      "loss": 4.7374,
      "step": 16220
    },
    {
      "epoch": 2.2214461791290057,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000572754278630573,
      "loss": 4.8133,
      "step": 16221
    },
    {
      "epoch": 2.2215831279101614,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000572714866339881,
      "loss": 4.7154,
      "step": 16222
    },
    {
      "epoch": 2.2217200766913177,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005726754530322217,
      "loss": 4.6834,
      "step": 16223
    },
    {
      "epoch": 2.2218570254724734,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005726360387079216,
      "loss": 4.7677,
      "step": 16224
    },
    {
      "epoch": 2.221993974253629,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005725966233673075,
      "loss": 4.7296,
      "step": 16225
    },
    {
      "epoch": 2.222130923034785,
      "grad_norm": 0.375,
      "learning_rate": 0.0005725572070107059,
      "loss": 4.7664,
      "step": 16226
    },
    {
      "epoch": 2.2222678718159408,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005725177896384436,
      "loss": 4.7837,
      "step": 16227
    },
    {
      "epoch": 2.2224048205970965,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005724783712508472,
      "loss": 4.7184,
      "step": 16228
    },
    {
      "epoch": 2.2225417693782523,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005724389518482432,
      "loss": 4.7782,
      "step": 16229
    },
    {
      "epoch": 2.2226787181594085,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005723995314309586,
      "loss": 4.7892,
      "step": 16230
    },
    {
      "epoch": 2.2228156669405643,
      "grad_norm": 0.421875,
      "learning_rate": 0.0005723601099993199,
      "loss": 4.7202,
      "step": 16231
    },
    {
      "epoch": 2.22295261572172,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005723206875536537,
      "loss": 4.7152,
      "step": 16232
    },
    {
      "epoch": 2.223089564502876,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005722812640942868,
      "loss": 4.714,
      "step": 16233
    },
    {
      "epoch": 2.2232265132840316,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005722418396215461,
      "loss": 4.6752,
      "step": 16234
    },
    {
      "epoch": 2.2233634620651874,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005722024141357581,
      "loss": 4.7533,
      "step": 16235
    },
    {
      "epoch": 2.2235004108463436,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005721629876372495,
      "loss": 4.7968,
      "step": 16236
    },
    {
      "epoch": 2.2236373596274994,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005721235601263471,
      "loss": 4.7884,
      "step": 16237
    },
    {
      "epoch": 2.223774308408655,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005720841316033778,
      "loss": 4.7825,
      "step": 16238
    },
    {
      "epoch": 2.223911257189811,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005720447020686679,
      "loss": 4.7477,
      "step": 16239
    },
    {
      "epoch": 2.2240482059709668,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005720052715225449,
      "loss": 4.7492,
      "step": 16240
    },
    {
      "epoch": 2.2241851547521225,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005719658399653348,
      "loss": 4.7343,
      "step": 16241
    },
    {
      "epoch": 2.2243221035332787,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005719264073973648,
      "loss": 4.6977,
      "step": 16242
    },
    {
      "epoch": 2.2244590523144345,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005718869738189617,
      "loss": 4.6609,
      "step": 16243
    },
    {
      "epoch": 2.2245960010955903,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005718475392304522,
      "loss": 4.7522,
      "step": 16244
    },
    {
      "epoch": 2.224732949876746,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005718081036321631,
      "loss": 4.7016,
      "step": 16245
    },
    {
      "epoch": 2.224869898657902,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000571768667024421,
      "loss": 4.7144,
      "step": 16246
    },
    {
      "epoch": 2.2250068474390576,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005717292294075533,
      "loss": 4.6897,
      "step": 16247
    },
    {
      "epoch": 2.2251437962202134,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005716897907818862,
      "loss": 4.6921,
      "step": 16248
    },
    {
      "epoch": 2.2252807450013696,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000571650351147747,
      "loss": 4.7538,
      "step": 16249
    },
    {
      "epoch": 2.2254176937825254,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005716109105054621,
      "loss": 4.672,
      "step": 16250
    },
    {
      "epoch": 2.225554642563681,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005715714688553589,
      "loss": 4.7787,
      "step": 16251
    },
    {
      "epoch": 2.225691591344837,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005715320261977637,
      "loss": 4.7118,
      "step": 16252
    },
    {
      "epoch": 2.2258285401259927,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005714925825330038,
      "loss": 4.706,
      "step": 16253
    },
    {
      "epoch": 2.225965488907149,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005714531378614059,
      "loss": 4.7308,
      "step": 16254
    },
    {
      "epoch": 2.2261024376883047,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005714136921832969,
      "loss": 4.6811,
      "step": 16255
    },
    {
      "epoch": 2.2262393864694605,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005713742454990036,
      "loss": 4.687,
      "step": 16256
    },
    {
      "epoch": 2.2263763352506163,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005713347978088531,
      "loss": 4.7676,
      "step": 16257
    },
    {
      "epoch": 2.226513284031772,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005712953491131723,
      "loss": 4.7767,
      "step": 16258
    },
    {
      "epoch": 2.226650232812928,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005712558994122879,
      "loss": 4.7478,
      "step": 16259
    },
    {
      "epoch": 2.2267871815940836,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000571216448706527,
      "loss": 4.7279,
      "step": 16260
    },
    {
      "epoch": 2.22692413037524,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005711769969962165,
      "loss": 4.7579,
      "step": 16261
    },
    {
      "epoch": 2.2270610791563956,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005711375442816834,
      "loss": 4.7405,
      "step": 16262
    },
    {
      "epoch": 2.2271980279375514,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005710980905632546,
      "loss": 4.7379,
      "step": 16263
    },
    {
      "epoch": 2.227334976718707,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005710586358412571,
      "loss": 4.7144,
      "step": 16264
    },
    {
      "epoch": 2.227471925499863,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005710191801160178,
      "loss": 4.6949,
      "step": 16265
    },
    {
      "epoch": 2.2276088742810187,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005709797233878637,
      "loss": 4.7736,
      "step": 16266
    },
    {
      "epoch": 2.227745823062175,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005709402656571218,
      "loss": 4.7936,
      "step": 16267
    },
    {
      "epoch": 2.2278827718433307,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000570900806924119,
      "loss": 4.7365,
      "step": 16268
    },
    {
      "epoch": 2.2280197206244865,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005708613471891827,
      "loss": 4.8541,
      "step": 16269
    },
    {
      "epoch": 2.2281566694056423,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005708218864526394,
      "loss": 4.7867,
      "step": 16270
    },
    {
      "epoch": 2.228293618186798,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005707824247148162,
      "loss": 4.7596,
      "step": 16271
    },
    {
      "epoch": 2.228430566967954,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005707429619760406,
      "loss": 4.72,
      "step": 16272
    },
    {
      "epoch": 2.22856751574911,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000570703498236639,
      "loss": 4.6942,
      "step": 16273
    },
    {
      "epoch": 2.228704464530266,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000570664033496939,
      "loss": 4.8545,
      "step": 16274
    },
    {
      "epoch": 2.2288414133114216,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005706245677572672,
      "loss": 4.7219,
      "step": 16275
    },
    {
      "epoch": 2.2289783620925774,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000570585101017951,
      "loss": 4.7886,
      "step": 16276
    },
    {
      "epoch": 2.229115310873733,
      "grad_norm": 0.375,
      "learning_rate": 0.0005705456332793172,
      "loss": 4.7418,
      "step": 16277
    },
    {
      "epoch": 2.229252259654889,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005705061645416931,
      "loss": 4.7091,
      "step": 16278
    },
    {
      "epoch": 2.2293892084360447,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005704666948054058,
      "loss": 4.7788,
      "step": 16279
    },
    {
      "epoch": 2.229526157217201,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005704272240707821,
      "loss": 4.8363,
      "step": 16280
    },
    {
      "epoch": 2.2296631059983567,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005703877523381495,
      "loss": 4.7227,
      "step": 16281
    },
    {
      "epoch": 2.2298000547795125,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005703482796078348,
      "loss": 4.75,
      "step": 16282
    },
    {
      "epoch": 2.2299370035606683,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005703088058801652,
      "loss": 4.776,
      "step": 16283
    },
    {
      "epoch": 2.230073952341824,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0005702693311554677,
      "loss": 4.7598,
      "step": 16284
    },
    {
      "epoch": 2.23021090112298,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005702298554340699,
      "loss": 4.7923,
      "step": 16285
    },
    {
      "epoch": 2.230347849904136,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005701903787162985,
      "loss": 4.7738,
      "step": 16286
    },
    {
      "epoch": 2.230484798685292,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005701509010024808,
      "loss": 4.7095,
      "step": 16287
    },
    {
      "epoch": 2.2306217474664476,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000570111422292944,
      "loss": 4.7879,
      "step": 16288
    },
    {
      "epoch": 2.2307586962476034,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000570071942588015,
      "loss": 4.7507,
      "step": 16289
    },
    {
      "epoch": 2.230895645028759,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005700324618880216,
      "loss": 4.7903,
      "step": 16290
    },
    {
      "epoch": 2.231032593809915,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005699929801932902,
      "loss": 4.7572,
      "step": 16291
    },
    {
      "epoch": 2.231169542591071,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005699534975041486,
      "loss": 4.7324,
      "step": 16292
    },
    {
      "epoch": 2.231306491372227,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005699140138209238,
      "loss": 4.6514,
      "step": 16293
    },
    {
      "epoch": 2.2314434401533827,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005698745291439428,
      "loss": 4.7029,
      "step": 16294
    },
    {
      "epoch": 2.2315803889345385,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005698350434735332,
      "loss": 4.727,
      "step": 16295
    },
    {
      "epoch": 2.2317173377156942,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000569795556810022,
      "loss": 4.7553,
      "step": 16296
    },
    {
      "epoch": 2.23185428649685,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005697560691537365,
      "loss": 4.7154,
      "step": 16297
    },
    {
      "epoch": 2.231991235278006,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005697165805050037,
      "loss": 4.7004,
      "step": 16298
    },
    {
      "epoch": 2.232128184059162,
      "grad_norm": 0.375,
      "learning_rate": 0.0005696770908641513,
      "loss": 4.778,
      "step": 16299
    },
    {
      "epoch": 2.232265132840318,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005696376002315063,
      "loss": 4.6383,
      "step": 16300
    },
    {
      "epoch": 2.2324020816214736,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000569598108607396,
      "loss": 4.7668,
      "step": 16301
    },
    {
      "epoch": 2.2325390304026294,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005695586159921477,
      "loss": 4.732,
      "step": 16302
    },
    {
      "epoch": 2.232675979183785,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005695191223860885,
      "loss": 4.8446,
      "step": 16303
    },
    {
      "epoch": 2.232812927964941,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005694796277895461,
      "loss": 4.7508,
      "step": 16304
    },
    {
      "epoch": 2.232949876746097,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005694401322028474,
      "loss": 4.6826,
      "step": 16305
    },
    {
      "epoch": 2.233086825527253,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005694006356263197,
      "loss": 4.7366,
      "step": 16306
    },
    {
      "epoch": 2.2332237743084087,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005693611380602907,
      "loss": 4.7334,
      "step": 16307
    },
    {
      "epoch": 2.2333607230895645,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005693216395050875,
      "loss": 4.6462,
      "step": 16308
    },
    {
      "epoch": 2.2334976718707202,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005692821399610374,
      "loss": 4.7505,
      "step": 16309
    },
    {
      "epoch": 2.233634620651876,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005692426394284677,
      "loss": 4.7722,
      "step": 16310
    },
    {
      "epoch": 2.2337715694330322,
      "grad_norm": 0.359375,
      "learning_rate": 0.000569203137907706,
      "loss": 4.7276,
      "step": 16311
    },
    {
      "epoch": 2.233908518214188,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005691636353990794,
      "loss": 4.7354,
      "step": 16312
    },
    {
      "epoch": 2.234045466995344,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005691241319029153,
      "loss": 4.6843,
      "step": 16313
    },
    {
      "epoch": 2.2341824157764996,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005690846274195413,
      "loss": 4.8316,
      "step": 16314
    },
    {
      "epoch": 2.2343193645576553,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005690451219492844,
      "loss": 4.7263,
      "step": 16315
    },
    {
      "epoch": 2.234456313338811,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005690056154924725,
      "loss": 4.7419,
      "step": 16316
    },
    {
      "epoch": 2.2345932621199673,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005689661080494324,
      "loss": 4.7391,
      "step": 16317
    },
    {
      "epoch": 2.234730210901123,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005689265996204921,
      "loss": 4.7116,
      "step": 16318
    },
    {
      "epoch": 2.234867159682279,
      "grad_norm": 0.375,
      "learning_rate": 0.0005688870902059784,
      "loss": 4.7536,
      "step": 16319
    },
    {
      "epoch": 2.2350041084634347,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005688475798062192,
      "loss": 4.7616,
      "step": 16320
    },
    {
      "epoch": 2.2351410572445904,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005688080684215419,
      "loss": 4.7857,
      "step": 16321
    },
    {
      "epoch": 2.2352780060257462,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005687685560522737,
      "loss": 4.7642,
      "step": 16322
    },
    {
      "epoch": 2.2354149548069024,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005687290426987424,
      "loss": 4.7567,
      "step": 16323
    },
    {
      "epoch": 2.235551903588058,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005686895283612749,
      "loss": 4.7279,
      "step": 16324
    },
    {
      "epoch": 2.235688852369214,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005686500130401992,
      "loss": 4.689,
      "step": 16325
    },
    {
      "epoch": 2.2358258011503698,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005686104967358426,
      "loss": 4.7623,
      "step": 16326
    },
    {
      "epoch": 2.2359627499315256,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005685709794485324,
      "loss": 4.731,
      "step": 16327
    },
    {
      "epoch": 2.2360996987126813,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005685314611785962,
      "loss": 4.8186,
      "step": 16328
    },
    {
      "epoch": 2.236236647493837,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005684919419263616,
      "loss": 4.8441,
      "step": 16329
    },
    {
      "epoch": 2.2363735962749933,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005684524216921562,
      "loss": 4.7865,
      "step": 16330
    },
    {
      "epoch": 2.236510545056149,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005684129004763073,
      "loss": 4.7334,
      "step": 16331
    },
    {
      "epoch": 2.236647493837305,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005683733782791424,
      "loss": 4.7724,
      "step": 16332
    },
    {
      "epoch": 2.2367844426184607,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000568333855100989,
      "loss": 4.7806,
      "step": 16333
    },
    {
      "epoch": 2.2369213913996164,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000568294330942175,
      "loss": 4.7818,
      "step": 16334
    },
    {
      "epoch": 2.237058340180772,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005682548058030275,
      "loss": 4.7887,
      "step": 16335
    },
    {
      "epoch": 2.2371952889619284,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005682152796838743,
      "loss": 4.6879,
      "step": 16336
    },
    {
      "epoch": 2.237332237743084,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000568175752585043,
      "loss": 4.7187,
      "step": 16337
    },
    {
      "epoch": 2.23746918652424,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005681362245068609,
      "loss": 4.7427,
      "step": 16338
    },
    {
      "epoch": 2.2376061353053958,
      "grad_norm": 0.375,
      "learning_rate": 0.0005680966954496559,
      "loss": 4.7764,
      "step": 16339
    },
    {
      "epoch": 2.2377430840865515,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005680571654137555,
      "loss": 4.7375,
      "step": 16340
    },
    {
      "epoch": 2.2378800328677073,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005680176343994871,
      "loss": 4.6708,
      "step": 16341
    },
    {
      "epoch": 2.2380169816488635,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005679781024071785,
      "loss": 4.7343,
      "step": 16342
    },
    {
      "epoch": 2.2381539304300193,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005679385694371572,
      "loss": 4.6897,
      "step": 16343
    },
    {
      "epoch": 2.238290879211175,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005678990354897509,
      "loss": 4.7686,
      "step": 16344
    },
    {
      "epoch": 2.238427827992331,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005678595005652871,
      "loss": 4.7145,
      "step": 16345
    },
    {
      "epoch": 2.2385647767734866,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005678199646640936,
      "loss": 4.729,
      "step": 16346
    },
    {
      "epoch": 2.2387017255546424,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000567780427786498,
      "loss": 4.6843,
      "step": 16347
    },
    {
      "epoch": 2.238838674335798,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005677408899328281,
      "loss": 4.744,
      "step": 16348
    },
    {
      "epoch": 2.2389756231169544,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005677013511034111,
      "loss": 4.7096,
      "step": 16349
    },
    {
      "epoch": 2.23911257189811,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005676618112985751,
      "loss": 4.7173,
      "step": 16350
    },
    {
      "epoch": 2.239249520679266,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005676222705186477,
      "loss": 4.7727,
      "step": 16351
    },
    {
      "epoch": 2.2393864694604217,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005675827287639565,
      "loss": 4.7779,
      "step": 16352
    },
    {
      "epoch": 2.2395234182415775,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005675431860348291,
      "loss": 4.7225,
      "step": 16353
    },
    {
      "epoch": 2.2396603670227333,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005675036423315933,
      "loss": 4.7007,
      "step": 16354
    },
    {
      "epoch": 2.2397973158038895,
      "grad_norm": 0.375,
      "learning_rate": 0.000567464097654577,
      "loss": 4.7551,
      "step": 16355
    },
    {
      "epoch": 2.2399342645850453,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005674245520041075,
      "loss": 4.7898,
      "step": 16356
    },
    {
      "epoch": 2.240071213366201,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005673850053805129,
      "loss": 4.6861,
      "step": 16357
    },
    {
      "epoch": 2.240208162147357,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005673454577841206,
      "loss": 4.7425,
      "step": 16358
    },
    {
      "epoch": 2.2403451109285126,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005673059092152588,
      "loss": 4.7339,
      "step": 16359
    },
    {
      "epoch": 2.2404820597096684,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005672663596742548,
      "loss": 4.7977,
      "step": 16360
    },
    {
      "epoch": 2.2406190084908246,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005672268091614366,
      "loss": 4.6552,
      "step": 16361
    },
    {
      "epoch": 2.2407559572719804,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005671872576771319,
      "loss": 4.7855,
      "step": 16362
    },
    {
      "epoch": 2.240892906053136,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005671477052216685,
      "loss": 4.799,
      "step": 16363
    },
    {
      "epoch": 2.241029854834292,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005671081517953741,
      "loss": 4.7501,
      "step": 16364
    },
    {
      "epoch": 2.2411668036154477,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005670685973985766,
      "loss": 4.7151,
      "step": 16365
    },
    {
      "epoch": 2.2413037523966035,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005670290420316037,
      "loss": 4.7058,
      "step": 16366
    },
    {
      "epoch": 2.2414407011777593,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005669894856947832,
      "loss": 4.7209,
      "step": 16367
    },
    {
      "epoch": 2.2415776499589155,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000566949928388443,
      "loss": 4.7746,
      "step": 16368
    },
    {
      "epoch": 2.2417145987400713,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005669103701129108,
      "loss": 4.7639,
      "step": 16369
    },
    {
      "epoch": 2.241851547521227,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005668708108685147,
      "loss": 4.6857,
      "step": 16370
    },
    {
      "epoch": 2.241988496302383,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005668312506555821,
      "loss": 4.6862,
      "step": 16371
    },
    {
      "epoch": 2.2421254450835386,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005667916894744413,
      "loss": 4.7298,
      "step": 16372
    },
    {
      "epoch": 2.242262393864695,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005667521273254197,
      "loss": 4.6994,
      "step": 16373
    },
    {
      "epoch": 2.2423993426458506,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005667125642088456,
      "loss": 4.65,
      "step": 16374
    },
    {
      "epoch": 2.2425362914270064,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005666730001250466,
      "loss": 4.7383,
      "step": 16375
    },
    {
      "epoch": 2.242673240208162,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005666334350743507,
      "loss": 4.7285,
      "step": 16376
    },
    {
      "epoch": 2.242810188989318,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005665938690570856,
      "loss": 4.7679,
      "step": 16377
    },
    {
      "epoch": 2.2429471377704737,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005665543020735794,
      "loss": 4.7719,
      "step": 16378
    },
    {
      "epoch": 2.2430840865516295,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00056651473412416,
      "loss": 4.6683,
      "step": 16379
    },
    {
      "epoch": 2.2432210353327857,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005664751652091552,
      "loss": 4.7139,
      "step": 16380
    },
    {
      "epoch": 2.2433579841139415,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005664355953288929,
      "loss": 4.718,
      "step": 16381
    },
    {
      "epoch": 2.2434949328950973,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000566396024483701,
      "loss": 4.8,
      "step": 16382
    },
    {
      "epoch": 2.243631881676253,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005663564526739076,
      "loss": 4.7481,
      "step": 16383
    },
    {
      "epoch": 2.243768830457409,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005663168798998406,
      "loss": 4.7323,
      "step": 16384
    },
    {
      "epoch": 2.2439057792385646,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005662773061618279,
      "loss": 4.6868,
      "step": 16385
    },
    {
      "epoch": 2.244042728019721,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005662377314601974,
      "loss": 4.7508,
      "step": 16386
    },
    {
      "epoch": 2.2441796768008766,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005661981557952771,
      "loss": 4.7406,
      "step": 16387
    },
    {
      "epoch": 2.2443166255820324,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000566158579167395,
      "loss": 4.8175,
      "step": 16388
    },
    {
      "epoch": 2.244453574363188,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005661190015768792,
      "loss": 4.6989,
      "step": 16389
    },
    {
      "epoch": 2.244590523144344,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005660794230240574,
      "loss": 4.7764,
      "step": 16390
    },
    {
      "epoch": 2.2447274719254997,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005660398435092578,
      "loss": 4.6982,
      "step": 16391
    },
    {
      "epoch": 2.244864420706656,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005660002630328084,
      "loss": 4.6721,
      "step": 16392
    },
    {
      "epoch": 2.2450013694878117,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005659606815950371,
      "loss": 4.7298,
      "step": 16393
    },
    {
      "epoch": 2.2451383182689675,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005659210991962721,
      "loss": 4.709,
      "step": 16394
    },
    {
      "epoch": 2.2452752670501233,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005658815158368414,
      "loss": 4.7659,
      "step": 16395
    },
    {
      "epoch": 2.245412215831279,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005658419315170728,
      "loss": 4.7472,
      "step": 16396
    },
    {
      "epoch": 2.245549164612435,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005658023462372947,
      "loss": 4.7553,
      "step": 16397
    },
    {
      "epoch": 2.2456861133935906,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005657627599978348,
      "loss": 4.7471,
      "step": 16398
    },
    {
      "epoch": 2.245823062174747,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005657231727990214,
      "loss": 4.779,
      "step": 16399
    },
    {
      "epoch": 2.2459600109559026,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005656835846411823,
      "loss": 4.7653,
      "step": 16400
    },
    {
      "epoch": 2.2460969597370584,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005656439955246459,
      "loss": 4.7631,
      "step": 16401
    },
    {
      "epoch": 2.246233908518214,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005656044054497403,
      "loss": 4.7217,
      "step": 16402
    },
    {
      "epoch": 2.24637085729937,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005655648144167933,
      "loss": 4.6249,
      "step": 16403
    },
    {
      "epoch": 2.2465078060805257,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005655252224261332,
      "loss": 4.7951,
      "step": 16404
    },
    {
      "epoch": 2.246644754861682,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005654856294780879,
      "loss": 4.7693,
      "step": 16405
    },
    {
      "epoch": 2.2467817036428377,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005654460355729859,
      "loss": 4.7338,
      "step": 16406
    },
    {
      "epoch": 2.2469186524239935,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005654064407111549,
      "loss": 4.8312,
      "step": 16407
    },
    {
      "epoch": 2.2470556012051492,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005653668448929234,
      "loss": 4.715,
      "step": 16408
    },
    {
      "epoch": 2.247192549986305,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005653272481186192,
      "loss": 4.6726,
      "step": 16409
    },
    {
      "epoch": 2.247329498767461,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005652876503885706,
      "loss": 4.7635,
      "step": 16410
    },
    {
      "epoch": 2.247466447548617,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000565248051703106,
      "loss": 4.6728,
      "step": 16411
    },
    {
      "epoch": 2.247603396329773,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005652084520625532,
      "loss": 4.7626,
      "step": 16412
    },
    {
      "epoch": 2.2477403451109286,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005651688514672405,
      "loss": 4.7388,
      "step": 16413
    },
    {
      "epoch": 2.2478772938920843,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000565129249917496,
      "loss": 4.7915,
      "step": 16414
    },
    {
      "epoch": 2.24801424267324,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000565089647413648,
      "loss": 4.8153,
      "step": 16415
    },
    {
      "epoch": 2.248151191454396,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005650500439560247,
      "loss": 4.739,
      "step": 16416
    },
    {
      "epoch": 2.2482881402355517,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005650104395449545,
      "loss": 4.7132,
      "step": 16417
    },
    {
      "epoch": 2.248425089016708,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005649708341807651,
      "loss": 4.7484,
      "step": 16418
    },
    {
      "epoch": 2.2485620377978637,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005649312278637851,
      "loss": 4.6616,
      "step": 16419
    },
    {
      "epoch": 2.2486989865790195,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005648916205943426,
      "loss": 4.7853,
      "step": 16420
    },
    {
      "epoch": 2.2488359353601752,
      "grad_norm": 0.375,
      "learning_rate": 0.0005648520123727658,
      "loss": 4.6369,
      "step": 16421
    },
    {
      "epoch": 2.248972884141331,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005648124031993831,
      "loss": 4.7479,
      "step": 16422
    },
    {
      "epoch": 2.249109832922487,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005647727930745226,
      "loss": 4.7652,
      "step": 16423
    },
    {
      "epoch": 2.249246781703643,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005647331819985127,
      "loss": 4.6948,
      "step": 16424
    },
    {
      "epoch": 2.249383730484799,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005646935699716814,
      "loss": 4.7236,
      "step": 16425
    },
    {
      "epoch": 2.2495206792659546,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005646539569943574,
      "loss": 4.7976,
      "step": 16426
    },
    {
      "epoch": 2.2496576280471103,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005646143430668686,
      "loss": 4.7211,
      "step": 16427
    },
    {
      "epoch": 2.249794576828266,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005645747281895435,
      "loss": 4.7525,
      "step": 16428
    },
    {
      "epoch": 2.249931525609422,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005645351123627102,
      "loss": 4.7276,
      "step": 16429
    },
    {
      "epoch": 2.250068474390578,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005644954955866971,
      "loss": 4.7643,
      "step": 16430
    },
    {
      "epoch": 2.250205423171734,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005644558778618328,
      "loss": 4.7276,
      "step": 16431
    },
    {
      "epoch": 2.2503423719528897,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005644162591884451,
      "loss": 4.7136,
      "step": 16432
    },
    {
      "epoch": 2.2504793207340454,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005643766395668628,
      "loss": 4.7509,
      "step": 16433
    },
    {
      "epoch": 2.250616269515201,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005643370189974138,
      "loss": 4.7887,
      "step": 16434
    },
    {
      "epoch": 2.250753218296357,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005642973974804269,
      "loss": 4.7026,
      "step": 16435
    },
    {
      "epoch": 2.2508901670775128,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005642577750162302,
      "loss": 4.676,
      "step": 16436
    },
    {
      "epoch": 2.251027115858669,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005642181516051521,
      "loss": 4.69,
      "step": 16437
    },
    {
      "epoch": 2.2511640646398248,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005641785272475208,
      "loss": 4.7774,
      "step": 16438
    },
    {
      "epoch": 2.2513010134209805,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005641389019436649,
      "loss": 4.7721,
      "step": 16439
    },
    {
      "epoch": 2.2514379622021363,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005640992756939129,
      "loss": 4.8079,
      "step": 16440
    },
    {
      "epoch": 2.251574910983292,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005640596484985928,
      "loss": 4.7538,
      "step": 16441
    },
    {
      "epoch": 2.2517118597644483,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005640200203580333,
      "loss": 4.7125,
      "step": 16442
    },
    {
      "epoch": 2.251848808545604,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005639803912725627,
      "loss": 4.7197,
      "step": 16443
    },
    {
      "epoch": 2.25198575732676,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005639407612425094,
      "loss": 4.6827,
      "step": 16444
    },
    {
      "epoch": 2.2521227061079157,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000563901130268202,
      "loss": 4.6828,
      "step": 16445
    },
    {
      "epoch": 2.2522596548890714,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005638614983499688,
      "loss": 4.6957,
      "step": 16446
    },
    {
      "epoch": 2.252396603670227,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000563821865488138,
      "loss": 4.7491,
      "step": 16447
    },
    {
      "epoch": 2.252533552451383,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005637822316830384,
      "loss": 4.7725,
      "step": 16448
    },
    {
      "epoch": 2.252670501232539,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005637425969349984,
      "loss": 4.7688,
      "step": 16449
    },
    {
      "epoch": 2.252807450013695,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005637029612443463,
      "loss": 4.848,
      "step": 16450
    },
    {
      "epoch": 2.2529443987948508,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005636633246114107,
      "loss": 4.6723,
      "step": 16451
    },
    {
      "epoch": 2.2530813475760065,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005636236870365201,
      "loss": 4.7531,
      "step": 16452
    },
    {
      "epoch": 2.2532182963571623,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005635840485200029,
      "loss": 4.6665,
      "step": 16453
    },
    {
      "epoch": 2.253355245138318,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005635444090621876,
      "loss": 4.7131,
      "step": 16454
    },
    {
      "epoch": 2.2534921939194743,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005635047686634028,
      "loss": 4.7821,
      "step": 16455
    },
    {
      "epoch": 2.25362914270063,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005634651273239769,
      "loss": 4.6658,
      "step": 16456
    },
    {
      "epoch": 2.253766091481786,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005634254850442383,
      "loss": 4.7496,
      "step": 16457
    },
    {
      "epoch": 2.2539030402629416,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005633858418245159,
      "loss": 4.6832,
      "step": 16458
    },
    {
      "epoch": 2.2540399890440974,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005633461976651378,
      "loss": 4.7692,
      "step": 16459
    },
    {
      "epoch": 2.254176937825253,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000563306552566433,
      "loss": 4.7004,
      "step": 16460
    },
    {
      "epoch": 2.2543138866064094,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005632669065287296,
      "loss": 4.7552,
      "step": 16461
    },
    {
      "epoch": 2.254450835387565,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005632272595523564,
      "loss": 4.8034,
      "step": 16462
    },
    {
      "epoch": 2.254587784168721,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005631876116376419,
      "loss": 4.6813,
      "step": 16463
    },
    {
      "epoch": 2.2547247329498767,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005631479627849148,
      "loss": 4.7394,
      "step": 16464
    },
    {
      "epoch": 2.2548616817310325,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005631083129945035,
      "loss": 4.8078,
      "step": 16465
    },
    {
      "epoch": 2.2549986305121883,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005630686622667366,
      "loss": 4.7176,
      "step": 16466
    },
    {
      "epoch": 2.255135579293344,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005630290106019429,
      "loss": 4.7555,
      "step": 16467
    },
    {
      "epoch": 2.2552725280745003,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005629893580004506,
      "loss": 4.6836,
      "step": 16468
    },
    {
      "epoch": 2.255409476855656,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005629497044625889,
      "loss": 4.7312,
      "step": 16469
    },
    {
      "epoch": 2.255546425636812,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005629100499886858,
      "loss": 4.7281,
      "step": 16470
    },
    {
      "epoch": 2.2556833744179676,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005628703945790704,
      "loss": 4.681,
      "step": 16471
    },
    {
      "epoch": 2.2558203231991234,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000562830738234071,
      "loss": 4.6925,
      "step": 16472
    },
    {
      "epoch": 2.2559572719802796,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005627910809540166,
      "loss": 4.7511,
      "step": 16473
    },
    {
      "epoch": 2.2560942207614354,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005627514227392354,
      "loss": 4.7456,
      "step": 16474
    },
    {
      "epoch": 2.256231169542591,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005627117635900564,
      "loss": 4.7318,
      "step": 16475
    },
    {
      "epoch": 2.256368118323747,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005626721035068081,
      "loss": 4.7626,
      "step": 16476
    },
    {
      "epoch": 2.2565050671049027,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005626324424898193,
      "loss": 4.7095,
      "step": 16477
    },
    {
      "epoch": 2.2566420158860585,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005625927805394187,
      "loss": 4.651,
      "step": 16478
    },
    {
      "epoch": 2.2567789646672143,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005625531176559347,
      "loss": 4.7529,
      "step": 16479
    },
    {
      "epoch": 2.2569159134483705,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005625134538396963,
      "loss": 4.7794,
      "step": 16480
    },
    {
      "epoch": 2.2570528622295263,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005624737890910321,
      "loss": 4.7673,
      "step": 16481
    },
    {
      "epoch": 2.257189811010682,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005624341234102707,
      "loss": 4.7011,
      "step": 16482
    },
    {
      "epoch": 2.257326759791838,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000562394456797741,
      "loss": 4.6569,
      "step": 16483
    },
    {
      "epoch": 2.2574637085729936,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005623547892537718,
      "loss": 4.7206,
      "step": 16484
    },
    {
      "epoch": 2.2576006573541494,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005623151207786915,
      "loss": 4.7025,
      "step": 16485
    },
    {
      "epoch": 2.257737606135305,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000562275451372829,
      "loss": 4.6654,
      "step": 16486
    },
    {
      "epoch": 2.2578745549164614,
      "grad_norm": 0.375,
      "learning_rate": 0.0005622357810365133,
      "loss": 4.6884,
      "step": 16487
    },
    {
      "epoch": 2.258011503697617,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005621961097700727,
      "loss": 4.7203,
      "step": 16488
    },
    {
      "epoch": 2.258148452478773,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005621564375738364,
      "loss": 4.6865,
      "step": 16489
    },
    {
      "epoch": 2.2582854012599287,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0005621167644481328,
      "loss": 4.7294,
      "step": 16490
    },
    {
      "epoch": 2.2584223500410845,
      "grad_norm": 0.375,
      "learning_rate": 0.000562077090393291,
      "loss": 4.6941,
      "step": 16491
    },
    {
      "epoch": 2.2585592988222407,
      "grad_norm": 0.421875,
      "learning_rate": 0.0005620374154096396,
      "loss": 4.8061,
      "step": 16492
    },
    {
      "epoch": 2.2586962476033965,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005619977394975075,
      "loss": 4.8602,
      "step": 16493
    },
    {
      "epoch": 2.2588331963845523,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005619580626572233,
      "loss": 4.7536,
      "step": 16494
    },
    {
      "epoch": 2.258970145165708,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005619183848891159,
      "loss": 4.738,
      "step": 16495
    },
    {
      "epoch": 2.259107093946864,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005618787061935144,
      "loss": 4.7187,
      "step": 16496
    },
    {
      "epoch": 2.2592440427280196,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005618390265707473,
      "loss": 4.7393,
      "step": 16497
    },
    {
      "epoch": 2.2593809915091754,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005617993460211436,
      "loss": 4.7351,
      "step": 16498
    },
    {
      "epoch": 2.2595179402903316,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000561759664545032,
      "loss": 4.7421,
      "step": 16499
    },
    {
      "epoch": 2.2596548890714874,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005617199821427415,
      "loss": 4.6462,
      "step": 16500
    },
    {
      "epoch": 2.259791837852643,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005616802988146007,
      "loss": 4.7641,
      "step": 16501
    },
    {
      "epoch": 2.259928786633799,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005616406145609389,
      "loss": 4.7885,
      "step": 16502
    },
    {
      "epoch": 2.2600657354149547,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005616009293820846,
      "loss": 4.6573,
      "step": 16503
    },
    {
      "epoch": 2.2602026841961105,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005615612432783668,
      "loss": 4.7769,
      "step": 16504
    },
    {
      "epoch": 2.2603396329772663,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005615215562501145,
      "loss": 4.6774,
      "step": 16505
    },
    {
      "epoch": 2.2604765817584225,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005614818682976563,
      "loss": 4.7252,
      "step": 16506
    },
    {
      "epoch": 2.2606135305395783,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005614421794213214,
      "loss": 4.7366,
      "step": 16507
    },
    {
      "epoch": 2.260750479320734,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005614024896214386,
      "loss": 4.7643,
      "step": 16508
    },
    {
      "epoch": 2.26088742810189,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000561362798898337,
      "loss": 4.7676,
      "step": 16509
    },
    {
      "epoch": 2.2610243768830456,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000561323107252345,
      "loss": 4.7886,
      "step": 16510
    },
    {
      "epoch": 2.261161325664202,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005612834146837921,
      "loss": 4.608,
      "step": 16511
    },
    {
      "epoch": 2.2612982744453576,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005612437211930069,
      "loss": 4.7518,
      "step": 16512
    },
    {
      "epoch": 2.2614352232265134,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005612040267803186,
      "loss": 4.7298,
      "step": 16513
    },
    {
      "epoch": 2.261572172007669,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000561164331446056,
      "loss": 4.7232,
      "step": 16514
    },
    {
      "epoch": 2.261709120788825,
      "grad_norm": 0.408203125,
      "learning_rate": 0.000561124635190548,
      "loss": 4.7002,
      "step": 16515
    },
    {
      "epoch": 2.2618460695699807,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005610849380141238,
      "loss": 4.6809,
      "step": 16516
    },
    {
      "epoch": 2.2619830183511365,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005610452399171123,
      "loss": 4.647,
      "step": 16517
    },
    {
      "epoch": 2.2621199671322927,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005610055408998423,
      "loss": 4.7251,
      "step": 16518
    },
    {
      "epoch": 2.2622569159134485,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005609658409626429,
      "loss": 4.8093,
      "step": 16519
    },
    {
      "epoch": 2.2623938646946042,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005609261401058432,
      "loss": 4.6883,
      "step": 16520
    },
    {
      "epoch": 2.26253081347576,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005608864383297721,
      "loss": 4.7574,
      "step": 16521
    },
    {
      "epoch": 2.262667762256916,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0005608467356347588,
      "loss": 4.751,
      "step": 16522
    },
    {
      "epoch": 2.2628047110380716,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005608070320211321,
      "loss": 4.7513,
      "step": 16523
    },
    {
      "epoch": 2.262941659819228,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0005607673274892211,
      "loss": 4.6665,
      "step": 16524
    },
    {
      "epoch": 2.2630786086003836,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005607276220393549,
      "loss": 4.6055,
      "step": 16525
    },
    {
      "epoch": 2.2632155573815393,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005606879156718625,
      "loss": 4.7142,
      "step": 16526
    },
    {
      "epoch": 2.263352506162695,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005606482083870729,
      "loss": 4.6701,
      "step": 16527
    },
    {
      "epoch": 2.263489454943851,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005606085001853153,
      "loss": 4.7714,
      "step": 16528
    },
    {
      "epoch": 2.2636264037250067,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005605687910669187,
      "loss": 4.7124,
      "step": 16529
    },
    {
      "epoch": 2.263763352506163,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005605290810322122,
      "loss": 4.722,
      "step": 16530
    },
    {
      "epoch": 2.2639003012873187,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005604893700815248,
      "loss": 4.7246,
      "step": 16531
    },
    {
      "epoch": 2.2640372500684744,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005604496582151858,
      "loss": 4.7405,
      "step": 16532
    },
    {
      "epoch": 2.2641741988496302,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000560409945433524,
      "loss": 4.7029,
      "step": 16533
    },
    {
      "epoch": 2.264311147630786,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005603702317368688,
      "loss": 4.8039,
      "step": 16534
    },
    {
      "epoch": 2.264448096411942,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005603305171255491,
      "loss": 4.6744,
      "step": 16535
    },
    {
      "epoch": 2.2645850451930976,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005602908015998941,
      "loss": 4.7731,
      "step": 16536
    },
    {
      "epoch": 2.2647219939742538,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000560251085160233,
      "loss": 4.6943,
      "step": 16537
    },
    {
      "epoch": 2.2648589427554096,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005602113678068949,
      "loss": 4.7189,
      "step": 16538
    },
    {
      "epoch": 2.2649958915365653,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000560171649540209,
      "loss": 4.75,
      "step": 16539
    },
    {
      "epoch": 2.265132840317721,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005601319303605042,
      "loss": 4.7774,
      "step": 16540
    },
    {
      "epoch": 2.265269789098877,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005600922102681099,
      "loss": 4.779,
      "step": 16541
    },
    {
      "epoch": 2.265406737880033,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005600524892633553,
      "loss": 4.6491,
      "step": 16542
    },
    {
      "epoch": 2.265543686661189,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005600127673465695,
      "loss": 4.7512,
      "step": 16543
    },
    {
      "epoch": 2.2656806354423447,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005599730445180817,
      "loss": 4.7753,
      "step": 16544
    },
    {
      "epoch": 2.2658175842235004,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005599333207782209,
      "loss": 4.7783,
      "step": 16545
    },
    {
      "epoch": 2.265954533004656,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005598935961273167,
      "loss": 4.7887,
      "step": 16546
    },
    {
      "epoch": 2.266091481785812,
      "grad_norm": 0.3203125,
      "learning_rate": 0.000559853870565698,
      "loss": 4.7653,
      "step": 16547
    },
    {
      "epoch": 2.2662284305669678,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000559814144093694,
      "loss": 4.7872,
      "step": 16548
    },
    {
      "epoch": 2.266365379348124,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005597744167116341,
      "loss": 4.6616,
      "step": 16549
    },
    {
      "epoch": 2.2665023281292798,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005597346884198474,
      "loss": 4.7208,
      "step": 16550
    },
    {
      "epoch": 2.2666392769104355,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005596949592186633,
      "loss": 4.743,
      "step": 16551
    },
    {
      "epoch": 2.2667762256915913,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005596552291084109,
      "loss": 4.7853,
      "step": 16552
    },
    {
      "epoch": 2.266913174472747,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005596154980894195,
      "loss": 4.6854,
      "step": 16553
    },
    {
      "epoch": 2.267050123253903,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005595757661620182,
      "loss": 4.7994,
      "step": 16554
    },
    {
      "epoch": 2.2671870720350586,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005595360333265365,
      "loss": 4.784,
      "step": 16555
    },
    {
      "epoch": 2.267324020816215,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005594962995833036,
      "loss": 4.7955,
      "step": 16556
    },
    {
      "epoch": 2.2674609695973706,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005594565649326488,
      "loss": 4.7669,
      "step": 16557
    },
    {
      "epoch": 2.2675979183785264,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005594168293749014,
      "loss": 4.6997,
      "step": 16558
    },
    {
      "epoch": 2.267734867159682,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005593770929103905,
      "loss": 4.729,
      "step": 16559
    },
    {
      "epoch": 2.267871815940838,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005593373555394458,
      "loss": 4.7087,
      "step": 16560
    },
    {
      "epoch": 2.268008764721994,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005592976172623963,
      "loss": 4.7717,
      "step": 16561
    },
    {
      "epoch": 2.26814571350315,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005592578780795713,
      "loss": 4.7702,
      "step": 16562
    },
    {
      "epoch": 2.2682826622843058,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005592181379913004,
      "loss": 4.7507,
      "step": 16563
    },
    {
      "epoch": 2.2684196110654615,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005591783969979125,
      "loss": 4.7532,
      "step": 16564
    },
    {
      "epoch": 2.2685565598466173,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005591386550997375,
      "loss": 4.7545,
      "step": 16565
    },
    {
      "epoch": 2.268693508627773,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005590989122971042,
      "loss": 4.7913,
      "step": 16566
    },
    {
      "epoch": 2.268830457408929,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005590591685903425,
      "loss": 4.7437,
      "step": 16567
    },
    {
      "epoch": 2.268967406190085,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005590194239797812,
      "loss": 4.773,
      "step": 16568
    },
    {
      "epoch": 2.269104354971241,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005589796784657501,
      "loss": 4.7111,
      "step": 16569
    },
    {
      "epoch": 2.2692413037523966,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005589399320485784,
      "loss": 4.6953,
      "step": 16570
    },
    {
      "epoch": 2.2693782525335524,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005589001847285956,
      "loss": 4.7433,
      "step": 16571
    },
    {
      "epoch": 2.269515201314708,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005588604365061309,
      "loss": 4.7969,
      "step": 16572
    },
    {
      "epoch": 2.269652150095864,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005588206873815139,
      "loss": 4.6868,
      "step": 16573
    },
    {
      "epoch": 2.26978909887702,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005587809373550739,
      "loss": 4.6912,
      "step": 16574
    },
    {
      "epoch": 2.269926047658176,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005587411864271405,
      "loss": 4.7374,
      "step": 16575
    },
    {
      "epoch": 2.2700629964393317,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005587014345980429,
      "loss": 4.6916,
      "step": 16576
    },
    {
      "epoch": 2.2701999452204875,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0005586616818681104,
      "loss": 4.6618,
      "step": 16577
    },
    {
      "epoch": 2.2703368940016433,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005586219282376727,
      "loss": 4.6664,
      "step": 16578
    },
    {
      "epoch": 2.270473842782799,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005585821737070594,
      "loss": 4.7825,
      "step": 16579
    },
    {
      "epoch": 2.2706107915639553,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005585424182765996,
      "loss": 4.6733,
      "step": 16580
    },
    {
      "epoch": 2.270747740345111,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005585026619466231,
      "loss": 4.7901,
      "step": 16581
    },
    {
      "epoch": 2.270884689126267,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000558462904717459,
      "loss": 4.7662,
      "step": 16582
    },
    {
      "epoch": 2.2710216379074226,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000558423146589437,
      "loss": 4.6416,
      "step": 16583
    },
    {
      "epoch": 2.2711585866885784,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005583833875628866,
      "loss": 4.6629,
      "step": 16584
    },
    {
      "epoch": 2.271295535469734,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005583436276381373,
      "loss": 4.7648,
      "step": 16585
    },
    {
      "epoch": 2.27143248425089,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005583038668155184,
      "loss": 4.7195,
      "step": 16586
    },
    {
      "epoch": 2.271569433032046,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005582641050953597,
      "loss": 4.627,
      "step": 16587
    },
    {
      "epoch": 2.271706381813202,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005582243424779903,
      "loss": 4.7718,
      "step": 16588
    },
    {
      "epoch": 2.2718433305943577,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005581845789637401,
      "loss": 4.8,
      "step": 16589
    },
    {
      "epoch": 2.2719802793755135,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005581448145529387,
      "loss": 4.7123,
      "step": 16590
    },
    {
      "epoch": 2.2721172281566693,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005581050492459152,
      "loss": 4.8533,
      "step": 16591
    },
    {
      "epoch": 2.2722541769378255,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005580652830429996,
      "loss": 4.7579,
      "step": 16592
    },
    {
      "epoch": 2.2723911257189813,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005580255159445211,
      "loss": 4.7307,
      "step": 16593
    },
    {
      "epoch": 2.272528074500137,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005579857479508095,
      "loss": 4.6523,
      "step": 16594
    },
    {
      "epoch": 2.272665023281293,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005579459790621943,
      "loss": 4.7071,
      "step": 16595
    },
    {
      "epoch": 2.2728019720624486,
      "grad_norm": 0.359375,
      "learning_rate": 0.000557906209279005,
      "loss": 4.7401,
      "step": 16596
    },
    {
      "epoch": 2.2729389208436044,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005578664386015711,
      "loss": 4.7232,
      "step": 16597
    },
    {
      "epoch": 2.27307586962476,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005578266670302224,
      "loss": 4.6983,
      "step": 16598
    },
    {
      "epoch": 2.2732128184059164,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005577868945652885,
      "loss": 4.7243,
      "step": 16599
    },
    {
      "epoch": 2.273349767187072,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005577471212070989,
      "loss": 4.7233,
      "step": 16600
    },
    {
      "epoch": 2.273486715968228,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005577073469559833,
      "loss": 4.7664,
      "step": 16601
    },
    {
      "epoch": 2.2736236647493837,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005576675718122711,
      "loss": 4.6881,
      "step": 16602
    },
    {
      "epoch": 2.2737606135305395,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005576277957762922,
      "loss": 4.7246,
      "step": 16603
    },
    {
      "epoch": 2.2738975623116953,
      "grad_norm": 0.34375,
      "learning_rate": 0.000557588018848376,
      "loss": 4.6846,
      "step": 16604
    },
    {
      "epoch": 2.274034511092851,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005575482410288524,
      "loss": 4.7549,
      "step": 16605
    },
    {
      "epoch": 2.2741714598740073,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005575084623180507,
      "loss": 4.7633,
      "step": 16606
    },
    {
      "epoch": 2.274308408655163,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005574686827163008,
      "loss": 4.8152,
      "step": 16607
    },
    {
      "epoch": 2.274445357436319,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005574289022239325,
      "loss": 4.6692,
      "step": 16608
    },
    {
      "epoch": 2.2745823062174746,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005573891208412751,
      "loss": 4.7779,
      "step": 16609
    },
    {
      "epoch": 2.2747192549986304,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005573493385686585,
      "loss": 4.7202,
      "step": 16610
    },
    {
      "epoch": 2.2748562037797866,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005573095554064125,
      "loss": 4.6855,
      "step": 16611
    },
    {
      "epoch": 2.2749931525609424,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005572697713548665,
      "loss": 4.7205,
      "step": 16612
    },
    {
      "epoch": 2.275130101342098,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005572299864143503,
      "loss": 4.6527,
      "step": 16613
    },
    {
      "epoch": 2.275267050123254,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005571902005851937,
      "loss": 4.6982,
      "step": 16614
    },
    {
      "epoch": 2.2754039989044097,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005571504138677264,
      "loss": 4.7506,
      "step": 16615
    },
    {
      "epoch": 2.2755409476855655,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000557110626262278,
      "loss": 4.7421,
      "step": 16616
    },
    {
      "epoch": 2.2756778964667213,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005570708377691784,
      "loss": 4.6724,
      "step": 16617
    },
    {
      "epoch": 2.2758148452478775,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005570310483887572,
      "loss": 4.7499,
      "step": 16618
    },
    {
      "epoch": 2.2759517940290332,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005569912581213443,
      "loss": 4.6874,
      "step": 16619
    },
    {
      "epoch": 2.276088742810189,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005569514669672693,
      "loss": 4.7885,
      "step": 16620
    },
    {
      "epoch": 2.276225691591345,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000556911674926862,
      "loss": 4.7462,
      "step": 16621
    },
    {
      "epoch": 2.2763626403725006,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005568718820004522,
      "loss": 4.6782,
      "step": 16622
    },
    {
      "epoch": 2.2764995891536564,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005568320881883696,
      "loss": 4.7286,
      "step": 16623
    },
    {
      "epoch": 2.276636537934812,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005567922934909441,
      "loss": 4.7456,
      "step": 16624
    },
    {
      "epoch": 2.2767734867159684,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005567524979085053,
      "loss": 4.8293,
      "step": 16625
    },
    {
      "epoch": 2.276910435497124,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005567127014413832,
      "loss": 4.7222,
      "step": 16626
    },
    {
      "epoch": 2.27704738427828,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005566729040899075,
      "loss": 4.7555,
      "step": 16627
    },
    {
      "epoch": 2.2771843330594357,
      "grad_norm": 0.375,
      "learning_rate": 0.0005566331058544081,
      "loss": 4.7251,
      "step": 16628
    },
    {
      "epoch": 2.2773212818405915,
      "grad_norm": 0.375,
      "learning_rate": 0.0005565933067352145,
      "loss": 4.6948,
      "step": 16629
    },
    {
      "epoch": 2.2774582306217477,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000556553506732657,
      "loss": 4.6886,
      "step": 16630
    },
    {
      "epoch": 2.2775951794029035,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005565137058470652,
      "loss": 4.764,
      "step": 16631
    },
    {
      "epoch": 2.2777321281840592,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005564739040787688,
      "loss": 4.7255,
      "step": 16632
    },
    {
      "epoch": 2.277869076965215,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005564341014280979,
      "loss": 4.6899,
      "step": 16633
    },
    {
      "epoch": 2.278006025746371,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005563942978953821,
      "loss": 4.6461,
      "step": 16634
    },
    {
      "epoch": 2.2781429745275266,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005563544934809515,
      "loss": 4.7842,
      "step": 16635
    },
    {
      "epoch": 2.2782799233086823,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005563146881851359,
      "loss": 4.7741,
      "step": 16636
    },
    {
      "epoch": 2.2784168720898386,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005562748820082651,
      "loss": 4.7278,
      "step": 16637
    },
    {
      "epoch": 2.2785538208709943,
      "grad_norm": 0.359375,
      "learning_rate": 0.000556235074950669,
      "loss": 4.6891,
      "step": 16638
    },
    {
      "epoch": 2.27869076965215,
      "grad_norm": 0.375,
      "learning_rate": 0.0005561952670126776,
      "loss": 4.7665,
      "step": 16639
    },
    {
      "epoch": 2.278827718433306,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005561554581946206,
      "loss": 4.6439,
      "step": 16640
    },
    {
      "epoch": 2.2789646672144617,
      "grad_norm": 0.375,
      "learning_rate": 0.0005561156484968281,
      "loss": 4.6474,
      "step": 16641
    },
    {
      "epoch": 2.2791016159956174,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005560758379196299,
      "loss": 4.7515,
      "step": 16642
    },
    {
      "epoch": 2.2792385647767737,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005560360264633561,
      "loss": 4.8098,
      "step": 16643
    },
    {
      "epoch": 2.2793755135579294,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005559962141283363,
      "loss": 4.7423,
      "step": 16644
    },
    {
      "epoch": 2.279512462339085,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005559564009149007,
      "loss": 4.8052,
      "step": 16645
    },
    {
      "epoch": 2.279649411120241,
      "grad_norm": 0.375,
      "learning_rate": 0.0005559165868233791,
      "loss": 4.7695,
      "step": 16646
    },
    {
      "epoch": 2.2797863599013968,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005558767718541015,
      "loss": 4.62,
      "step": 16647
    },
    {
      "epoch": 2.2799233086825526,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000555836956007398,
      "loss": 4.7436,
      "step": 16648
    },
    {
      "epoch": 2.2800602574637088,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005557971392835983,
      "loss": 4.6974,
      "step": 16649
    },
    {
      "epoch": 2.2801972062448645,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005557573216830326,
      "loss": 4.7456,
      "step": 16650
    },
    {
      "epoch": 2.2803341550260203,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005557175032060306,
      "loss": 4.7853,
      "step": 16651
    },
    {
      "epoch": 2.280471103807176,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005556776838529226,
      "loss": 4.7336,
      "step": 16652
    },
    {
      "epoch": 2.280608052588332,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005556378636240385,
      "loss": 4.7057,
      "step": 16653
    },
    {
      "epoch": 2.2807450013694877,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000555598042519708,
      "loss": 4.6515,
      "step": 16654
    },
    {
      "epoch": 2.2808819501506434,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005555582205402617,
      "loss": 4.7866,
      "step": 16655
    },
    {
      "epoch": 2.2810188989317997,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000555518397686029,
      "loss": 4.6725,
      "step": 16656
    },
    {
      "epoch": 2.2811558477129554,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005554785739573404,
      "loss": 4.7903,
      "step": 16657
    },
    {
      "epoch": 2.281292796494111,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005554387493545255,
      "loss": 4.6432,
      "step": 16658
    },
    {
      "epoch": 2.281429745275267,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005553989238779148,
      "loss": 4.7883,
      "step": 16659
    },
    {
      "epoch": 2.2815666940564228,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005553590975278379,
      "loss": 4.7195,
      "step": 16660
    },
    {
      "epoch": 2.281703642837579,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005553192703046252,
      "loss": 4.6841,
      "step": 16661
    },
    {
      "epoch": 2.2818405916187348,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005552794422086064,
      "loss": 4.7544,
      "step": 16662
    },
    {
      "epoch": 2.2819775403998905,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005552396132401119,
      "loss": 4.7863,
      "step": 16663
    },
    {
      "epoch": 2.2821144891810463,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005551997833994718,
      "loss": 4.7609,
      "step": 16664
    },
    {
      "epoch": 2.282251437962202,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005551599526870158,
      "loss": 4.7514,
      "step": 16665
    },
    {
      "epoch": 2.282388386743358,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005551201211030744,
      "loss": 4.7154,
      "step": 16666
    },
    {
      "epoch": 2.2825253355245136,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005550802886479774,
      "loss": 4.7069,
      "step": 16667
    },
    {
      "epoch": 2.28266228430567,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000555040455322055,
      "loss": 4.7434,
      "step": 16668
    },
    {
      "epoch": 2.2827992330868256,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005550006211256373,
      "loss": 4.7717,
      "step": 16669
    },
    {
      "epoch": 2.2829361818679814,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005549607860590546,
      "loss": 4.6999,
      "step": 16670
    },
    {
      "epoch": 2.283073130649137,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005549209501226367,
      "loss": 4.7426,
      "step": 16671
    },
    {
      "epoch": 2.283210079430293,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005548811133167139,
      "loss": 4.7492,
      "step": 16672
    },
    {
      "epoch": 2.2833470282114487,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005548412756416163,
      "loss": 4.7349,
      "step": 16673
    },
    {
      "epoch": 2.2834839769926045,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005548014370976741,
      "loss": 4.7605,
      "step": 16674
    },
    {
      "epoch": 2.2836209257737607,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005547615976852175,
      "loss": 4.7852,
      "step": 16675
    },
    {
      "epoch": 2.2837578745549165,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005547217574045764,
      "loss": 4.6951,
      "step": 16676
    },
    {
      "epoch": 2.2838948233360723,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005546819162560813,
      "loss": 4.784,
      "step": 16677
    },
    {
      "epoch": 2.284031772117228,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005546420742400622,
      "loss": 4.69,
      "step": 16678
    },
    {
      "epoch": 2.284168720898384,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005546022313568493,
      "loss": 4.7759,
      "step": 16679
    },
    {
      "epoch": 2.28430566967954,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005545623876067726,
      "loss": 4.7054,
      "step": 16680
    },
    {
      "epoch": 2.284442618460696,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005545225429901626,
      "loss": 4.6877,
      "step": 16681
    },
    {
      "epoch": 2.2845795672418516,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005544826975073495,
      "loss": 4.6655,
      "step": 16682
    },
    {
      "epoch": 2.2847165160230074,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005544428511586632,
      "loss": 4.7999,
      "step": 16683
    },
    {
      "epoch": 2.284853464804163,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005544030039444343,
      "loss": 4.7011,
      "step": 16684
    },
    {
      "epoch": 2.284990413585319,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005543631558649927,
      "loss": 4.7292,
      "step": 16685
    },
    {
      "epoch": 2.2851273623664747,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005543233069206688,
      "loss": 4.7453,
      "step": 16686
    },
    {
      "epoch": 2.285264311147631,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005542834571117927,
      "loss": 4.7591,
      "step": 16687
    },
    {
      "epoch": 2.2854012599287867,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005542436064386948,
      "loss": 4.7661,
      "step": 16688
    },
    {
      "epoch": 2.2855382087099425,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005542037549017053,
      "loss": 4.7665,
      "step": 16689
    },
    {
      "epoch": 2.2856751574910983,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005541639025011544,
      "loss": 4.6987,
      "step": 16690
    },
    {
      "epoch": 2.285812106272254,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005541240492373726,
      "loss": 4.7401,
      "step": 16691
    },
    {
      "epoch": 2.28594905505341,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005540841951106898,
      "loss": 4.6884,
      "step": 16692
    },
    {
      "epoch": 2.286086003834566,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005540443401214365,
      "loss": 4.675,
      "step": 16693
    },
    {
      "epoch": 2.286222952615722,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000554004484269943,
      "loss": 4.6687,
      "step": 16694
    },
    {
      "epoch": 2.2863599013968776,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005539646275565396,
      "loss": 4.6861,
      "step": 16695
    },
    {
      "epoch": 2.2864968501780334,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005539247699815566,
      "loss": 4.7439,
      "step": 16696
    },
    {
      "epoch": 2.286633798959189,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005538849115453242,
      "loss": 4.7118,
      "step": 16697
    },
    {
      "epoch": 2.286770747740345,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005538450522481728,
      "loss": 4.7792,
      "step": 16698
    },
    {
      "epoch": 2.286907696521501,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005538051920904326,
      "loss": 4.7348,
      "step": 16699
    },
    {
      "epoch": 2.287044645302657,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005537653310724341,
      "loss": 4.6646,
      "step": 16700
    },
    {
      "epoch": 2.2871815940838127,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005537254691945076,
      "loss": 4.7957,
      "step": 16701
    },
    {
      "epoch": 2.2873185428649685,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005536856064569834,
      "loss": 4.6975,
      "step": 16702
    },
    {
      "epoch": 2.2874554916461243,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005536457428601919,
      "loss": 4.7374,
      "step": 16703
    },
    {
      "epoch": 2.28759244042728,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005536058784044634,
      "loss": 4.7611,
      "step": 16704
    },
    {
      "epoch": 2.287729389208436,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005535660130901283,
      "loss": 4.7593,
      "step": 16705
    },
    {
      "epoch": 2.287866337989592,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005535261469175171,
      "loss": 4.6696,
      "step": 16706
    },
    {
      "epoch": 2.288003286770748,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005534862798869599,
      "loss": 4.8861,
      "step": 16707
    },
    {
      "epoch": 2.2881402355519036,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005534464119987873,
      "loss": 4.7182,
      "step": 16708
    },
    {
      "epoch": 2.2882771843330594,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005534065432533295,
      "loss": 4.796,
      "step": 16709
    },
    {
      "epoch": 2.288414133114215,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005533666736509172,
      "loss": 4.7118,
      "step": 16710
    },
    {
      "epoch": 2.2885510818953714,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005533268031918806,
      "loss": 4.7792,
      "step": 16711
    },
    {
      "epoch": 2.288688030676527,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00055328693187655,
      "loss": 4.7454,
      "step": 16712
    },
    {
      "epoch": 2.288824979457683,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005532470597052562,
      "loss": 4.6781,
      "step": 16713
    },
    {
      "epoch": 2.2889619282388387,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005532071866783294,
      "loss": 4.745,
      "step": 16714
    },
    {
      "epoch": 2.2890988770199945,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005531673127961,
      "loss": 4.7215,
      "step": 16715
    },
    {
      "epoch": 2.2892358258011503,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005531274380588984,
      "loss": 4.6725,
      "step": 16716
    },
    {
      "epoch": 2.289372774582306,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005530875624670554,
      "loss": 4.7004,
      "step": 16717
    },
    {
      "epoch": 2.2895097233634623,
      "grad_norm": 0.359375,
      "learning_rate": 0.000553047686020901,
      "loss": 4.676,
      "step": 16718
    },
    {
      "epoch": 2.289646672144618,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005530078087207659,
      "loss": 4.702,
      "step": 16719
    },
    {
      "epoch": 2.289783620925774,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005529679305669806,
      "loss": 4.7296,
      "step": 16720
    },
    {
      "epoch": 2.2899205697069296,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005529280515598753,
      "loss": 4.6799,
      "step": 16721
    },
    {
      "epoch": 2.2900575184880854,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005528881716997811,
      "loss": 4.7707,
      "step": 16722
    },
    {
      "epoch": 2.290194467269241,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005528482909870278,
      "loss": 4.751,
      "step": 16723
    },
    {
      "epoch": 2.290331416050397,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005528084094219463,
      "loss": 4.7714,
      "step": 16724
    },
    {
      "epoch": 2.290468364831553,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005527685270048671,
      "loss": 4.7642,
      "step": 16725
    },
    {
      "epoch": 2.290605313612709,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005527286437361206,
      "loss": 4.6468,
      "step": 16726
    },
    {
      "epoch": 2.2907422623938647,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005526887596160372,
      "loss": 4.7618,
      "step": 16727
    },
    {
      "epoch": 2.2908792111750205,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005526488746449477,
      "loss": 4.7094,
      "step": 16728
    },
    {
      "epoch": 2.2910161599561762,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005526089888231824,
      "loss": 4.6832,
      "step": 16729
    },
    {
      "epoch": 2.2911531087373325,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000552569102151072,
      "loss": 4.7196,
      "step": 16730
    },
    {
      "epoch": 2.2912900575184882,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005525292146289473,
      "loss": 4.8412,
      "step": 16731
    },
    {
      "epoch": 2.291427006299644,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005524893262571381,
      "loss": 4.6585,
      "step": 16732
    },
    {
      "epoch": 2.2915639550808,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005524494370359758,
      "loss": 4.7397,
      "step": 16733
    },
    {
      "epoch": 2.2917009038619556,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005524095469657904,
      "loss": 4.7323,
      "step": 16734
    },
    {
      "epoch": 2.2918378526431114,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005523696560469128,
      "loss": 4.7376,
      "step": 16735
    },
    {
      "epoch": 2.291974801424267,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005523297642796734,
      "loss": 4.7193,
      "step": 16736
    },
    {
      "epoch": 2.2921117502054233,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005522898716644028,
      "loss": 4.7093,
      "step": 16737
    },
    {
      "epoch": 2.292248698986579,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005522499782014317,
      "loss": 4.7402,
      "step": 16738
    },
    {
      "epoch": 2.292385647767735,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005522100838910907,
      "loss": 4.7781,
      "step": 16739
    },
    {
      "epoch": 2.2925225965488907,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005521701887337105,
      "loss": 4.7365,
      "step": 16740
    },
    {
      "epoch": 2.2926595453300465,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005521302927296213,
      "loss": 4.6657,
      "step": 16741
    },
    {
      "epoch": 2.2927964941112022,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005520903958791543,
      "loss": 4.7487,
      "step": 16742
    },
    {
      "epoch": 2.292933442892358,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005520504981826397,
      "loss": 4.7779,
      "step": 16743
    },
    {
      "epoch": 2.2930703916735142,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005520105996404084,
      "loss": 4.7548,
      "step": 16744
    },
    {
      "epoch": 2.29320734045467,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005519707002527909,
      "loss": 4.7021,
      "step": 16745
    },
    {
      "epoch": 2.293344289235826,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005519308000201179,
      "loss": 4.8116,
      "step": 16746
    },
    {
      "epoch": 2.2934812380169816,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005518908989427202,
      "loss": 4.733,
      "step": 16747
    },
    {
      "epoch": 2.2936181867981373,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005518509970209282,
      "loss": 4.7052,
      "step": 16748
    },
    {
      "epoch": 2.2937551355792936,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005518110942550729,
      "loss": 4.7753,
      "step": 16749
    },
    {
      "epoch": 2.2938920843604493,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005517711906454846,
      "loss": 4.7454,
      "step": 16750
    },
    {
      "epoch": 2.294029033141605,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005517312861924943,
      "loss": 4.7176,
      "step": 16751
    },
    {
      "epoch": 2.294165981922761,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005516913808964324,
      "loss": 4.7746,
      "step": 16752
    },
    {
      "epoch": 2.2943029307039167,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00055165147475763,
      "loss": 4.7554,
      "step": 16753
    },
    {
      "epoch": 2.2944398794850724,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005516115677764176,
      "loss": 4.6214,
      "step": 16754
    },
    {
      "epoch": 2.294576828266228,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005515716599531257,
      "loss": 4.8391,
      "step": 16755
    },
    {
      "epoch": 2.2947137770473844,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005515317512880855,
      "loss": 4.7931,
      "step": 16756
    },
    {
      "epoch": 2.29485072582854,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005514918417816275,
      "loss": 4.7511,
      "step": 16757
    },
    {
      "epoch": 2.294987674609696,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005514519314340823,
      "loss": 4.6548,
      "step": 16758
    },
    {
      "epoch": 2.2951246233908518,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005514120202457808,
      "loss": 4.7678,
      "step": 16759
    },
    {
      "epoch": 2.2952615721720075,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005513721082170538,
      "loss": 4.7326,
      "step": 16760
    },
    {
      "epoch": 2.2953985209531633,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005513321953482317,
      "loss": 4.6669,
      "step": 16761
    },
    {
      "epoch": 2.2955354697343195,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005512922816396458,
      "loss": 4.8358,
      "step": 16762
    },
    {
      "epoch": 2.2956724185154753,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005512523670916266,
      "loss": 4.7533,
      "step": 16763
    },
    {
      "epoch": 2.295809367296631,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005512124517045049,
      "loss": 4.7459,
      "step": 16764
    },
    {
      "epoch": 2.295946316077787,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005511725354786115,
      "loss": 4.8013,
      "step": 16765
    },
    {
      "epoch": 2.2960832648589427,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005511326184142771,
      "loss": 4.7099,
      "step": 16766
    },
    {
      "epoch": 2.2962202136400984,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005510927005118327,
      "loss": 4.6964,
      "step": 16767
    },
    {
      "epoch": 2.2963571624212546,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005510527817716089,
      "loss": 4.7018,
      "step": 16768
    },
    {
      "epoch": 2.2964941112024104,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005510128621939367,
      "loss": 4.7059,
      "step": 16769
    },
    {
      "epoch": 2.296631059983566,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005509729417791469,
      "loss": 4.6574,
      "step": 16770
    },
    {
      "epoch": 2.296768008764722,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005509330205275702,
      "loss": 4.7392,
      "step": 16771
    },
    {
      "epoch": 2.2969049575458778,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005508930984395376,
      "loss": 4.7202,
      "step": 16772
    },
    {
      "epoch": 2.2970419063270335,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005508531755153797,
      "loss": 4.7398,
      "step": 16773
    },
    {
      "epoch": 2.2971788551081893,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005508132517554274,
      "loss": 4.7573,
      "step": 16774
    },
    {
      "epoch": 2.2973158038893455,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005507733271600119,
      "loss": 4.7459,
      "step": 16775
    },
    {
      "epoch": 2.2974527526705013,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005507334017294637,
      "loss": 4.7278,
      "step": 16776
    },
    {
      "epoch": 2.297589701451657,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0005506934754641138,
      "loss": 4.7719,
      "step": 16777
    },
    {
      "epoch": 2.297726650232813,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005506535483642932,
      "loss": 4.8006,
      "step": 16778
    },
    {
      "epoch": 2.2978635990139686,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005506136204303326,
      "loss": 4.7621,
      "step": 16779
    },
    {
      "epoch": 2.298000547795125,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005505736916625629,
      "loss": 4.7742,
      "step": 16780
    },
    {
      "epoch": 2.2981374965762806,
      "grad_norm": 0.4296875,
      "learning_rate": 0.000550533762061315,
      "loss": 4.6751,
      "step": 16781
    },
    {
      "epoch": 2.2982744453574364,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00055049383162692,
      "loss": 4.7378,
      "step": 16782
    },
    {
      "epoch": 2.298411394138592,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005504539003597086,
      "loss": 4.788,
      "step": 16783
    },
    {
      "epoch": 2.298548342919748,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005504139682600118,
      "loss": 4.7617,
      "step": 16784
    },
    {
      "epoch": 2.2986852917009037,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005503740353281605,
      "loss": 4.738,
      "step": 16785
    },
    {
      "epoch": 2.2988222404820595,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005503341015644858,
      "loss": 4.788,
      "step": 16786
    },
    {
      "epoch": 2.2989591892632157,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005502941669693183,
      "loss": 4.7662,
      "step": 16787
    },
    {
      "epoch": 2.2990961380443715,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005502542315429891,
      "loss": 4.7774,
      "step": 16788
    },
    {
      "epoch": 2.2992330868255273,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005502142952858294,
      "loss": 4.6926,
      "step": 16789
    },
    {
      "epoch": 2.299370035606683,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005501743581981698,
      "loss": 4.7808,
      "step": 16790
    },
    {
      "epoch": 2.299506984387839,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005501344202803414,
      "loss": 4.7433,
      "step": 16791
    },
    {
      "epoch": 2.2996439331689946,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005500944815326753,
      "loss": 4.7323,
      "step": 16792
    },
    {
      "epoch": 2.2997808819501504,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005500545419555024,
      "loss": 4.7685,
      "step": 16793
    },
    {
      "epoch": 2.2999178307313066,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005500146015491536,
      "loss": 4.7208,
      "step": 16794
    },
    {
      "epoch": 2.3000547795124624,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005499746603139599,
      "loss": 4.7926,
      "step": 16795
    },
    {
      "epoch": 2.300191728293618,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005499347182502525,
      "loss": 4.7494,
      "step": 16796
    },
    {
      "epoch": 2.300328677074774,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005498947753583622,
      "loss": 4.7854,
      "step": 16797
    },
    {
      "epoch": 2.3004656258559297,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005498548316386203,
      "loss": 4.6825,
      "step": 16798
    },
    {
      "epoch": 2.300602574637086,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005498148870913573,
      "loss": 4.7976,
      "step": 16799
    },
    {
      "epoch": 2.3007395234182417,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005497749417169047,
      "loss": 4.7916,
      "step": 16800
    },
    {
      "epoch": 2.3008764721993975,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005497349955155933,
      "loss": 4.7169,
      "step": 16801
    },
    {
      "epoch": 2.3010134209805533,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005496950484877544,
      "loss": 4.773,
      "step": 16802
    },
    {
      "epoch": 2.301150369761709,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005496551006337187,
      "loss": 4.7194,
      "step": 16803
    },
    {
      "epoch": 2.301287318542865,
      "grad_norm": 0.375,
      "learning_rate": 0.0005496151519538175,
      "loss": 4.7334,
      "step": 16804
    },
    {
      "epoch": 2.3014242673240206,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005495752024483819,
      "loss": 4.7314,
      "step": 16805
    },
    {
      "epoch": 2.301561216105177,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005495352521177427,
      "loss": 4.7443,
      "step": 16806
    },
    {
      "epoch": 2.3016981648863326,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005494953009622313,
      "loss": 4.7771,
      "step": 16807
    },
    {
      "epoch": 2.3018351136674884,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005494553489821785,
      "loss": 4.7163,
      "step": 16808
    },
    {
      "epoch": 2.301972062448644,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005494153961779157,
      "loss": 4.6911,
      "step": 16809
    },
    {
      "epoch": 2.3021090112298,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005493754425497736,
      "loss": 4.7564,
      "step": 16810
    },
    {
      "epoch": 2.3022459600109557,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005493354880980835,
      "loss": 4.7623,
      "step": 16811
    },
    {
      "epoch": 2.302382908792112,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005492955328231767,
      "loss": 4.7827,
      "step": 16812
    },
    {
      "epoch": 2.3025198575732677,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005492555767253842,
      "loss": 4.6919,
      "step": 16813
    },
    {
      "epoch": 2.3026568063544235,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005492156198050371,
      "loss": 4.823,
      "step": 16814
    },
    {
      "epoch": 2.3027937551355793,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0005491756620624663,
      "loss": 4.6772,
      "step": 16815
    },
    {
      "epoch": 2.302930703916735,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005491357034980034,
      "loss": 4.6396,
      "step": 16816
    },
    {
      "epoch": 2.303067652697891,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0005490957441119792,
      "loss": 4.7736,
      "step": 16817
    },
    {
      "epoch": 2.303204601479047,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005490557839047249,
      "loss": 4.6538,
      "step": 16818
    },
    {
      "epoch": 2.303341550260203,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0005490158228765717,
      "loss": 4.6819,
      "step": 16819
    },
    {
      "epoch": 2.3034784990413586,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000548975861027851,
      "loss": 4.7654,
      "step": 16820
    },
    {
      "epoch": 2.3036154478225144,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0005489358983588934,
      "loss": 4.728,
      "step": 16821
    },
    {
      "epoch": 2.30375239660367,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005488959348700307,
      "loss": 4.6869,
      "step": 16822
    },
    {
      "epoch": 2.303889345384826,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005488559705615938,
      "loss": 4.7433,
      "step": 16823
    },
    {
      "epoch": 2.3040262941659817,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005488160054339138,
      "loss": 4.7736,
      "step": 16824
    },
    {
      "epoch": 2.304163242947138,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005487760394873221,
      "loss": 4.7825,
      "step": 16825
    },
    {
      "epoch": 2.3043001917282937,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005487360727221498,
      "loss": 4.665,
      "step": 16826
    },
    {
      "epoch": 2.3044371405094495,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005486961051387282,
      "loss": 4.7598,
      "step": 16827
    },
    {
      "epoch": 2.3045740892906053,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005486561367373884,
      "loss": 4.7489,
      "step": 16828
    },
    {
      "epoch": 2.304711038071761,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005486161675184617,
      "loss": 4.7777,
      "step": 16829
    },
    {
      "epoch": 2.3048479868529173,
      "grad_norm": 0.375,
      "learning_rate": 0.0005485761974822792,
      "loss": 4.7446,
      "step": 16830
    },
    {
      "epoch": 2.304984935634073,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005485362266291723,
      "loss": 4.6945,
      "step": 16831
    },
    {
      "epoch": 2.305121884415229,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005484962549594724,
      "loss": 4.7511,
      "step": 16832
    },
    {
      "epoch": 2.3052588331963846,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005484562824735104,
      "loss": 4.7606,
      "step": 16833
    },
    {
      "epoch": 2.3053957819775404,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005484163091716178,
      "loss": 4.7733,
      "step": 16834
    },
    {
      "epoch": 2.305532730758696,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005483763350541259,
      "loss": 4.714,
      "step": 16835
    },
    {
      "epoch": 2.305669679539852,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005483363601213656,
      "loss": 4.6435,
      "step": 16836
    },
    {
      "epoch": 2.305806628321008,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005482963843736686,
      "loss": 4.7243,
      "step": 16837
    },
    {
      "epoch": 2.305943577102164,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005482564078113661,
      "loss": 4.7517,
      "step": 16838
    },
    {
      "epoch": 2.3060805258833197,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005482164304347893,
      "loss": 4.8184,
      "step": 16839
    },
    {
      "epoch": 2.3062174746644755,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005481764522442694,
      "loss": 4.7371,
      "step": 16840
    },
    {
      "epoch": 2.3063544234456312,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000548136473240138,
      "loss": 4.6905,
      "step": 16841
    },
    {
      "epoch": 2.306491372226787,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005480964934227262,
      "loss": 4.713,
      "step": 16842
    },
    {
      "epoch": 2.306628321007943,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005480565127923653,
      "loss": 4.6511,
      "step": 16843
    },
    {
      "epoch": 2.306765269789099,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005480165313493869,
      "loss": 4.7573,
      "step": 16844
    },
    {
      "epoch": 2.306902218570255,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000547976549094122,
      "loss": 4.7267,
      "step": 16845
    },
    {
      "epoch": 2.3070391673514106,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005479365660269022,
      "loss": 4.7344,
      "step": 16846
    },
    {
      "epoch": 2.3071761161325663,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005478965821480587,
      "loss": 4.6943,
      "step": 16847
    },
    {
      "epoch": 2.307313064913722,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005478565974579229,
      "loss": 4.7642,
      "step": 16848
    },
    {
      "epoch": 2.3074500136948783,
      "grad_norm": 0.34375,
      "learning_rate": 0.000547816611956826,
      "loss": 4.7147,
      "step": 16849
    },
    {
      "epoch": 2.307586962476034,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005477766256450998,
      "loss": 4.7997,
      "step": 16850
    },
    {
      "epoch": 2.30772391125719,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005477366385230752,
      "loss": 4.7303,
      "step": 16851
    },
    {
      "epoch": 2.3078608600383457,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005476966505910839,
      "loss": 4.782,
      "step": 16852
    },
    {
      "epoch": 2.3079978088195015,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005476566618494571,
      "loss": 4.7154,
      "step": 16853
    },
    {
      "epoch": 2.3081347576006572,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005476166722985263,
      "loss": 4.7848,
      "step": 16854
    },
    {
      "epoch": 2.308271706381813,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005475766819386229,
      "loss": 4.683,
      "step": 16855
    },
    {
      "epoch": 2.3084086551629692,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005475366907700783,
      "loss": 4.6235,
      "step": 16856
    },
    {
      "epoch": 2.308545603944125,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005474966987932239,
      "loss": 4.714,
      "step": 16857
    },
    {
      "epoch": 2.308682552725281,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000547456706008391,
      "loss": 4.787,
      "step": 16858
    },
    {
      "epoch": 2.3088195015064366,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005474167124159113,
      "loss": 4.6445,
      "step": 16859
    },
    {
      "epoch": 2.3089564502875923,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005473767180161161,
      "loss": 4.7394,
      "step": 16860
    },
    {
      "epoch": 2.309093399068748,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005473367228093367,
      "loss": 4.7431,
      "step": 16861
    },
    {
      "epoch": 2.3092303478499043,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005472967267959048,
      "loss": 4.7354,
      "step": 16862
    },
    {
      "epoch": 2.30936729663106,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005472567299761518,
      "loss": 4.7357,
      "step": 16863
    },
    {
      "epoch": 2.309504245412216,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005472167323504089,
      "loss": 4.737,
      "step": 16864
    },
    {
      "epoch": 2.3096411941933717,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005471767339190079,
      "loss": 4.7483,
      "step": 16865
    },
    {
      "epoch": 2.3097781429745274,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005471367346822801,
      "loss": 4.7397,
      "step": 16866
    },
    {
      "epoch": 2.309915091755683,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005470967346405571,
      "loss": 4.7617,
      "step": 16867
    },
    {
      "epoch": 2.3100520405368394,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005470567337941704,
      "loss": 4.7369,
      "step": 16868
    },
    {
      "epoch": 2.310188989317995,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005470167321434511,
      "loss": 4.6872,
      "step": 16869
    },
    {
      "epoch": 2.310325938099151,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005469767296887313,
      "loss": 4.7397,
      "step": 16870
    },
    {
      "epoch": 2.3104628868803068,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005469367264303422,
      "loss": 4.7039,
      "step": 16871
    },
    {
      "epoch": 2.3105998356614625,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005468967223686154,
      "loss": 4.7723,
      "step": 16872
    },
    {
      "epoch": 2.3107367844426183,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005468567175038822,
      "loss": 4.7353,
      "step": 16873
    },
    {
      "epoch": 2.310873733223774,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005468167118364744,
      "loss": 4.7245,
      "step": 16874
    },
    {
      "epoch": 2.3110106820049303,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005467767053667234,
      "loss": 4.5738,
      "step": 16875
    },
    {
      "epoch": 2.311147630786086,
      "grad_norm": 0.375,
      "learning_rate": 0.0005467366980949609,
      "loss": 4.7799,
      "step": 16876
    },
    {
      "epoch": 2.311284579567242,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005466966900215182,
      "loss": 4.7014,
      "step": 16877
    },
    {
      "epoch": 2.3114215283483976,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005466566811467269,
      "loss": 4.735,
      "step": 16878
    },
    {
      "epoch": 2.3115584771295534,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005466166714709189,
      "loss": 4.7868,
      "step": 16879
    },
    {
      "epoch": 2.311695425910709,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005465766609944253,
      "loss": 4.7031,
      "step": 16880
    },
    {
      "epoch": 2.3118323746918654,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000546536649717578,
      "loss": 4.7399,
      "step": 16881
    },
    {
      "epoch": 2.311969323473021,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005464966376407085,
      "loss": 4.7648,
      "step": 16882
    },
    {
      "epoch": 2.312106272254177,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005464566247641484,
      "loss": 4.7482,
      "step": 16883
    },
    {
      "epoch": 2.3122432210353328,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005464166110882293,
      "loss": 4.6789,
      "step": 16884
    },
    {
      "epoch": 2.3123801698164885,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005463765966132827,
      "loss": 4.7673,
      "step": 16885
    },
    {
      "epoch": 2.3125171185976443,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005463365813396402,
      "loss": 4.7351,
      "step": 16886
    },
    {
      "epoch": 2.3126540673788005,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005462965652676334,
      "loss": 4.772,
      "step": 16887
    },
    {
      "epoch": 2.3127910161599563,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005462565483975943,
      "loss": 4.6747,
      "step": 16888
    },
    {
      "epoch": 2.312927964941112,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005462165307298541,
      "loss": 4.6522,
      "step": 16889
    },
    {
      "epoch": 2.313064913722268,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005461765122647447,
      "loss": 4.8254,
      "step": 16890
    },
    {
      "epoch": 2.3132018625034236,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005461364930025974,
      "loss": 4.743,
      "step": 16891
    },
    {
      "epoch": 2.3133388112845794,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005460964729437443,
      "loss": 4.794,
      "step": 16892
    },
    {
      "epoch": 2.313475760065735,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005460564520885167,
      "loss": 4.778,
      "step": 16893
    },
    {
      "epoch": 2.3136127088468914,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005460164304372465,
      "loss": 4.6545,
      "step": 16894
    },
    {
      "epoch": 2.313749657628047,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005459764079902652,
      "loss": 4.786,
      "step": 16895
    },
    {
      "epoch": 2.313886606409203,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005459363847479044,
      "loss": 4.7572,
      "step": 16896
    },
    {
      "epoch": 2.3140235551903587,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005458963607104961,
      "loss": 4.7126,
      "step": 16897
    },
    {
      "epoch": 2.3141605039715145,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005458563358783717,
      "loss": 4.7409,
      "step": 16898
    },
    {
      "epoch": 2.3142974527526707,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005458163102518631,
      "loss": 4.7935,
      "step": 16899
    },
    {
      "epoch": 2.3144344015338265,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005457762838313018,
      "loss": 4.8018,
      "step": 16900
    },
    {
      "epoch": 2.3145713503149823,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005457362566170197,
      "loss": 4.7416,
      "step": 16901
    },
    {
      "epoch": 2.314708299096138,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005456962286093484,
      "loss": 4.8417,
      "step": 16902
    },
    {
      "epoch": 2.314845247877294,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005456561998086197,
      "loss": 4.702,
      "step": 16903
    },
    {
      "epoch": 2.3149821966584496,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005456161702151653,
      "loss": 4.6358,
      "step": 16904
    },
    {
      "epoch": 2.3151191454396054,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005455761398293167,
      "loss": 4.7471,
      "step": 16905
    },
    {
      "epoch": 2.3152560942207616,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000545536108651406,
      "loss": 4.682,
      "step": 16906
    },
    {
      "epoch": 2.3153930430019174,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005454960766817648,
      "loss": 4.6728,
      "step": 16907
    },
    {
      "epoch": 2.315529991783073,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005454560439207248,
      "loss": 4.7778,
      "step": 16908
    },
    {
      "epoch": 2.315666940564229,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005454160103686179,
      "loss": 4.7315,
      "step": 16909
    },
    {
      "epoch": 2.3158038893453847,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005453759760257759,
      "loss": 4.7313,
      "step": 16910
    },
    {
      "epoch": 2.3159408381265405,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005453359408925302,
      "loss": 4.6978,
      "step": 16911
    },
    {
      "epoch": 2.3160777869076963,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005452959049692131,
      "loss": 4.7852,
      "step": 16912
    },
    {
      "epoch": 2.3162147356888525,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005452558682561559,
      "loss": 4.7173,
      "step": 16913
    },
    {
      "epoch": 2.3163516844700083,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005452158307536907,
      "loss": 4.7834,
      "step": 16914
    },
    {
      "epoch": 2.316488633251164,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005451757924621492,
      "loss": 4.7841,
      "step": 16915
    },
    {
      "epoch": 2.31662558203232,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005451357533818633,
      "loss": 4.6843,
      "step": 16916
    },
    {
      "epoch": 2.3167625308134756,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005450957135131647,
      "loss": 4.7774,
      "step": 16917
    },
    {
      "epoch": 2.316899479594632,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005450556728563853,
      "loss": 4.6888,
      "step": 16918
    },
    {
      "epoch": 2.3170364283757876,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005450156314118569,
      "loss": 4.7243,
      "step": 16919
    },
    {
      "epoch": 2.3171733771569434,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005449755891799112,
      "loss": 4.7895,
      "step": 16920
    },
    {
      "epoch": 2.317310325938099,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005449355461608804,
      "loss": 4.7196,
      "step": 16921
    },
    {
      "epoch": 2.317447274719255,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005448955023550958,
      "loss": 4.6953,
      "step": 16922
    },
    {
      "epoch": 2.3175842235004107,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005448554577628897,
      "loss": 4.6891,
      "step": 16923
    },
    {
      "epoch": 2.3177211722815665,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005448154123845938,
      "loss": 4.6915,
      "step": 16924
    },
    {
      "epoch": 2.3178581210627227,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005447753662205401,
      "loss": 4.673,
      "step": 16925
    },
    {
      "epoch": 2.3179950698438785,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005447353192710602,
      "loss": 4.7319,
      "step": 16926
    },
    {
      "epoch": 2.3181320186250343,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005446952715364861,
      "loss": 4.7942,
      "step": 16927
    },
    {
      "epoch": 2.31826896740619,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005446552230171499,
      "loss": 4.6773,
      "step": 16928
    },
    {
      "epoch": 2.318405916187346,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005446151737133832,
      "loss": 4.7599,
      "step": 16929
    },
    {
      "epoch": 2.3185428649685016,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005445751236255182,
      "loss": 4.8497,
      "step": 16930
    },
    {
      "epoch": 2.318679813749658,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005445350727538863,
      "loss": 4.7284,
      "step": 16931
    },
    {
      "epoch": 2.3188167625308136,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00054449502109882,
      "loss": 4.7015,
      "step": 16932
    },
    {
      "epoch": 2.3189537113119694,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005444549686606508,
      "loss": 4.6637,
      "step": 16933
    },
    {
      "epoch": 2.319090660093125,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005444149154397107,
      "loss": 4.6979,
      "step": 16934
    },
    {
      "epoch": 2.319227608874281,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005443748614363318,
      "loss": 4.7255,
      "step": 16935
    },
    {
      "epoch": 2.3193645576554367,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000544334806650846,
      "loss": 4.5759,
      "step": 16936
    },
    {
      "epoch": 2.319501506436593,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005442947510835852,
      "loss": 4.7332,
      "step": 16937
    },
    {
      "epoch": 2.3196384552177487,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000544254694734881,
      "loss": 4.7047,
      "step": 16938
    },
    {
      "epoch": 2.3197754039989045,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005442146376050661,
      "loss": 4.7465,
      "step": 16939
    },
    {
      "epoch": 2.3199123527800602,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005441745796944717,
      "loss": 4.7345,
      "step": 16940
    },
    {
      "epoch": 2.320049301561216,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005441345210034303,
      "loss": 4.7853,
      "step": 16941
    },
    {
      "epoch": 2.320186250342372,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005440944615322737,
      "loss": 4.6891,
      "step": 16942
    },
    {
      "epoch": 2.3203231991235276,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005440544012813338,
      "loss": 4.6886,
      "step": 16943
    },
    {
      "epoch": 2.320460147904684,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005440143402509428,
      "loss": 4.7293,
      "step": 16944
    },
    {
      "epoch": 2.3205970966858396,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005439742784414325,
      "loss": 4.7042,
      "step": 16945
    },
    {
      "epoch": 2.3207340454669954,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005439342158531349,
      "loss": 4.7659,
      "step": 16946
    },
    {
      "epoch": 2.320870994248151,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005438941524863821,
      "loss": 4.6681,
      "step": 16947
    },
    {
      "epoch": 2.321007943029307,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005438540883415062,
      "loss": 4.6732,
      "step": 16948
    },
    {
      "epoch": 2.321144891810463,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005438140234188389,
      "loss": 4.7196,
      "step": 16949
    },
    {
      "epoch": 2.321281840591619,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005437739577187126,
      "loss": 4.7633,
      "step": 16950
    },
    {
      "epoch": 2.3214187893727747,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005437338912414591,
      "loss": 4.7031,
      "step": 16951
    },
    {
      "epoch": 2.3215557381539305,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005436938239874104,
      "loss": 4.7562,
      "step": 16952
    },
    {
      "epoch": 2.3216926869350862,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005436537559568988,
      "loss": 4.644,
      "step": 16953
    },
    {
      "epoch": 2.321829635716242,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005436136871502562,
      "loss": 4.6336,
      "step": 16954
    },
    {
      "epoch": 2.321966584497398,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005435736175678147,
      "loss": 4.7532,
      "step": 16955
    },
    {
      "epoch": 2.322103533278554,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005435335472099062,
      "loss": 4.695,
      "step": 16956
    },
    {
      "epoch": 2.32224048205971,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000543493476076863,
      "loss": 4.7036,
      "step": 16957
    },
    {
      "epoch": 2.3223774308408656,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005434534041690171,
      "loss": 4.7243,
      "step": 16958
    },
    {
      "epoch": 2.3225143796220213,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005434133314867006,
      "loss": 4.6732,
      "step": 16959
    },
    {
      "epoch": 2.322651328403177,
      "grad_norm": 0.375,
      "learning_rate": 0.0005433732580302454,
      "loss": 4.6768,
      "step": 16960
    },
    {
      "epoch": 2.322788277184333,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005433331837999837,
      "loss": 4.783,
      "step": 16961
    },
    {
      "epoch": 2.3229252259654887,
      "grad_norm": 0.390625,
      "learning_rate": 0.000543293108796248,
      "loss": 4.7293,
      "step": 16962
    },
    {
      "epoch": 2.323062174746645,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005432530330193698,
      "loss": 4.7646,
      "step": 16963
    },
    {
      "epoch": 2.3231991235278007,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0005432129564696817,
      "loss": 4.8405,
      "step": 16964
    },
    {
      "epoch": 2.3233360723089564,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005431728791475155,
      "loss": 4.7421,
      "step": 16965
    },
    {
      "epoch": 2.3234730210901122,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005431328010532035,
      "loss": 4.7562,
      "step": 16966
    },
    {
      "epoch": 2.323609969871268,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005430927221870777,
      "loss": 4.7494,
      "step": 16967
    },
    {
      "epoch": 2.323746918652424,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005430526425494704,
      "loss": 4.6303,
      "step": 16968
    },
    {
      "epoch": 2.32388386743358,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005430125621407136,
      "loss": 4.6557,
      "step": 16969
    },
    {
      "epoch": 2.3240208162147358,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005429724809611395,
      "loss": 4.6885,
      "step": 16970
    },
    {
      "epoch": 2.3241577649958916,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005429323990110805,
      "loss": 4.8017,
      "step": 16971
    },
    {
      "epoch": 2.3242947137770473,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005428923162908683,
      "loss": 4.7064,
      "step": 16972
    },
    {
      "epoch": 2.324431662558203,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005428522328008356,
      "loss": 4.7066,
      "step": 16973
    },
    {
      "epoch": 2.324568611339359,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005428121485413142,
      "loss": 4.726,
      "step": 16974
    },
    {
      "epoch": 2.324705560120515,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005427720635126364,
      "loss": 4.7958,
      "step": 16975
    },
    {
      "epoch": 2.324842508901671,
      "grad_norm": 0.375,
      "learning_rate": 0.0005427319777151344,
      "loss": 4.6762,
      "step": 16976
    },
    {
      "epoch": 2.3249794576828267,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005426918911491405,
      "loss": 4.7718,
      "step": 16977
    },
    {
      "epoch": 2.3251164064639824,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005426518038149867,
      "loss": 4.7809,
      "step": 16978
    },
    {
      "epoch": 2.325253355245138,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005426117157130053,
      "loss": 4.684,
      "step": 16979
    },
    {
      "epoch": 2.325390304026294,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005425716268435286,
      "loss": 4.6579,
      "step": 16980
    },
    {
      "epoch": 2.32552725280745,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005425315372068889,
      "loss": 4.6698,
      "step": 16981
    },
    {
      "epoch": 2.325664201588606,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000542491446803418,
      "loss": 4.7185,
      "step": 16982
    },
    {
      "epoch": 2.3258011503697618,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005424513556334487,
      "loss": 4.5918,
      "step": 16983
    },
    {
      "epoch": 2.3259380991509175,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005424112636973129,
      "loss": 4.7349,
      "step": 16984
    },
    {
      "epoch": 2.3260750479320733,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005423711709953429,
      "loss": 4.7624,
      "step": 16985
    },
    {
      "epoch": 2.326211996713229,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005423310775278711,
      "loss": 4.6547,
      "step": 16986
    },
    {
      "epoch": 2.3263489454943853,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005422909832952296,
      "loss": 4.815,
      "step": 16987
    },
    {
      "epoch": 2.326485894275541,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005422508882977507,
      "loss": 4.6553,
      "step": 16988
    },
    {
      "epoch": 2.326622843056697,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005422107925357667,
      "loss": 4.7282,
      "step": 16989
    },
    {
      "epoch": 2.3267597918378526,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005421706960096099,
      "loss": 4.7479,
      "step": 16990
    },
    {
      "epoch": 2.3268967406190084,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005421305987196127,
      "loss": 4.7035,
      "step": 16991
    },
    {
      "epoch": 2.327033689400164,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005420905006661072,
      "loss": 4.7383,
      "step": 16992
    },
    {
      "epoch": 2.32717063818132,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005420504018494258,
      "loss": 4.7505,
      "step": 16993
    },
    {
      "epoch": 2.327307586962476,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005420103022699005,
      "loss": 4.6935,
      "step": 16994
    },
    {
      "epoch": 2.327444535743632,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005419702019278642,
      "loss": 4.6345,
      "step": 16995
    },
    {
      "epoch": 2.3275814845247877,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000541930100823649,
      "loss": 4.6815,
      "step": 16996
    },
    {
      "epoch": 2.3277184333059435,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005418899989575869,
      "loss": 4.7289,
      "step": 16997
    },
    {
      "epoch": 2.3278553820870993,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005418498963300106,
      "loss": 4.7719,
      "step": 16998
    },
    {
      "epoch": 2.3279923308682555,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005418097929412522,
      "loss": 4.6716,
      "step": 16999
    },
    {
      "epoch": 2.3281292796494113,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005417696887916444,
      "loss": 4.6825,
      "step": 17000
    },
    {
      "epoch": 2.328266228430567,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005417295838815191,
      "loss": 4.6722,
      "step": 17001
    },
    {
      "epoch": 2.328403177211723,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000541689478211209,
      "loss": 4.7069,
      "step": 17002
    },
    {
      "epoch": 2.3285401259928786,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005416493717810464,
      "loss": 4.7539,
      "step": 17003
    },
    {
      "epoch": 2.3286770747740344,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005416092645913634,
      "loss": 4.7536,
      "step": 17004
    },
    {
      "epoch": 2.32881402355519,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005415691566424929,
      "loss": 4.7056,
      "step": 17005
    },
    {
      "epoch": 2.3289509723363464,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005415290479347667,
      "loss": 4.7256,
      "step": 17006
    },
    {
      "epoch": 2.329087921117502,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005414889384685178,
      "loss": 4.7581,
      "step": 17007
    },
    {
      "epoch": 2.329224869898658,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005414488282440779,
      "loss": 4.7176,
      "step": 17008
    },
    {
      "epoch": 2.3293618186798137,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005414087172617799,
      "loss": 4.8143,
      "step": 17009
    },
    {
      "epoch": 2.3294987674609695,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005413686055219561,
      "loss": 4.7761,
      "step": 17010
    },
    {
      "epoch": 2.3296357162421253,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005413284930249389,
      "loss": 4.7375,
      "step": 17011
    },
    {
      "epoch": 2.329772665023281,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005412883797710607,
      "loss": 4.6878,
      "step": 17012
    },
    {
      "epoch": 2.3299096138044373,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000541248265760654,
      "loss": 4.7138,
      "step": 17013
    },
    {
      "epoch": 2.330046562585593,
      "grad_norm": 0.375,
      "learning_rate": 0.0005412081509940512,
      "loss": 4.7028,
      "step": 17014
    },
    {
      "epoch": 2.330183511366749,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005411680354715846,
      "loss": 4.8027,
      "step": 17015
    },
    {
      "epoch": 2.3303204601479046,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005411279191935867,
      "loss": 4.7072,
      "step": 17016
    },
    {
      "epoch": 2.3304574089290604,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005410878021603902,
      "loss": 4.7852,
      "step": 17017
    },
    {
      "epoch": 2.3305943577102166,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005410476843723274,
      "loss": 4.7732,
      "step": 17018
    },
    {
      "epoch": 2.3307313064913724,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005410075658297307,
      "loss": 4.7207,
      "step": 17019
    },
    {
      "epoch": 2.330868255272528,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005409674465329324,
      "loss": 4.6852,
      "step": 17020
    },
    {
      "epoch": 2.331005204053684,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005409273264822655,
      "loss": 4.7742,
      "step": 17021
    },
    {
      "epoch": 2.3311421528348397,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005408872056780619,
      "loss": 4.7244,
      "step": 17022
    },
    {
      "epoch": 2.3312791016159955,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005408470841206546,
      "loss": 4.7221,
      "step": 17023
    },
    {
      "epoch": 2.3314160503971513,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005408069618103758,
      "loss": 4.737,
      "step": 17024
    },
    {
      "epoch": 2.3315529991783075,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000540766838747558,
      "loss": 4.7927,
      "step": 17025
    },
    {
      "epoch": 2.3316899479594633,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005407267149325339,
      "loss": 4.7515,
      "step": 17026
    },
    {
      "epoch": 2.331826896740619,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005406865903656358,
      "loss": 4.6959,
      "step": 17027
    },
    {
      "epoch": 2.331963845521775,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005406464650471964,
      "loss": 4.6847,
      "step": 17028
    },
    {
      "epoch": 2.3321007943029306,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005406063389775481,
      "loss": 4.6761,
      "step": 17029
    },
    {
      "epoch": 2.3322377430840864,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005405662121570233,
      "loss": 4.6824,
      "step": 17030
    },
    {
      "epoch": 2.332374691865242,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005405260845859549,
      "loss": 4.7592,
      "step": 17031
    },
    {
      "epoch": 2.3325116406463984,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005404859562646752,
      "loss": 4.7345,
      "step": 17032
    },
    {
      "epoch": 2.332648589427554,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005404458271935171,
      "loss": 4.7528,
      "step": 17033
    },
    {
      "epoch": 2.33278553820871,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005404056973728125,
      "loss": 4.7366,
      "step": 17034
    },
    {
      "epoch": 2.3329224869898657,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005403655668028946,
      "loss": 4.6964,
      "step": 17035
    },
    {
      "epoch": 2.3330594357710215,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005403254354840955,
      "loss": 4.6974,
      "step": 17036
    },
    {
      "epoch": 2.3331963845521777,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000540285303416748,
      "loss": 4.678,
      "step": 17037
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005402451706011849,
      "loss": 4.7506,
      "step": 17038
    },
    {
      "epoch": 2.3334702821144893,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005402050370377385,
      "loss": 4.8133,
      "step": 17039
    },
    {
      "epoch": 2.333607230895645,
      "grad_norm": 0.375,
      "learning_rate": 0.0005401649027267414,
      "loss": 4.7103,
      "step": 17040
    },
    {
      "epoch": 2.333744179676801,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005401247676685262,
      "loss": 4.7921,
      "step": 17041
    },
    {
      "epoch": 2.3338811284579566,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005400846318634257,
      "loss": 4.72,
      "step": 17042
    },
    {
      "epoch": 2.3340180772391124,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005400444953117723,
      "loss": 4.6296,
      "step": 17043
    },
    {
      "epoch": 2.3341550260202686,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005400043580138989,
      "loss": 4.6833,
      "step": 17044
    },
    {
      "epoch": 2.3342919748014244,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005399642199701377,
      "loss": 4.6716,
      "step": 17045
    },
    {
      "epoch": 2.33442892358258,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005399240811808216,
      "loss": 4.7417,
      "step": 17046
    },
    {
      "epoch": 2.334565872363736,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005398839416462833,
      "loss": 4.7826,
      "step": 17047
    },
    {
      "epoch": 2.3347028211448917,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005398438013668553,
      "loss": 4.7191,
      "step": 17048
    },
    {
      "epoch": 2.3348397699260475,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005398036603428704,
      "loss": 4.688,
      "step": 17049
    },
    {
      "epoch": 2.3349767187072037,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000539763518574661,
      "loss": 4.7576,
      "step": 17050
    },
    {
      "epoch": 2.3351136674883595,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005397233760625601,
      "loss": 4.7509,
      "step": 17051
    },
    {
      "epoch": 2.3352506162695152,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005396832328069002,
      "loss": 4.7278,
      "step": 17052
    },
    {
      "epoch": 2.335387565050671,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005396430888080139,
      "loss": 4.6987,
      "step": 17053
    },
    {
      "epoch": 2.335524513831827,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000539602944066234,
      "loss": 4.7061,
      "step": 17054
    },
    {
      "epoch": 2.3356614626129826,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000539562798581893,
      "loss": 4.7689,
      "step": 17055
    },
    {
      "epoch": 2.335798411394139,
      "grad_norm": 0.3828125,
      "learning_rate": 0.000539522652355324,
      "loss": 4.6924,
      "step": 17056
    },
    {
      "epoch": 2.3359353601752946,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005394825053868592,
      "loss": 4.7505,
      "step": 17057
    },
    {
      "epoch": 2.3360723089564503,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005394423576768317,
      "loss": 4.7421,
      "step": 17058
    },
    {
      "epoch": 2.336209257737606,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005394022092255741,
      "loss": 4.6763,
      "step": 17059
    },
    {
      "epoch": 2.336346206518762,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005393620600334191,
      "loss": 4.7313,
      "step": 17060
    },
    {
      "epoch": 2.3364831552999177,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005393219101006993,
      "loss": 4.6923,
      "step": 17061
    },
    {
      "epoch": 2.3366201040810735,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005392817594277476,
      "loss": 4.7118,
      "step": 17062
    },
    {
      "epoch": 2.3367570528622297,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005392416080148968,
      "loss": 4.7675,
      "step": 17063
    },
    {
      "epoch": 2.3368940016433855,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005392014558624793,
      "loss": 4.772,
      "step": 17064
    },
    {
      "epoch": 2.3370309504245412,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005391613029708281,
      "loss": 4.703,
      "step": 17065
    },
    {
      "epoch": 2.337167899205697,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005391211493402761,
      "loss": 4.7743,
      "step": 17066
    },
    {
      "epoch": 2.337304847986853,
      "grad_norm": 0.375,
      "learning_rate": 0.0005390809949711559,
      "loss": 4.6792,
      "step": 17067
    },
    {
      "epoch": 2.337441796768009,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005390408398638001,
      "loss": 4.784,
      "step": 17068
    },
    {
      "epoch": 2.337578745549165,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005390006840185418,
      "loss": 4.7411,
      "step": 17069
    },
    {
      "epoch": 2.3377156943303206,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005389605274357136,
      "loss": 4.7582,
      "step": 17070
    },
    {
      "epoch": 2.3378526431114763,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005389203701156483,
      "loss": 4.7144,
      "step": 17071
    },
    {
      "epoch": 2.337989591892632,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005388802120586786,
      "loss": 4.6829,
      "step": 17072
    },
    {
      "epoch": 2.338126540673788,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005388400532651375,
      "loss": 4.7821,
      "step": 17073
    },
    {
      "epoch": 2.3382634894549437,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005387998937353577,
      "loss": 4.662,
      "step": 17074
    },
    {
      "epoch": 2.3384004382361,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005387597334696719,
      "loss": 4.7184,
      "step": 17075
    },
    {
      "epoch": 2.3385373870172557,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005387195724684133,
      "loss": 4.7978,
      "step": 17076
    },
    {
      "epoch": 2.3386743357984114,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005386794107319143,
      "loss": 4.8273,
      "step": 17077
    },
    {
      "epoch": 2.338811284579567,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005386392482605079,
      "loss": 4.7233,
      "step": 17078
    },
    {
      "epoch": 2.338948233360723,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005385990850545269,
      "loss": 4.7056,
      "step": 17079
    },
    {
      "epoch": 2.3390851821418788,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005385589211143043,
      "loss": 4.7635,
      "step": 17080
    },
    {
      "epoch": 2.3392221309230345,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005385187564401726,
      "loss": 4.6604,
      "step": 17081
    },
    {
      "epoch": 2.3393590797041908,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005384785910324649,
      "loss": 4.6534,
      "step": 17082
    },
    {
      "epoch": 2.3394960284853465,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005384384248915142,
      "loss": 4.7883,
      "step": 17083
    },
    {
      "epoch": 2.3396329772665023,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005383982580176532,
      "loss": 4.738,
      "step": 17084
    },
    {
      "epoch": 2.339769926047658,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005383580904112147,
      "loss": 4.7069,
      "step": 17085
    },
    {
      "epoch": 2.339906874828814,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005383179220725316,
      "loss": 4.7537,
      "step": 17086
    },
    {
      "epoch": 2.34004382360997,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005382777530019369,
      "loss": 4.7352,
      "step": 17087
    },
    {
      "epoch": 2.340180772391126,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005382375831997634,
      "loss": 4.6722,
      "step": 17088
    },
    {
      "epoch": 2.3403177211722817,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005381974126663442,
      "loss": 4.7118,
      "step": 17089
    },
    {
      "epoch": 2.3404546699534374,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005381572414020118,
      "loss": 4.6958,
      "step": 17090
    },
    {
      "epoch": 2.340591618734593,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005381170694070993,
      "loss": 4.7798,
      "step": 17091
    },
    {
      "epoch": 2.340728567515749,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005380768966819397,
      "loss": 4.7315,
      "step": 17092
    },
    {
      "epoch": 2.3408655162969048,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005380367232268659,
      "loss": 4.7741,
      "step": 17093
    },
    {
      "epoch": 2.341002465078061,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005379965490422109,
      "loss": 4.731,
      "step": 17094
    },
    {
      "epoch": 2.3411394138592168,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005379563741283074,
      "loss": 4.807,
      "step": 17095
    },
    {
      "epoch": 2.3412763626403725,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005379161984854885,
      "loss": 4.7205,
      "step": 17096
    },
    {
      "epoch": 2.3414133114215283,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005378760221140871,
      "loss": 4.7158,
      "step": 17097
    },
    {
      "epoch": 2.341550260202684,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005378358450144362,
      "loss": 4.7328,
      "step": 17098
    },
    {
      "epoch": 2.34168720898384,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005377956671868687,
      "loss": 4.6847,
      "step": 17099
    },
    {
      "epoch": 2.341824157764996,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005377554886317177,
      "loss": 4.7728,
      "step": 17100
    },
    {
      "epoch": 2.341961106546152,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005377153093493158,
      "loss": 4.6676,
      "step": 17101
    },
    {
      "epoch": 2.3420980553273076,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005376751293399963,
      "loss": 4.7451,
      "step": 17102
    },
    {
      "epoch": 2.3422350041084634,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005376349486040924,
      "loss": 4.6739,
      "step": 17103
    },
    {
      "epoch": 2.342371952889619,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005375947671419364,
      "loss": 4.6798,
      "step": 17104
    },
    {
      "epoch": 2.342508901670775,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000537554584953862,
      "loss": 4.6541,
      "step": 17105
    },
    {
      "epoch": 2.342645850451931,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005375144020402017,
      "loss": 4.7433,
      "step": 17106
    },
    {
      "epoch": 2.342782799233087,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005374742184012888,
      "loss": 4.7782,
      "step": 17107
    },
    {
      "epoch": 2.3429197480142427,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005374340340374562,
      "loss": 4.7256,
      "step": 17108
    },
    {
      "epoch": 2.3430566967953985,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000537393848949037,
      "loss": 4.7381,
      "step": 17109
    },
    {
      "epoch": 2.3431936455765543,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005373536631363639,
      "loss": 4.7288,
      "step": 17110
    },
    {
      "epoch": 2.34333059435771,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005373134765997703,
      "loss": 4.6687,
      "step": 17111
    },
    {
      "epoch": 2.343467543138866,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005372732893395892,
      "loss": 4.7969,
      "step": 17112
    },
    {
      "epoch": 2.343604491920022,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005372331013561534,
      "loss": 4.751,
      "step": 17113
    },
    {
      "epoch": 2.343741440701178,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005371929126497963,
      "loss": 4.6671,
      "step": 17114
    },
    {
      "epoch": 2.3438783894823336,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005371527232208506,
      "loss": 4.7747,
      "step": 17115
    },
    {
      "epoch": 2.3440153382634894,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005371125330696497,
      "loss": 4.6839,
      "step": 17116
    },
    {
      "epoch": 2.344152287044645,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005370723421965262,
      "loss": 4.7345,
      "step": 17117
    },
    {
      "epoch": 2.3442892358258014,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005370321506018137,
      "loss": 4.7519,
      "step": 17118
    },
    {
      "epoch": 2.344426184606957,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005369919582858449,
      "loss": 4.7477,
      "step": 17119
    },
    {
      "epoch": 2.344563133388113,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000536951765248953,
      "loss": 4.6633,
      "step": 17120
    },
    {
      "epoch": 2.3447000821692687,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005369115714914712,
      "loss": 4.7588,
      "step": 17121
    },
    {
      "epoch": 2.3448370309504245,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005368713770137324,
      "loss": 4.7013,
      "step": 17122
    },
    {
      "epoch": 2.3449739797315803,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005368311818160699,
      "loss": 4.7793,
      "step": 17123
    },
    {
      "epoch": 2.345110928512736,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005367909858988166,
      "loss": 4.7372,
      "step": 17124
    },
    {
      "epoch": 2.3452478772938923,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005367507892623058,
      "loss": 4.8012,
      "step": 17125
    },
    {
      "epoch": 2.345384826075048,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005367105919068704,
      "loss": 4.705,
      "step": 17126
    },
    {
      "epoch": 2.345521774856204,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005366703938328439,
      "loss": 4.6189,
      "step": 17127
    },
    {
      "epoch": 2.3456587236373596,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000536630195040559,
      "loss": 4.6705,
      "step": 17128
    },
    {
      "epoch": 2.3457956724185154,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005365899955303491,
      "loss": 4.7609,
      "step": 17129
    },
    {
      "epoch": 2.345932621199671,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005365497953025472,
      "loss": 4.7045,
      "step": 17130
    },
    {
      "epoch": 2.346069569980827,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005365095943574866,
      "loss": 4.721,
      "step": 17131
    },
    {
      "epoch": 2.346206518761983,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005364693926955004,
      "loss": 4.7002,
      "step": 17132
    },
    {
      "epoch": 2.346343467543139,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005364291903169217,
      "loss": 4.6582,
      "step": 17133
    },
    {
      "epoch": 2.3464804163242947,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005363889872220838,
      "loss": 4.6728,
      "step": 17134
    },
    {
      "epoch": 2.3466173651054505,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005363487834113195,
      "loss": 4.7591,
      "step": 17135
    },
    {
      "epoch": 2.3467543138866063,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005363085788849625,
      "loss": 4.7935,
      "step": 17136
    },
    {
      "epoch": 2.3468912626677625,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005362683736433457,
      "loss": 4.7751,
      "step": 17137
    },
    {
      "epoch": 2.3470282114489183,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005362281676868023,
      "loss": 4.7228,
      "step": 17138
    },
    {
      "epoch": 2.347165160230074,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005361879610156655,
      "loss": 4.7183,
      "step": 17139
    },
    {
      "epoch": 2.34730210901123,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005361477536302686,
      "loss": 4.7431,
      "step": 17140
    },
    {
      "epoch": 2.3474390577923856,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005361075455309447,
      "loss": 4.692,
      "step": 17141
    },
    {
      "epoch": 2.3475760065735414,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000536067336718027,
      "loss": 4.7916,
      "step": 17142
    },
    {
      "epoch": 2.347712955354697,
      "grad_norm": 0.328125,
      "learning_rate": 0.000536027127191849,
      "loss": 4.6759,
      "step": 17143
    },
    {
      "epoch": 2.3478499041358534,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005359869169527435,
      "loss": 4.6854,
      "step": 17144
    },
    {
      "epoch": 2.347986852917009,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005359467060010439,
      "loss": 4.7369,
      "step": 17145
    },
    {
      "epoch": 2.348123801698165,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005359064943370835,
      "loss": 4.7767,
      "step": 17146
    },
    {
      "epoch": 2.3482607504793207,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005358662819611956,
      "loss": 4.6812,
      "step": 17147
    },
    {
      "epoch": 2.3483976992604765,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005358260688737133,
      "loss": 4.6726,
      "step": 17148
    },
    {
      "epoch": 2.3485346480416323,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005357858550749698,
      "loss": 4.6994,
      "step": 17149
    },
    {
      "epoch": 2.348671596822788,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005357456405652986,
      "loss": 4.7224,
      "step": 17150
    },
    {
      "epoch": 2.3488085456039443,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005357054253450329,
      "loss": 4.7542,
      "step": 17151
    },
    {
      "epoch": 2.3489454943851,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005356652094145059,
      "loss": 4.6842,
      "step": 17152
    },
    {
      "epoch": 2.349082443166256,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005356249927740507,
      "loss": 4.6315,
      "step": 17153
    },
    {
      "epoch": 2.3492193919474116,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0005355847754240011,
      "loss": 4.7011,
      "step": 17154
    },
    {
      "epoch": 2.3493563407285674,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005355445573646899,
      "loss": 4.6811,
      "step": 17155
    },
    {
      "epoch": 2.3494932895097236,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005355043385964506,
      "loss": 4.6671,
      "step": 17156
    },
    {
      "epoch": 2.3496302382908794,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005354641191196164,
      "loss": 4.6996,
      "step": 17157
    },
    {
      "epoch": 2.349767187072035,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005354238989345207,
      "loss": 4.7265,
      "step": 17158
    },
    {
      "epoch": 2.349904135853191,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005353836780414968,
      "loss": 4.7931,
      "step": 17159
    },
    {
      "epoch": 2.3500410846343467,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000535343456440878,
      "loss": 4.733,
      "step": 17160
    },
    {
      "epoch": 2.3501780334155025,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005353032341329977,
      "loss": 4.6651,
      "step": 17161
    },
    {
      "epoch": 2.3503149821966582,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005352630111181892,
      "loss": 4.7049,
      "step": 17162
    },
    {
      "epoch": 2.3504519309778145,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005352227873967857,
      "loss": 4.7994,
      "step": 17163
    },
    {
      "epoch": 2.3505888797589702,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005351825629691207,
      "loss": 4.7506,
      "step": 17164
    },
    {
      "epoch": 2.350725828540126,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005351423378355275,
      "loss": 4.7343,
      "step": 17165
    },
    {
      "epoch": 2.350862777321282,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005351021119963395,
      "loss": 4.7626,
      "step": 17166
    },
    {
      "epoch": 2.3509997261024376,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005350618854518898,
      "loss": 4.8058,
      "step": 17167
    },
    {
      "epoch": 2.3511366748835933,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005350216582025123,
      "loss": 4.6784,
      "step": 17168
    },
    {
      "epoch": 2.3512736236647496,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005349814302485399,
      "loss": 4.7135,
      "step": 17169
    },
    {
      "epoch": 2.3514105724459053,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005349412015903062,
      "loss": 4.7879,
      "step": 17170
    },
    {
      "epoch": 2.351547521227061,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005349009722281445,
      "loss": 4.7266,
      "step": 17171
    },
    {
      "epoch": 2.351684470008217,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005348607421623884,
      "loss": 4.6676,
      "step": 17172
    },
    {
      "epoch": 2.3518214187893727,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005348205113933708,
      "loss": 4.675,
      "step": 17173
    },
    {
      "epoch": 2.3519583675705285,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005347802799214254,
      "loss": 4.6833,
      "step": 17174
    },
    {
      "epoch": 2.3520953163516847,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005347400477468858,
      "loss": 4.7478,
      "step": 17175
    },
    {
      "epoch": 2.3522322651328404,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005346998148700852,
      "loss": 4.7714,
      "step": 17176
    },
    {
      "epoch": 2.3523692139139962,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000534659581291357,
      "loss": 4.767,
      "step": 17177
    },
    {
      "epoch": 2.352506162695152,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005346193470110347,
      "loss": 4.7675,
      "step": 17178
    },
    {
      "epoch": 2.352643111476308,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005345791120294518,
      "loss": 4.8145,
      "step": 17179
    },
    {
      "epoch": 2.3527800602574636,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005345388763469415,
      "loss": 4.7373,
      "step": 17180
    },
    {
      "epoch": 2.3529170090386193,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005344986399638375,
      "loss": 4.7328,
      "step": 17181
    },
    {
      "epoch": 2.3530539578197756,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000534458402880473,
      "loss": 4.7377,
      "step": 17182
    },
    {
      "epoch": 2.3531909066009313,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005344181650971817,
      "loss": 4.759,
      "step": 17183
    },
    {
      "epoch": 2.353327855382087,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005343779266142968,
      "loss": 4.7013,
      "step": 17184
    },
    {
      "epoch": 2.353464804163243,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005343376874321521,
      "loss": 4.6322,
      "step": 17185
    },
    {
      "epoch": 2.3536017529443987,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005342974475510809,
      "loss": 4.7733,
      "step": 17186
    },
    {
      "epoch": 2.353738701725555,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005342572069714165,
      "loss": 4.7911,
      "step": 17187
    },
    {
      "epoch": 2.3538756505067107,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005342169656934927,
      "loss": 4.7074,
      "step": 17188
    },
    {
      "epoch": 2.3540125992878664,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005341767237176427,
      "loss": 4.75,
      "step": 17189
    },
    {
      "epoch": 2.354149548069022,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005341364810442003,
      "loss": 4.735,
      "step": 17190
    },
    {
      "epoch": 2.354286496850178,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005340962376734987,
      "loss": 4.6811,
      "step": 17191
    },
    {
      "epoch": 2.3544234456313338,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005340559936058716,
      "loss": 4.7743,
      "step": 17192
    },
    {
      "epoch": 2.3545603944124895,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005340157488416524,
      "loss": 4.7191,
      "step": 17193
    },
    {
      "epoch": 2.3546973431936458,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005339755033811746,
      "loss": 4.6196,
      "step": 17194
    },
    {
      "epoch": 2.3548342919748015,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000533935257224772,
      "loss": 4.7326,
      "step": 17195
    },
    {
      "epoch": 2.3549712407559573,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005338950103727778,
      "loss": 4.6805,
      "step": 17196
    },
    {
      "epoch": 2.355108189537113,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005338547628255257,
      "loss": 4.739,
      "step": 17197
    },
    {
      "epoch": 2.355245138318269,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000533814514583349,
      "loss": 4.725,
      "step": 17198
    },
    {
      "epoch": 2.3553820870994246,
      "grad_norm": 0.375,
      "learning_rate": 0.0005337742656465818,
      "loss": 4.6927,
      "step": 17199
    },
    {
      "epoch": 2.3555190358805804,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005337340160155571,
      "loss": 4.6755,
      "step": 17200
    },
    {
      "epoch": 2.3556559846617366,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005336937656906086,
      "loss": 4.7262,
      "step": 17201
    },
    {
      "epoch": 2.3557929334428924,
      "grad_norm": 0.375,
      "learning_rate": 0.00053365351467207,
      "loss": 4.645,
      "step": 17202
    },
    {
      "epoch": 2.355929882224048,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005336132629602747,
      "loss": 4.7664,
      "step": 17203
    },
    {
      "epoch": 2.356066831005204,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005335730105555565,
      "loss": 4.6859,
      "step": 17204
    },
    {
      "epoch": 2.3562037797863598,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005335327574582488,
      "loss": 4.7411,
      "step": 17205
    },
    {
      "epoch": 2.356340728567516,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005334925036686852,
      "loss": 4.6542,
      "step": 17206
    },
    {
      "epoch": 2.3564776773486718,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005334522491871994,
      "loss": 4.6756,
      "step": 17207
    },
    {
      "epoch": 2.3566146261298275,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005334119940141248,
      "loss": 4.7211,
      "step": 17208
    },
    {
      "epoch": 2.3567515749109833,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005333717381497953,
      "loss": 4.6666,
      "step": 17209
    },
    {
      "epoch": 2.356888523692139,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005333314815945443,
      "loss": 4.7085,
      "step": 17210
    },
    {
      "epoch": 2.357025472473295,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005332912243487054,
      "loss": 4.7335,
      "step": 17211
    },
    {
      "epoch": 2.3571624212544506,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005332509664126122,
      "loss": 4.6347,
      "step": 17212
    },
    {
      "epoch": 2.357299370035607,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005332107077865984,
      "loss": 4.8086,
      "step": 17213
    },
    {
      "epoch": 2.3574363188167626,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005331704484709978,
      "loss": 4.7266,
      "step": 17214
    },
    {
      "epoch": 2.3575732675979184,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005331301884661438,
      "loss": 4.7779,
      "step": 17215
    },
    {
      "epoch": 2.357710216379074,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00053308992777237,
      "loss": 4.7355,
      "step": 17216
    },
    {
      "epoch": 2.35784716516023,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005330496663900105,
      "loss": 4.6861,
      "step": 17217
    },
    {
      "epoch": 2.3579841139413857,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005330094043193984,
      "loss": 4.7578,
      "step": 17218
    },
    {
      "epoch": 2.358121062722542,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005329691415608676,
      "loss": 4.7153,
      "step": 17219
    },
    {
      "epoch": 2.3582580115036977,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005329288781147517,
      "loss": 4.7863,
      "step": 17220
    },
    {
      "epoch": 2.3583949602848535,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005328886139813845,
      "loss": 4.7453,
      "step": 17221
    },
    {
      "epoch": 2.3585319090660093,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005328483491610996,
      "loss": 4.7991,
      "step": 17222
    },
    {
      "epoch": 2.358668857847165,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005328080836542305,
      "loss": 4.8166,
      "step": 17223
    },
    {
      "epoch": 2.358805806628321,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005327678174611115,
      "loss": 4.75,
      "step": 17224
    },
    {
      "epoch": 2.358942755409477,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0005327275505820755,
      "loss": 4.7589,
      "step": 17225
    },
    {
      "epoch": 2.359079704190633,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005326872830174567,
      "loss": 4.6562,
      "step": 17226
    },
    {
      "epoch": 2.3592166529717886,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005326470147675887,
      "loss": 4.7491,
      "step": 17227
    },
    {
      "epoch": 2.3593536017529444,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005326067458328052,
      "loss": 4.669,
      "step": 17228
    },
    {
      "epoch": 2.3594905505341,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005325664762134398,
      "loss": 4.6658,
      "step": 17229
    },
    {
      "epoch": 2.359627499315256,
      "grad_norm": 0.375,
      "learning_rate": 0.0005325262059098265,
      "loss": 4.7554,
      "step": 17230
    },
    {
      "epoch": 2.3597644480964117,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005324859349222987,
      "loss": 4.7152,
      "step": 17231
    },
    {
      "epoch": 2.359901396877568,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005324456632511904,
      "loss": 4.7568,
      "step": 17232
    },
    {
      "epoch": 2.3600383456587237,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005324053908968353,
      "loss": 4.6285,
      "step": 17233
    },
    {
      "epoch": 2.3601752944398795,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000532365117859567,
      "loss": 4.7746,
      "step": 17234
    },
    {
      "epoch": 2.3603122432210353,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0005323248441397194,
      "loss": 4.722,
      "step": 17235
    },
    {
      "epoch": 2.360449192002191,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005322845697376262,
      "loss": 4.7745,
      "step": 17236
    },
    {
      "epoch": 2.3605861407833473,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000532244294653621,
      "loss": 4.7628,
      "step": 17237
    },
    {
      "epoch": 2.360723089564503,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005322040188880379,
      "loss": 4.7138,
      "step": 17238
    },
    {
      "epoch": 2.360860038345659,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005321637424412105,
      "loss": 4.7073,
      "step": 17239
    },
    {
      "epoch": 2.3609969871268146,
      "grad_norm": 0.375,
      "learning_rate": 0.0005321234653134726,
      "loss": 4.6964,
      "step": 17240
    },
    {
      "epoch": 2.3611339359079704,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005320831875051578,
      "loss": 4.687,
      "step": 17241
    },
    {
      "epoch": 2.361270884689126,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005320429090166003,
      "loss": 4.7019,
      "step": 17242
    },
    {
      "epoch": 2.361407833470282,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005320026298481336,
      "loss": 4.7342,
      "step": 17243
    },
    {
      "epoch": 2.361544782251438,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005319623500000915,
      "loss": 4.7723,
      "step": 17244
    },
    {
      "epoch": 2.361681731032594,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005319220694728079,
      "loss": 4.696,
      "step": 17245
    },
    {
      "epoch": 2.3618186798137497,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005318817882666165,
      "loss": 4.7749,
      "step": 17246
    },
    {
      "epoch": 2.3619556285949055,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005318415063818514,
      "loss": 4.7139,
      "step": 17247
    },
    {
      "epoch": 2.3620925773760613,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005318012238188461,
      "loss": 4.7553,
      "step": 17248
    },
    {
      "epoch": 2.362229526157217,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005317609405779346,
      "loss": 4.6937,
      "step": 17249
    },
    {
      "epoch": 2.362366474938373,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005317206566594507,
      "loss": 4.7854,
      "step": 17250
    },
    {
      "epoch": 2.362503423719529,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005316803720637282,
      "loss": 4.7455,
      "step": 17251
    },
    {
      "epoch": 2.362640372500685,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005316400867911011,
      "loss": 4.6788,
      "step": 17252
    },
    {
      "epoch": 2.3627773212818406,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005315998008419031,
      "loss": 4.7152,
      "step": 17253
    },
    {
      "epoch": 2.3629142700629964,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005315595142164682,
      "loss": 4.6923,
      "step": 17254
    },
    {
      "epoch": 2.363051218844152,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00053151922691513,
      "loss": 4.7962,
      "step": 17255
    },
    {
      "epoch": 2.3631881676253084,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005314789389382225,
      "loss": 4.641,
      "step": 17256
    },
    {
      "epoch": 2.363325116406464,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005314386502860798,
      "loss": 4.8476,
      "step": 17257
    },
    {
      "epoch": 2.36346206518762,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005313983609590354,
      "loss": 4.738,
      "step": 17258
    },
    {
      "epoch": 2.3635990139687757,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005313580709574236,
      "loss": 4.7611,
      "step": 17259
    },
    {
      "epoch": 2.3637359627499315,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000531317780281578,
      "loss": 4.6467,
      "step": 17260
    },
    {
      "epoch": 2.3638729115310873,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005312774889318326,
      "loss": 4.7234,
      "step": 17261
    },
    {
      "epoch": 2.364009860312243,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005312371969085212,
      "loss": 4.7476,
      "step": 17262
    },
    {
      "epoch": 2.3641468090933992,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005311969042119778,
      "loss": 4.7459,
      "step": 17263
    },
    {
      "epoch": 2.364283757874555,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005311566108425365,
      "loss": 4.7141,
      "step": 17264
    },
    {
      "epoch": 2.364420706655711,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005311163168005307,
      "loss": 4.6464,
      "step": 17265
    },
    {
      "epoch": 2.3645576554368666,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005310760220862949,
      "loss": 4.6877,
      "step": 17266
    },
    {
      "epoch": 2.3646946042180224,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005310357267001627,
      "loss": 4.7729,
      "step": 17267
    },
    {
      "epoch": 2.364831552999178,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005309954306424682,
      "loss": 4.6749,
      "step": 17268
    },
    {
      "epoch": 2.364968501780334,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005309551339135452,
      "loss": 4.7522,
      "step": 17269
    },
    {
      "epoch": 2.36510545056149,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005309148365137277,
      "loss": 4.6868,
      "step": 17270
    },
    {
      "epoch": 2.365242399342646,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005308745384433498,
      "loss": 4.7667,
      "step": 17271
    },
    {
      "epoch": 2.3653793481238017,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005308342397027451,
      "loss": 4.755,
      "step": 17272
    },
    {
      "epoch": 2.3655162969049575,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005307939402922479,
      "loss": 4.715,
      "step": 17273
    },
    {
      "epoch": 2.3656532456861132,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005307536402121921,
      "loss": 4.821,
      "step": 17274
    },
    {
      "epoch": 2.3657901944672695,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005307133394629117,
      "loss": 4.7676,
      "step": 17275
    },
    {
      "epoch": 2.3659271432484252,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005306730380447405,
      "loss": 4.7005,
      "step": 17276
    },
    {
      "epoch": 2.366064092029581,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005306327359580127,
      "loss": 4.7028,
      "step": 17277
    },
    {
      "epoch": 2.366201040810737,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005305924332030622,
      "loss": 4.7312,
      "step": 17278
    },
    {
      "epoch": 2.3663379895918926,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005305521297802228,
      "loss": 4.6995,
      "step": 17279
    },
    {
      "epoch": 2.3664749383730483,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005305118256898289,
      "loss": 4.7324,
      "step": 17280
    },
    {
      "epoch": 2.366611887154204,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005304715209322143,
      "loss": 4.682,
      "step": 17281
    },
    {
      "epoch": 2.3667488359353603,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000530431215507713,
      "loss": 4.6749,
      "step": 17282
    },
    {
      "epoch": 2.366885784716516,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000530390909416659,
      "loss": 4.7409,
      "step": 17283
    },
    {
      "epoch": 2.367022733497672,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005303506026593864,
      "loss": 4.5894,
      "step": 17284
    },
    {
      "epoch": 2.3671596822788277,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005303102952362293,
      "loss": 4.7309,
      "step": 17285
    },
    {
      "epoch": 2.3672966310599834,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005302699871475217,
      "loss": 4.7175,
      "step": 17286
    },
    {
      "epoch": 2.3674335798411392,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005302296783935974,
      "loss": 4.7172,
      "step": 17287
    },
    {
      "epoch": 2.3675705286222954,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005301893689747906,
      "loss": 4.6819,
      "step": 17288
    },
    {
      "epoch": 2.367707477403451,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005301490588914356,
      "loss": 4.7343,
      "step": 17289
    },
    {
      "epoch": 2.367844426184607,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005301087481438661,
      "loss": 4.7241,
      "step": 17290
    },
    {
      "epoch": 2.3679813749657628,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005300684367324165,
      "loss": 4.633,
      "step": 17291
    },
    {
      "epoch": 2.3681183237469186,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005300281246574204,
      "loss": 4.7114,
      "step": 17292
    },
    {
      "epoch": 2.3682552725280743,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005299878119192125,
      "loss": 4.7637,
      "step": 17293
    },
    {
      "epoch": 2.3683922213092305,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005299474985181263,
      "loss": 4.683,
      "step": 17294
    },
    {
      "epoch": 2.3685291700903863,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005299071844544963,
      "loss": 4.7617,
      "step": 17295
    },
    {
      "epoch": 2.368666118871542,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005298668697286562,
      "loss": 4.6863,
      "step": 17296
    },
    {
      "epoch": 2.368803067652698,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005298265543409404,
      "loss": 4.7357,
      "step": 17297
    },
    {
      "epoch": 2.3689400164338537,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005297862382916831,
      "loss": 4.7958,
      "step": 17298
    },
    {
      "epoch": 2.3690769652150094,
      "grad_norm": 0.326171875,
      "learning_rate": 0.000529745921581218,
      "loss": 4.7185,
      "step": 17299
    },
    {
      "epoch": 2.369213913996165,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005297056042098796,
      "loss": 4.7511,
      "step": 17300
    },
    {
      "epoch": 2.3693508627773214,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005296652861780017,
      "loss": 4.7677,
      "step": 17301
    },
    {
      "epoch": 2.369487811558477,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005296249674859188,
      "loss": 4.6632,
      "step": 17302
    },
    {
      "epoch": 2.369624760339633,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005295846481339646,
      "loss": 4.6906,
      "step": 17303
    },
    {
      "epoch": 2.3697617091207888,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005295443281224737,
      "loss": 4.6415,
      "step": 17304
    },
    {
      "epoch": 2.3698986579019445,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005295040074517798,
      "loss": 4.7458,
      "step": 17305
    },
    {
      "epoch": 2.3700356066831008,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005294636861222172,
      "loss": 4.6842,
      "step": 17306
    },
    {
      "epoch": 2.3701725554642565,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005294233641341203,
      "loss": 4.7769,
      "step": 17307
    },
    {
      "epoch": 2.3703095042454123,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000529383041487823,
      "loss": 4.6396,
      "step": 17308
    },
    {
      "epoch": 2.370446453026568,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005293427181836595,
      "loss": 4.6361,
      "step": 17309
    },
    {
      "epoch": 2.370583401807724,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005293023942219639,
      "loss": 4.6937,
      "step": 17310
    },
    {
      "epoch": 2.3707203505888796,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005292620696030706,
      "loss": 4.6947,
      "step": 17311
    },
    {
      "epoch": 2.3708572993700354,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005292217443273136,
      "loss": 4.695,
      "step": 17312
    },
    {
      "epoch": 2.3709942481511916,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005291814183950272,
      "loss": 4.7415,
      "step": 17313
    },
    {
      "epoch": 2.3711311969323474,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005291410918065453,
      "loss": 4.7213,
      "step": 17314
    },
    {
      "epoch": 2.371268145713503,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0005291007645622023,
      "loss": 4.7113,
      "step": 17315
    },
    {
      "epoch": 2.371405094494659,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005290604366623327,
      "loss": 4.7531,
      "step": 17316
    },
    {
      "epoch": 2.3715420432758147,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005290201081072703,
      "loss": 4.7873,
      "step": 17317
    },
    {
      "epoch": 2.3716789920569705,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005289797788973495,
      "loss": 4.7466,
      "step": 17318
    },
    {
      "epoch": 2.3718159408381263,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005289394490329043,
      "loss": 4.7461,
      "step": 17319
    },
    {
      "epoch": 2.3719528896192825,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000528899118514269,
      "loss": 4.7474,
      "step": 17320
    },
    {
      "epoch": 2.3720898384004383,
      "grad_norm": 0.359375,
      "learning_rate": 0.000528858787341778,
      "loss": 4.8025,
      "step": 17321
    },
    {
      "epoch": 2.372226787181594,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005288184555157656,
      "loss": 4.7873,
      "step": 17322
    },
    {
      "epoch": 2.37236373596275,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005287781230365657,
      "loss": 4.763,
      "step": 17323
    },
    {
      "epoch": 2.3725006847439056,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005287377899045129,
      "loss": 4.6792,
      "step": 17324
    },
    {
      "epoch": 2.372637633525062,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000528697456119941,
      "loss": 4.6436,
      "step": 17325
    },
    {
      "epoch": 2.3727745823062176,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005286571216831848,
      "loss": 4.7642,
      "step": 17326
    },
    {
      "epoch": 2.3729115310873734,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005286167865945782,
      "loss": 4.577,
      "step": 17327
    },
    {
      "epoch": 2.373048479868529,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005285764508544554,
      "loss": 4.8274,
      "step": 17328
    },
    {
      "epoch": 2.373185428649685,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005285361144631511,
      "loss": 4.6661,
      "step": 17329
    },
    {
      "epoch": 2.3733223774308407,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005284957774209991,
      "loss": 4.7558,
      "step": 17330
    },
    {
      "epoch": 2.3734593262119965,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000528455439728334,
      "loss": 4.6703,
      "step": 17331
    },
    {
      "epoch": 2.3735962749931527,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005284151013854899,
      "loss": 4.6452,
      "step": 17332
    },
    {
      "epoch": 2.3737332237743085,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005283747623928012,
      "loss": 4.7201,
      "step": 17333
    },
    {
      "epoch": 2.3738701725554643,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005283344227506021,
      "loss": 4.8206,
      "step": 17334
    },
    {
      "epoch": 2.37400712133662,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005282940824592271,
      "loss": 4.6817,
      "step": 17335
    },
    {
      "epoch": 2.374144070117776,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005282537415190104,
      "loss": 4.8082,
      "step": 17336
    },
    {
      "epoch": 2.3742810188989316,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005282133999302861,
      "loss": 4.7001,
      "step": 17337
    },
    {
      "epoch": 2.374417967680088,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005281730576933889,
      "loss": 4.6483,
      "step": 17338
    },
    {
      "epoch": 2.3745549164612436,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005281327148086529,
      "loss": 4.6816,
      "step": 17339
    },
    {
      "epoch": 2.3746918652423994,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005280923712764124,
      "loss": 4.7402,
      "step": 17340
    },
    {
      "epoch": 2.374828814023555,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005280520270970017,
      "loss": 4.6947,
      "step": 17341
    },
    {
      "epoch": 2.374965762804711,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005280116822707555,
      "loss": 4.7538,
      "step": 17342
    },
    {
      "epoch": 2.3751027115858667,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005279713367980079,
      "loss": 4.73,
      "step": 17343
    },
    {
      "epoch": 2.375239660367023,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000527930990679093,
      "loss": 4.7009,
      "step": 17344
    },
    {
      "epoch": 2.3753766091481787,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005278906439143457,
      "loss": 4.722,
      "step": 17345
    },
    {
      "epoch": 2.3755135579293345,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005278502965040999,
      "loss": 4.6987,
      "step": 17346
    },
    {
      "epoch": 2.3756505067104903,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005278099484486903,
      "loss": 4.6673,
      "step": 17347
    },
    {
      "epoch": 2.375787455491646,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000527769599748451,
      "loss": 4.6807,
      "step": 17348
    },
    {
      "epoch": 2.375924404272802,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005277292504037164,
      "loss": 4.7709,
      "step": 17349
    },
    {
      "epoch": 2.3760613530539576,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005276889004148212,
      "loss": 4.6785,
      "step": 17350
    },
    {
      "epoch": 2.376198301835114,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005276485497820994,
      "loss": 4.6903,
      "step": 17351
    },
    {
      "epoch": 2.3763352506162696,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005276081985058856,
      "loss": 4.7605,
      "step": 17352
    },
    {
      "epoch": 2.3764721993974254,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005275678465865142,
      "loss": 4.7456,
      "step": 17353
    },
    {
      "epoch": 2.376609148178581,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005275274940243196,
      "loss": 4.7044,
      "step": 17354
    },
    {
      "epoch": 2.376746096959737,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005274871408196361,
      "loss": 4.7044,
      "step": 17355
    },
    {
      "epoch": 2.376883045740893,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005274467869727983,
      "loss": 4.7153,
      "step": 17356
    },
    {
      "epoch": 2.377019994522049,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005274064324841404,
      "loss": 4.7348,
      "step": 17357
    },
    {
      "epoch": 2.3771569433032047,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000527366077353997,
      "loss": 4.6859,
      "step": 17358
    },
    {
      "epoch": 2.3772938920843605,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005273257215827025,
      "loss": 4.6678,
      "step": 17359
    },
    {
      "epoch": 2.3774308408655163,
      "grad_norm": 0.375,
      "learning_rate": 0.0005272853651705914,
      "loss": 4.7329,
      "step": 17360
    },
    {
      "epoch": 2.377567789646672,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005272450081179979,
      "loss": 4.7057,
      "step": 17361
    },
    {
      "epoch": 2.377704738427828,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005272046504252566,
      "loss": 4.6964,
      "step": 17362
    },
    {
      "epoch": 2.377841687208984,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005271642920927021,
      "loss": 4.7579,
      "step": 17363
    },
    {
      "epoch": 2.37797863599014,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005271239331206686,
      "loss": 4.7792,
      "step": 17364
    },
    {
      "epoch": 2.3781155847712956,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005270835735094909,
      "loss": 4.6626,
      "step": 17365
    },
    {
      "epoch": 2.3782525335524514,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005270432132595031,
      "loss": 4.6819,
      "step": 17366
    },
    {
      "epoch": 2.378389482333607,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005270028523710399,
      "loss": 4.6246,
      "step": 17367
    },
    {
      "epoch": 2.378526431114763,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005269624908444356,
      "loss": 4.6842,
      "step": 17368
    },
    {
      "epoch": 2.3786633798959187,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000526922128680025,
      "loss": 4.7542,
      "step": 17369
    },
    {
      "epoch": 2.378800328677075,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005268817658781421,
      "loss": 4.7457,
      "step": 17370
    },
    {
      "epoch": 2.3789372774582307,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005268414024391219,
      "loss": 4.8009,
      "step": 17371
    },
    {
      "epoch": 2.3790742262393865,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005268010383632987,
      "loss": 4.7258,
      "step": 17372
    },
    {
      "epoch": 2.3792111750205422,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005267606736510069,
      "loss": 4.7681,
      "step": 17373
    },
    {
      "epoch": 2.379348123801698,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005267203083025811,
      "loss": 4.6739,
      "step": 17374
    },
    {
      "epoch": 2.3794850725828542,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005266799423183559,
      "loss": 4.7101,
      "step": 17375
    },
    {
      "epoch": 2.37962202136401,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005266395756986657,
      "loss": 4.7497,
      "step": 17376
    },
    {
      "epoch": 2.379758970145166,
      "grad_norm": 0.34375,
      "learning_rate": 0.000526599208443845,
      "loss": 4.77,
      "step": 17377
    },
    {
      "epoch": 2.3798959189263216,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005265588405542285,
      "loss": 4.6989,
      "step": 17378
    },
    {
      "epoch": 2.3800328677074774,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005265184720301507,
      "loss": 4.7714,
      "step": 17379
    },
    {
      "epoch": 2.380169816488633,
      "grad_norm": 0.3984375,
      "learning_rate": 0.000526478102871946,
      "loss": 4.6804,
      "step": 17380
    },
    {
      "epoch": 2.380306765269789,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000526437733079949,
      "loss": 4.7207,
      "step": 17381
    },
    {
      "epoch": 2.380443714050945,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005263973626544943,
      "loss": 4.6227,
      "step": 17382
    },
    {
      "epoch": 2.380580662832101,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005263569915959166,
      "loss": 4.7196,
      "step": 17383
    },
    {
      "epoch": 2.3807176116132567,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005263166199045501,
      "loss": 4.7246,
      "step": 17384
    },
    {
      "epoch": 2.3808545603944125,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005262762475807298,
      "loss": 4.6624,
      "step": 17385
    },
    {
      "epoch": 2.3809915091755682,
      "grad_norm": 0.375,
      "learning_rate": 0.0005262358746247899,
      "loss": 4.7989,
      "step": 17386
    },
    {
      "epoch": 2.381128457956724,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005261955010370653,
      "loss": 4.6887,
      "step": 17387
    },
    {
      "epoch": 2.38126540673788,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005261551268178903,
      "loss": 4.81,
      "step": 17388
    },
    {
      "epoch": 2.381402355519036,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005261147519675996,
      "loss": 4.7617,
      "step": 17389
    },
    {
      "epoch": 2.381539304300192,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005260743764865279,
      "loss": 4.6228,
      "step": 17390
    },
    {
      "epoch": 2.3816762530813476,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005260340003750096,
      "loss": 4.7126,
      "step": 17391
    },
    {
      "epoch": 2.3818132018625033,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005259936236333795,
      "loss": 4.7731,
      "step": 17392
    },
    {
      "epoch": 2.381950150643659,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005259532462619722,
      "loss": 4.7533,
      "step": 17393
    },
    {
      "epoch": 2.3820870994248153,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005259128682611222,
      "loss": 4.6308,
      "step": 17394
    },
    {
      "epoch": 2.382224048205971,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005258724896311641,
      "loss": 4.7454,
      "step": 17395
    },
    {
      "epoch": 2.382360996987127,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005258321103724327,
      "loss": 4.6987,
      "step": 17396
    },
    {
      "epoch": 2.3824979457682827,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005257917304852624,
      "loss": 4.8522,
      "step": 17397
    },
    {
      "epoch": 2.3826348945494384,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005257513499699882,
      "loss": 4.75,
      "step": 17398
    },
    {
      "epoch": 2.382771843330594,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005257109688269442,
      "loss": 4.7399,
      "step": 17399
    },
    {
      "epoch": 2.38290879211175,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005256705870564655,
      "loss": 4.6632,
      "step": 17400
    },
    {
      "epoch": 2.383045740892906,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005256302046588866,
      "loss": 4.7203,
      "step": 17401
    },
    {
      "epoch": 2.383182689674062,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005255898216345423,
      "loss": 4.7001,
      "step": 17402
    },
    {
      "epoch": 2.3833196384552178,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000525549437983767,
      "loss": 4.6573,
      "step": 17403
    },
    {
      "epoch": 2.3834565872363735,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005255090537068956,
      "loss": 4.7353,
      "step": 17404
    },
    {
      "epoch": 2.3835935360175293,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005254686688042627,
      "loss": 4.7702,
      "step": 17405
    },
    {
      "epoch": 2.383730484798685,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005254282832762028,
      "loss": 4.7219,
      "step": 17406
    },
    {
      "epoch": 2.3838674335798413,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005253878971230509,
      "loss": 4.6102,
      "step": 17407
    },
    {
      "epoch": 2.384004382360997,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005253475103451414,
      "loss": 4.7123,
      "step": 17408
    },
    {
      "epoch": 2.384141331142153,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005253071229428092,
      "loss": 4.649,
      "step": 17409
    },
    {
      "epoch": 2.3842782799233087,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000525266734916389,
      "loss": 4.7905,
      "step": 17410
    },
    {
      "epoch": 2.3844152287044644,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005252263462662152,
      "loss": 4.7903,
      "step": 17411
    },
    {
      "epoch": 2.38455217748562,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000525185956992623,
      "loss": 4.7636,
      "step": 17412
    },
    {
      "epoch": 2.3846891262667764,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005251455670959467,
      "loss": 4.6114,
      "step": 17413
    },
    {
      "epoch": 2.384826075047932,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005251051765765212,
      "loss": 4.7715,
      "step": 17414
    },
    {
      "epoch": 2.384963023829088,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005250647854346813,
      "loss": 4.6884,
      "step": 17415
    },
    {
      "epoch": 2.3850999726102438,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005250243936707616,
      "loss": 4.702,
      "step": 17416
    },
    {
      "epoch": 2.3852369213913995,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005249840012850967,
      "loss": 4.8012,
      "step": 17417
    },
    {
      "epoch": 2.3853738701725553,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0005249436082780217,
      "loss": 4.6697,
      "step": 17418
    },
    {
      "epoch": 2.385510818953711,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005249032146498712,
      "loss": 4.6314,
      "step": 17419
    },
    {
      "epoch": 2.3856477677348673,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005248628204009798,
      "loss": 4.7367,
      "step": 17420
    },
    {
      "epoch": 2.385784716516023,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005248224255316825,
      "loss": 4.7691,
      "step": 17421
    },
    {
      "epoch": 2.385921665297179,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005247820300423137,
      "loss": 4.7635,
      "step": 17422
    },
    {
      "epoch": 2.3860586140783346,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005247416339332086,
      "loss": 4.7166,
      "step": 17423
    },
    {
      "epoch": 2.3861955628594904,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005247012372047017,
      "loss": 4.747,
      "step": 17424
    },
    {
      "epoch": 2.3863325116406466,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005246608398571278,
      "loss": 4.7382,
      "step": 17425
    },
    {
      "epoch": 2.3864694604218024,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005246204418908217,
      "loss": 4.7839,
      "step": 17426
    },
    {
      "epoch": 2.386606409202958,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005245800433061182,
      "loss": 4.7456,
      "step": 17427
    },
    {
      "epoch": 2.386743357984114,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0005245396441033523,
      "loss": 4.6703,
      "step": 17428
    },
    {
      "epoch": 2.3868803067652697,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005244992442828585,
      "loss": 4.7531,
      "step": 17429
    },
    {
      "epoch": 2.3870172555464255,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005244588438449718,
      "loss": 4.7212,
      "step": 17430
    },
    {
      "epoch": 2.3871542043275813,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005244184427900268,
      "loss": 4.7555,
      "step": 17431
    },
    {
      "epoch": 2.3872911531087375,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005243780411183584,
      "loss": 4.6961,
      "step": 17432
    },
    {
      "epoch": 2.3874281018898933,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005243376388303015,
      "loss": 4.7236,
      "step": 17433
    },
    {
      "epoch": 2.387565050671049,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000524297235926191,
      "loss": 4.7093,
      "step": 17434
    },
    {
      "epoch": 2.387701999452205,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005242568324063615,
      "loss": 4.7076,
      "step": 17435
    },
    {
      "epoch": 2.3878389482333606,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005242164282711479,
      "loss": 4.7634,
      "step": 17436
    },
    {
      "epoch": 2.3879758970145164,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000524176023520885,
      "loss": 4.6408,
      "step": 17437
    },
    {
      "epoch": 2.388112845795672,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005241356181559078,
      "loss": 4.7132,
      "step": 17438
    },
    {
      "epoch": 2.3882497945768284,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005240952121765512,
      "loss": 4.6358,
      "step": 17439
    },
    {
      "epoch": 2.388386743357984,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005240548055831497,
      "loss": 4.7053,
      "step": 17440
    },
    {
      "epoch": 2.38852369213914,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005240143983760386,
      "loss": 4.7174,
      "step": 17441
    },
    {
      "epoch": 2.3886606409202957,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005239739905555524,
      "loss": 4.7274,
      "step": 17442
    },
    {
      "epoch": 2.3887975897014515,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005239335821220261,
      "loss": 4.724,
      "step": 17443
    },
    {
      "epoch": 2.3889345384826077,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005238931730757947,
      "loss": 4.6806,
      "step": 17444
    },
    {
      "epoch": 2.3890714872637635,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005238527634171928,
      "loss": 4.6581,
      "step": 17445
    },
    {
      "epoch": 2.3892084360449193,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005238123531465556,
      "loss": 4.6732,
      "step": 17446
    },
    {
      "epoch": 2.389345384826075,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005237719422642177,
      "loss": 4.728,
      "step": 17447
    },
    {
      "epoch": 2.389482333607231,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005237315307705144,
      "loss": 4.8055,
      "step": 17448
    },
    {
      "epoch": 2.3896192823883866,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005236911186657801,
      "loss": 4.7295,
      "step": 17449
    },
    {
      "epoch": 2.3897562311695424,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00052365070595035,
      "loss": 4.6599,
      "step": 17450
    },
    {
      "epoch": 2.3898931799506986,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005236102926245589,
      "loss": 4.7476,
      "step": 17451
    },
    {
      "epoch": 2.3900301287318544,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005235698786887419,
      "loss": 4.6971,
      "step": 17452
    },
    {
      "epoch": 2.39016707751301,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005235294641432336,
      "loss": 4.6848,
      "step": 17453
    },
    {
      "epoch": 2.390304026294166,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005234890489883693,
      "loss": 4.7621,
      "step": 17454
    },
    {
      "epoch": 2.3904409750753217,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005234486332244836,
      "loss": 4.6939,
      "step": 17455
    },
    {
      "epoch": 2.3905779238564775,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005234082168519116,
      "loss": 4.7617,
      "step": 17456
    },
    {
      "epoch": 2.3907148726376337,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005233677998709883,
      "loss": 4.7819,
      "step": 17457
    },
    {
      "epoch": 2.3908518214187895,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005233273822820484,
      "loss": 4.6934,
      "step": 17458
    },
    {
      "epoch": 2.3909887701999453,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005232869640854271,
      "loss": 4.7748,
      "step": 17459
    },
    {
      "epoch": 2.391125718981101,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005232465452814593,
      "loss": 4.7658,
      "step": 17460
    },
    {
      "epoch": 2.391262667762257,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00052320612587048,
      "loss": 4.788,
      "step": 17461
    },
    {
      "epoch": 2.3913996165434126,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005231657058528239,
      "loss": 4.7464,
      "step": 17462
    },
    {
      "epoch": 2.391536565324569,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005231252852288262,
      "loss": 4.7652,
      "step": 17463
    },
    {
      "epoch": 2.3916735141057246,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000523084863998822,
      "loss": 4.7239,
      "step": 17464
    },
    {
      "epoch": 2.3918104628868804,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005230444421631459,
      "loss": 4.6369,
      "step": 17465
    },
    {
      "epoch": 2.391947411668036,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005230040197221333,
      "loss": 4.8021,
      "step": 17466
    },
    {
      "epoch": 2.392084360449192,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005229635966761189,
      "loss": 4.7419,
      "step": 17467
    },
    {
      "epoch": 2.3922213092303477,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005229231730254377,
      "loss": 4.7884,
      "step": 17468
    },
    {
      "epoch": 2.3923582580115035,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005228827487704249,
      "loss": 4.7285,
      "step": 17469
    },
    {
      "epoch": 2.3924952067926597,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005228423239114154,
      "loss": 4.724,
      "step": 17470
    },
    {
      "epoch": 2.3926321555738155,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005228018984487442,
      "loss": 4.7611,
      "step": 17471
    },
    {
      "epoch": 2.3927691043549713,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005227614723827463,
      "loss": 4.6948,
      "step": 17472
    },
    {
      "epoch": 2.392906053136127,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005227210457137568,
      "loss": 4.8108,
      "step": 17473
    },
    {
      "epoch": 2.393043001917283,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005226806184421104,
      "loss": 4.7189,
      "step": 17474
    },
    {
      "epoch": 2.393179950698439,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005226401905681428,
      "loss": 4.6142,
      "step": 17475
    },
    {
      "epoch": 2.393316899479595,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005225997620921883,
      "loss": 4.7778,
      "step": 17476
    },
    {
      "epoch": 2.3934538482607506,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005225593330145823,
      "loss": 4.6789,
      "step": 17477
    },
    {
      "epoch": 2.3935907970419064,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00052251890333566,
      "loss": 4.7012,
      "step": 17478
    },
    {
      "epoch": 2.393727745823062,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005224784730557562,
      "loss": 4.64,
      "step": 17479
    },
    {
      "epoch": 2.393864694604218,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000522438042175206,
      "loss": 4.7305,
      "step": 17480
    },
    {
      "epoch": 2.3940016433853737,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005223976106943445,
      "loss": 4.7706,
      "step": 17481
    },
    {
      "epoch": 2.39413859216653,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005223571786135068,
      "loss": 4.6388,
      "step": 17482
    },
    {
      "epoch": 2.3942755409476857,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005223167459330278,
      "loss": 4.7469,
      "step": 17483
    },
    {
      "epoch": 2.3944124897288415,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005222763126532426,
      "loss": 4.6342,
      "step": 17484
    },
    {
      "epoch": 2.3945494385099972,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005222358787744867,
      "loss": 4.6916,
      "step": 17485
    },
    {
      "epoch": 2.394686387291153,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005221954442970946,
      "loss": 4.7889,
      "step": 17486
    },
    {
      "epoch": 2.394823336072309,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005221550092214018,
      "loss": 4.7075,
      "step": 17487
    },
    {
      "epoch": 2.3949602848534646,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005221145735477431,
      "loss": 4.7896,
      "step": 17488
    },
    {
      "epoch": 2.395097233634621,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000522074137276454,
      "loss": 4.7531,
      "step": 17489
    },
    {
      "epoch": 2.3952341824157766,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000522033700407869,
      "loss": 4.7231,
      "step": 17490
    },
    {
      "epoch": 2.3953711311969323,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005219932629423237,
      "loss": 4.7031,
      "step": 17491
    },
    {
      "epoch": 2.395508079978088,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005219528248801532,
      "loss": 4.6708,
      "step": 17492
    },
    {
      "epoch": 2.395645028759244,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005219123862216924,
      "loss": 4.6288,
      "step": 17493
    },
    {
      "epoch": 2.3957819775404,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0005218719469672765,
      "loss": 4.7356,
      "step": 17494
    },
    {
      "epoch": 2.395918926321556,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005218315071172407,
      "loss": 4.7653,
      "step": 17495
    },
    {
      "epoch": 2.3960558751027117,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00052179106667192,
      "loss": 4.7532,
      "step": 17496
    },
    {
      "epoch": 2.3961928238838675,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005217506256316496,
      "loss": 4.7238,
      "step": 17497
    },
    {
      "epoch": 2.3963297726650232,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005217101839967649,
      "loss": 4.6911,
      "step": 17498
    },
    {
      "epoch": 2.396466721446179,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005216697417676006,
      "loss": 4.7571,
      "step": 17499
    },
    {
      "epoch": 2.396603670227335,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005216292989444922,
      "loss": 4.7148,
      "step": 17500
    },
    {
      "epoch": 2.396740619008491,
      "grad_norm": 0.4375,
      "learning_rate": 0.0005215888555277746,
      "loss": 4.6295,
      "step": 17501
    },
    {
      "epoch": 2.396877567789647,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005215484115177832,
      "loss": 4.7175,
      "step": 17502
    },
    {
      "epoch": 2.3970145165708026,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005215079669148531,
      "loss": 4.7128,
      "step": 17503
    },
    {
      "epoch": 2.3971514653519583,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005214675217193193,
      "loss": 4.7651,
      "step": 17504
    },
    {
      "epoch": 2.397288414133114,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005214270759315171,
      "loss": 4.7634,
      "step": 17505
    },
    {
      "epoch": 2.39742536291427,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005213866295517819,
      "loss": 4.732,
      "step": 17506
    },
    {
      "epoch": 2.3975623116954257,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0005213461825804485,
      "loss": 4.6696,
      "step": 17507
    },
    {
      "epoch": 2.397699260476582,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005213057350178525,
      "loss": 4.7123,
      "step": 17508
    },
    {
      "epoch": 2.3978362092577377,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005212652868643288,
      "loss": 4.7095,
      "step": 17509
    },
    {
      "epoch": 2.3979731580388934,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005212248381202126,
      "loss": 4.7111,
      "step": 17510
    },
    {
      "epoch": 2.398110106820049,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0005211843887858392,
      "loss": 4.7172,
      "step": 17511
    },
    {
      "epoch": 2.398247055601205,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005211439388615437,
      "loss": 4.6552,
      "step": 17512
    },
    {
      "epoch": 2.398384004382361,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005211034883476617,
      "loss": 4.6072,
      "step": 17513
    },
    {
      "epoch": 2.398520953163517,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000521063037244528,
      "loss": 4.7628,
      "step": 17514
    },
    {
      "epoch": 2.3986579019446728,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005210225855524779,
      "loss": 4.8253,
      "step": 17515
    },
    {
      "epoch": 2.3987948507258285,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005209821332718468,
      "loss": 4.7237,
      "step": 17516
    },
    {
      "epoch": 2.3989317995069843,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00052094168040297,
      "loss": 4.7923,
      "step": 17517
    },
    {
      "epoch": 2.39906874828814,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005209012269461823,
      "loss": 4.8405,
      "step": 17518
    },
    {
      "epoch": 2.399205697069296,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005208607729018194,
      "loss": 4.7495,
      "step": 17519
    },
    {
      "epoch": 2.399342645850452,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005208203182702164,
      "loss": 4.7308,
      "step": 17520
    },
    {
      "epoch": 2.399479594631608,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005207798630517084,
      "loss": 4.7081,
      "step": 17521
    },
    {
      "epoch": 2.3996165434127636,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000520739407246631,
      "loss": 4.7179,
      "step": 17522
    },
    {
      "epoch": 2.3997534921939194,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005206989508553192,
      "loss": 4.7346,
      "step": 17523
    },
    {
      "epoch": 2.399890440975075,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005206584938781083,
      "loss": 4.7345,
      "step": 17524
    },
    {
      "epoch": 2.400027389756231,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005206180363153337,
      "loss": 4.7263,
      "step": 17525
    },
    {
      "epoch": 2.400164338537387,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005205775781673307,
      "loss": 4.7259,
      "step": 17526
    },
    {
      "epoch": 2.400301287318543,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005205371194344342,
      "loss": 4.6984,
      "step": 17527
    },
    {
      "epoch": 2.4004382360996988,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00052049666011698,
      "loss": 4.6227,
      "step": 17528
    },
    {
      "epoch": 2.4005751848808545,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005204562002153032,
      "loss": 4.7169,
      "step": 17529
    },
    {
      "epoch": 2.4007121336620103,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005204157397297389,
      "loss": 4.6747,
      "step": 17530
    },
    {
      "epoch": 2.400849082443166,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005203752786606228,
      "loss": 4.6926,
      "step": 17531
    },
    {
      "epoch": 2.4009860312243223,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005203348170082897,
      "loss": 4.7317,
      "step": 17532
    },
    {
      "epoch": 2.401122980005478,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005202943547730755,
      "loss": 4.6826,
      "step": 17533
    },
    {
      "epoch": 2.401259928786634,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005202538919553151,
      "loss": 4.6413,
      "step": 17534
    },
    {
      "epoch": 2.4013968775677896,
      "grad_norm": 0.34375,
      "learning_rate": 0.000520213428555344,
      "loss": 4.6935,
      "step": 17535
    },
    {
      "epoch": 2.4015338263489454,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005201729645734975,
      "loss": 4.6866,
      "step": 17536
    },
    {
      "epoch": 2.401670775130101,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005201325000101109,
      "loss": 4.7513,
      "step": 17537
    },
    {
      "epoch": 2.401807723911257,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005200920348655196,
      "loss": 4.6339,
      "step": 17538
    },
    {
      "epoch": 2.401944672692413,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005200515691400588,
      "loss": 4.6419,
      "step": 17539
    },
    {
      "epoch": 2.402081621473569,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005200111028340641,
      "loss": 4.7519,
      "step": 17540
    },
    {
      "epoch": 2.4022185702547247,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005199706359478705,
      "loss": 4.8148,
      "step": 17541
    },
    {
      "epoch": 2.4023555190358805,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005199301684818138,
      "loss": 4.6627,
      "step": 17542
    },
    {
      "epoch": 2.4024924678170363,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005198897004362289,
      "loss": 4.6981,
      "step": 17543
    },
    {
      "epoch": 2.4026294165981925,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005198492318114516,
      "loss": 4.7636,
      "step": 17544
    },
    {
      "epoch": 2.4027663653793483,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005198087626078169,
      "loss": 4.8055,
      "step": 17545
    },
    {
      "epoch": 2.402903314160504,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005197682928256605,
      "loss": 4.7082,
      "step": 17546
    },
    {
      "epoch": 2.40304026294166,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0005197278224653176,
      "loss": 4.7178,
      "step": 17547
    },
    {
      "epoch": 2.4031772117228156,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005196873515271237,
      "loss": 4.7397,
      "step": 17548
    },
    {
      "epoch": 2.4033141605039714,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005196468800114139,
      "loss": 4.7342,
      "step": 17549
    },
    {
      "epoch": 2.403451109285127,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005196064079185239,
      "loss": 4.7095,
      "step": 17550
    },
    {
      "epoch": 2.4035880580662834,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000519565935248789,
      "loss": 4.6806,
      "step": 17551
    },
    {
      "epoch": 2.403725006847439,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005195254620025447,
      "loss": 4.8235,
      "step": 17552
    },
    {
      "epoch": 2.403861955628595,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005194849881801262,
      "loss": 4.7812,
      "step": 17553
    },
    {
      "epoch": 2.4039989044097507,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005194445137818691,
      "loss": 4.7533,
      "step": 17554
    },
    {
      "epoch": 2.4041358531909065,
      "grad_norm": 0.359375,
      "learning_rate": 0.000519404038808109,
      "loss": 4.7329,
      "step": 17555
    },
    {
      "epoch": 2.4042728019720623,
      "grad_norm": 0.375,
      "learning_rate": 0.0005193635632591807,
      "loss": 4.7731,
      "step": 17556
    },
    {
      "epoch": 2.404409750753218,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005193230871354203,
      "loss": 4.6965,
      "step": 17557
    },
    {
      "epoch": 2.4045466995343743,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005192826104371629,
      "loss": 4.7699,
      "step": 17558
    },
    {
      "epoch": 2.40468364831553,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0005192421331647438,
      "loss": 4.671,
      "step": 17559
    },
    {
      "epoch": 2.404820597096686,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005192016553184989,
      "loss": 4.7916,
      "step": 17560
    },
    {
      "epoch": 2.4049575458778416,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005191611768987633,
      "loss": 4.7182,
      "step": 17561
    },
    {
      "epoch": 2.4050944946589974,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005191206979058726,
      "loss": 4.7068,
      "step": 17562
    },
    {
      "epoch": 2.4052314434401536,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000519080218340162,
      "loss": 4.7387,
      "step": 17563
    },
    {
      "epoch": 2.4053683922213094,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005190397382019674,
      "loss": 4.7568,
      "step": 17564
    },
    {
      "epoch": 2.405505341002465,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005189992574916239,
      "loss": 4.7599,
      "step": 17565
    },
    {
      "epoch": 2.405642289783621,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005189587762094672,
      "loss": 4.7754,
      "step": 17566
    },
    {
      "epoch": 2.4057792385647767,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005189182943558326,
      "loss": 4.7486,
      "step": 17567
    },
    {
      "epoch": 2.4059161873459325,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005188778119310558,
      "loss": 4.6991,
      "step": 17568
    },
    {
      "epoch": 2.4060531361270883,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000518837328935472,
      "loss": 4.7007,
      "step": 17569
    },
    {
      "epoch": 2.4061900849082445,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000518796845369417,
      "loss": 4.7478,
      "step": 17570
    },
    {
      "epoch": 2.4063270336894003,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005187563612332262,
      "loss": 4.7225,
      "step": 17571
    },
    {
      "epoch": 2.406463982470556,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005187158765272348,
      "loss": 4.7337,
      "step": 17572
    },
    {
      "epoch": 2.406600931251712,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005186753912517787,
      "loss": 4.8003,
      "step": 17573
    },
    {
      "epoch": 2.4067378800328676,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005186349054071933,
      "loss": 4.72,
      "step": 17574
    },
    {
      "epoch": 2.4068748288140234,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005185944189938141,
      "loss": 4.6787,
      "step": 17575
    },
    {
      "epoch": 2.4070117775951796,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005185539320119764,
      "loss": 4.7105,
      "step": 17576
    },
    {
      "epoch": 2.4071487263763354,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005185134444620162,
      "loss": 4.6682,
      "step": 17577
    },
    {
      "epoch": 2.407285675157491,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005184729563442686,
      "loss": 4.7159,
      "step": 17578
    },
    {
      "epoch": 2.407422623938647,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005184324676590693,
      "loss": 4.6643,
      "step": 17579
    },
    {
      "epoch": 2.4075595727198027,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000518391978406754,
      "loss": 4.7006,
      "step": 17580
    },
    {
      "epoch": 2.4076965215009585,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005183514885876579,
      "loss": 4.6772,
      "step": 17581
    },
    {
      "epoch": 2.4078334702821147,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005183109982021168,
      "loss": 4.7337,
      "step": 17582
    },
    {
      "epoch": 2.4079704190632705,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005182705072504662,
      "loss": 4.795,
      "step": 17583
    },
    {
      "epoch": 2.4081073678444262,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005182300157330416,
      "loss": 4.7215,
      "step": 17584
    },
    {
      "epoch": 2.408244316625582,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005181895236501785,
      "loss": 4.7749,
      "step": 17585
    },
    {
      "epoch": 2.408381265406738,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005181490310022126,
      "loss": 4.7799,
      "step": 17586
    },
    {
      "epoch": 2.4085182141878936,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005181085377894796,
      "loss": 4.7639,
      "step": 17587
    },
    {
      "epoch": 2.4086551629690494,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005180680440123147,
      "loss": 4.6576,
      "step": 17588
    },
    {
      "epoch": 2.4087921117502056,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000518027549671054,
      "loss": 4.703,
      "step": 17589
    },
    {
      "epoch": 2.4089290605313614,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005179870547660326,
      "loss": 4.7446,
      "step": 17590
    },
    {
      "epoch": 2.409066009312517,
      "grad_norm": 0.375,
      "learning_rate": 0.0005179465592975862,
      "loss": 4.7611,
      "step": 17591
    },
    {
      "epoch": 2.409202958093673,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005179060632660507,
      "loss": 4.6849,
      "step": 17592
    },
    {
      "epoch": 2.4093399068748287,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0005178655666717612,
      "loss": 4.8349,
      "step": 17593
    },
    {
      "epoch": 2.409476855655985,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005178250695150537,
      "loss": 4.7118,
      "step": 17594
    },
    {
      "epoch": 2.4096138044371407,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0005177845717962636,
      "loss": 4.7137,
      "step": 17595
    },
    {
      "epoch": 2.4097507532182965,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005177440735157268,
      "loss": 4.7642,
      "step": 17596
    },
    {
      "epoch": 2.4098877019994522,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005177035746737786,
      "loss": 4.7213,
      "step": 17597
    },
    {
      "epoch": 2.410024650780608,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005176630752707548,
      "loss": 4.6514,
      "step": 17598
    },
    {
      "epoch": 2.410161599561764,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005176225753069909,
      "loss": 4.6767,
      "step": 17599
    },
    {
      "epoch": 2.4102985483429196,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005175820747828227,
      "loss": 4.7087,
      "step": 17600
    },
    {
      "epoch": 2.410435497124076,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005175415736985857,
      "loss": 4.709,
      "step": 17601
    },
    {
      "epoch": 2.4105724459052316,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005175010720546155,
      "loss": 4.7489,
      "step": 17602
    },
    {
      "epoch": 2.4107093946863873,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005174605698512479,
      "loss": 4.7042,
      "step": 17603
    },
    {
      "epoch": 2.410846343467543,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005174200670888185,
      "loss": 4.7063,
      "step": 17604
    },
    {
      "epoch": 2.410983292248699,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0005173795637676629,
      "loss": 4.7202,
      "step": 17605
    },
    {
      "epoch": 2.4111202410298547,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005173390598881167,
      "loss": 4.6519,
      "step": 17606
    },
    {
      "epoch": 2.4112571898110104,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005172985554505159,
      "loss": 4.7061,
      "step": 17607
    },
    {
      "epoch": 2.4113941385921667,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005172580504551958,
      "loss": 4.7455,
      "step": 17608
    },
    {
      "epoch": 2.4115310873733224,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005172175449024923,
      "loss": 4.7346,
      "step": 17609
    },
    {
      "epoch": 2.4116680361544782,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005171770387927409,
      "loss": 4.7303,
      "step": 17610
    },
    {
      "epoch": 2.411804984935634,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005171365321262774,
      "loss": 4.7378,
      "step": 17611
    },
    {
      "epoch": 2.4119419337167898,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005170960249034374,
      "loss": 4.7578,
      "step": 17612
    },
    {
      "epoch": 2.412078882497946,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005170555171245567,
      "loss": 4.7523,
      "step": 17613
    },
    {
      "epoch": 2.4122158312791018,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005170150087899709,
      "loss": 4.6829,
      "step": 17614
    },
    {
      "epoch": 2.4123527800602576,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005169744999000158,
      "loss": 4.7295,
      "step": 17615
    },
    {
      "epoch": 2.4124897288414133,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005169339904550271,
      "loss": 4.7565,
      "step": 17616
    },
    {
      "epoch": 2.412626677622569,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005168934804553405,
      "loss": 4.7001,
      "step": 17617
    },
    {
      "epoch": 2.412763626403725,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005168529699012916,
      "loss": 4.7919,
      "step": 17618
    },
    {
      "epoch": 2.4129005751848807,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005168124587932162,
      "loss": 4.6901,
      "step": 17619
    },
    {
      "epoch": 2.413037523966037,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005167719471314502,
      "loss": 4.7289,
      "step": 17620
    },
    {
      "epoch": 2.4131744727471927,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005167314349163288,
      "loss": 4.6621,
      "step": 17621
    },
    {
      "epoch": 2.4133114215283484,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005166909221481884,
      "loss": 4.7284,
      "step": 17622
    },
    {
      "epoch": 2.413448370309504,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005166504088273644,
      "loss": 4.7396,
      "step": 17623
    },
    {
      "epoch": 2.41358531909066,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005166098949541925,
      "loss": 4.6611,
      "step": 17624
    },
    {
      "epoch": 2.4137222678718158,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005165693805290086,
      "loss": 4.7191,
      "step": 17625
    },
    {
      "epoch": 2.4138592166529715,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005165288655521483,
      "loss": 4.7246,
      "step": 17626
    },
    {
      "epoch": 2.4139961654341278,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005164883500239475,
      "loss": 4.7721,
      "step": 17627
    },
    {
      "epoch": 2.4141331142152835,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005164478339447419,
      "loss": 4.7674,
      "step": 17628
    },
    {
      "epoch": 2.4142700629964393,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005164073173148674,
      "loss": 4.7669,
      "step": 17629
    },
    {
      "epoch": 2.414407011777595,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005163668001346595,
      "loss": 4.8426,
      "step": 17630
    },
    {
      "epoch": 2.414543960558751,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005163262824044541,
      "loss": 4.7146,
      "step": 17631
    },
    {
      "epoch": 2.414680909339907,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005162857641245869,
      "loss": 4.7206,
      "step": 17632
    },
    {
      "epoch": 2.414817858121063,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005162452452953938,
      "loss": 4.6464,
      "step": 17633
    },
    {
      "epoch": 2.4149548069022186,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005162047259172108,
      "loss": 4.746,
      "step": 17634
    },
    {
      "epoch": 2.4150917556833744,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005161642059903731,
      "loss": 4.7027,
      "step": 17635
    },
    {
      "epoch": 2.41522870446453,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005161236855152172,
      "loss": 4.6221,
      "step": 17636
    },
    {
      "epoch": 2.415365653245686,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005160831644920785,
      "loss": 4.7376,
      "step": 17637
    },
    {
      "epoch": 2.4155026020268417,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005160426429212928,
      "loss": 4.6455,
      "step": 17638
    },
    {
      "epoch": 2.415639550807998,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005160021208031959,
      "loss": 4.756,
      "step": 17639
    },
    {
      "epoch": 2.4157764995891537,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005159615981381238,
      "loss": 4.761,
      "step": 17640
    },
    {
      "epoch": 2.4159134483703095,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005159210749264122,
      "loss": 4.8014,
      "step": 17641
    },
    {
      "epoch": 2.4160503971514653,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005158805511683969,
      "loss": 4.7462,
      "step": 17642
    },
    {
      "epoch": 2.416187345932621,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005158400268644138,
      "loss": 4.7531,
      "step": 17643
    },
    {
      "epoch": 2.416324294713777,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005157995020147987,
      "loss": 4.687,
      "step": 17644
    },
    {
      "epoch": 2.416461243494933,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005157589766198874,
      "loss": 4.7518,
      "step": 17645
    },
    {
      "epoch": 2.416598192276089,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000515718450680016,
      "loss": 4.7209,
      "step": 17646
    },
    {
      "epoch": 2.4167351410572446,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00051567792419552,
      "loss": 4.7442,
      "step": 17647
    },
    {
      "epoch": 2.4168720898384004,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005156373971667354,
      "loss": 4.6709,
      "step": 17648
    },
    {
      "epoch": 2.417009038619556,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005155968695939979,
      "loss": 4.7506,
      "step": 17649
    },
    {
      "epoch": 2.417145987400712,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005155563414776437,
      "loss": 4.6891,
      "step": 17650
    },
    {
      "epoch": 2.417282936181868,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005155158128180085,
      "loss": 4.7258,
      "step": 17651
    },
    {
      "epoch": 2.417419884963024,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005154752836154281,
      "loss": 4.7092,
      "step": 17652
    },
    {
      "epoch": 2.4175568337441797,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005154347538702382,
      "loss": 4.7049,
      "step": 17653
    },
    {
      "epoch": 2.4176937825253355,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005153942235827752,
      "loss": 4.6754,
      "step": 17654
    },
    {
      "epoch": 2.4178307313064913,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005153536927533745,
      "loss": 4.5898,
      "step": 17655
    },
    {
      "epoch": 2.417967680087647,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005153131613823724,
      "loss": 4.6911,
      "step": 17656
    },
    {
      "epoch": 2.418104628868803,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005152726294701044,
      "loss": 4.665,
      "step": 17657
    },
    {
      "epoch": 2.418241577649959,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005152320970169066,
      "loss": 4.6777,
      "step": 17658
    },
    {
      "epoch": 2.418378526431115,
      "grad_norm": 0.375,
      "learning_rate": 0.0005151915640231148,
      "loss": 4.7127,
      "step": 17659
    },
    {
      "epoch": 2.4185154752122706,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005151510304890651,
      "loss": 4.6633,
      "step": 17660
    },
    {
      "epoch": 2.4186524239934264,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005151104964150932,
      "loss": 4.6702,
      "step": 17661
    },
    {
      "epoch": 2.418789372774582,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005150699618015352,
      "loss": 4.7275,
      "step": 17662
    },
    {
      "epoch": 2.4189263215557384,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005150294266487268,
      "loss": 4.7613,
      "step": 17663
    },
    {
      "epoch": 2.419063270336894,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000514988890957004,
      "loss": 4.7483,
      "step": 17664
    },
    {
      "epoch": 2.41920021911805,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000514948354726703,
      "loss": 4.7659,
      "step": 17665
    },
    {
      "epoch": 2.4193371678992057,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005149078179581592,
      "loss": 4.6811,
      "step": 17666
    },
    {
      "epoch": 2.4194741166803615,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005148672806517091,
      "loss": 4.8099,
      "step": 17667
    },
    {
      "epoch": 2.4196110654615173,
      "grad_norm": 0.375,
      "learning_rate": 0.0005148267428076884,
      "loss": 4.6372,
      "step": 17668
    },
    {
      "epoch": 2.419748014242673,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000514786204426433,
      "loss": 4.7375,
      "step": 17669
    },
    {
      "epoch": 2.4198849630238293,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005147456655082786,
      "loss": 4.7381,
      "step": 17670
    },
    {
      "epoch": 2.420021911804985,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005147051260535618,
      "loss": 4.6559,
      "step": 17671
    },
    {
      "epoch": 2.420158860586141,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0005146645860626181,
      "loss": 4.6525,
      "step": 17672
    },
    {
      "epoch": 2.4202958093672966,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0005146240455357835,
      "loss": 4.7384,
      "step": 17673
    },
    {
      "epoch": 2.4204327581484524,
      "grad_norm": 0.435546875,
      "learning_rate": 0.000514583504473394,
      "loss": 4.7243,
      "step": 17674
    },
    {
      "epoch": 2.420569706929608,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005145429628757857,
      "loss": 4.7812,
      "step": 17675
    },
    {
      "epoch": 2.420706655710764,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005145024207432946,
      "loss": 4.7772,
      "step": 17676
    },
    {
      "epoch": 2.42084360449192,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005144618780762564,
      "loss": 4.7944,
      "step": 17677
    },
    {
      "epoch": 2.420980553273076,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005144213348750073,
      "loss": 4.6795,
      "step": 17678
    },
    {
      "epoch": 2.4211175020542317,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005143807911398831,
      "loss": 4.6812,
      "step": 17679
    },
    {
      "epoch": 2.4212544508353875,
      "grad_norm": 0.375,
      "learning_rate": 0.0005143402468712201,
      "loss": 4.7643,
      "step": 17680
    },
    {
      "epoch": 2.4213913996165433,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005142997020693542,
      "loss": 4.6855,
      "step": 17681
    },
    {
      "epoch": 2.4215283483976995,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005142591567346213,
      "loss": 4.6869,
      "step": 17682
    },
    {
      "epoch": 2.4216652971788553,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005142186108673574,
      "loss": 4.7459,
      "step": 17683
    },
    {
      "epoch": 2.421802245960011,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005141780644678986,
      "loss": 4.792,
      "step": 17684
    },
    {
      "epoch": 2.421939194741167,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005141375175365809,
      "loss": 4.7118,
      "step": 17685
    },
    {
      "epoch": 2.4220761435223226,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005140969700737402,
      "loss": 4.7019,
      "step": 17686
    },
    {
      "epoch": 2.4222130923034784,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005140564220797128,
      "loss": 4.6984,
      "step": 17687
    },
    {
      "epoch": 2.422350041084634,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005140158735548345,
      "loss": 4.7374,
      "step": 17688
    },
    {
      "epoch": 2.4224869898657904,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005139753244994414,
      "loss": 4.8034,
      "step": 17689
    },
    {
      "epoch": 2.422623938646946,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005139347749138697,
      "loss": 4.7389,
      "step": 17690
    },
    {
      "epoch": 2.422760887428102,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000513894224798455,
      "loss": 4.718,
      "step": 17691
    },
    {
      "epoch": 2.4228978362092577,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005138536741535339,
      "loss": 4.7206,
      "step": 17692
    },
    {
      "epoch": 2.4230347849904135,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005138131229794419,
      "loss": 4.7313,
      "step": 17693
    },
    {
      "epoch": 2.4231717337715692,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005137725712765156,
      "loss": 4.6854,
      "step": 17694
    },
    {
      "epoch": 2.4233086825527255,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005137320190450905,
      "loss": 4.7307,
      "step": 17695
    },
    {
      "epoch": 2.4234456313338812,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005136914662855033,
      "loss": 4.717,
      "step": 17696
    },
    {
      "epoch": 2.423582580115037,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005136509129980896,
      "loss": 4.6635,
      "step": 17697
    },
    {
      "epoch": 2.423719528896193,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005136103591831855,
      "loss": 4.7006,
      "step": 17698
    },
    {
      "epoch": 2.4238564776773486,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005135698048411274,
      "loss": 4.743,
      "step": 17699
    },
    {
      "epoch": 2.4239934264585044,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005135292499722511,
      "loss": 4.6648,
      "step": 17700
    },
    {
      "epoch": 2.4241303752396606,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0005134886945768928,
      "loss": 4.7091,
      "step": 17701
    },
    {
      "epoch": 2.4242673240208163,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005134481386553885,
      "loss": 4.6701,
      "step": 17702
    },
    {
      "epoch": 2.424404272801972,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0005134075822080745,
      "loss": 4.6879,
      "step": 17703
    },
    {
      "epoch": 2.424541221583128,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005133670252352866,
      "loss": 4.7476,
      "step": 17704
    },
    {
      "epoch": 2.4246781703642837,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000513326467737361,
      "loss": 4.7262,
      "step": 17705
    },
    {
      "epoch": 2.4248151191454395,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005132859097146341,
      "loss": 4.674,
      "step": 17706
    },
    {
      "epoch": 2.4249520679265952,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0005132453511674415,
      "loss": 4.6952,
      "step": 17707
    },
    {
      "epoch": 2.4250890167077515,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005132047920961199,
      "loss": 4.7062,
      "step": 17708
    },
    {
      "epoch": 2.4252259654889072,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000513164232501005,
      "loss": 4.7723,
      "step": 17709
    },
    {
      "epoch": 2.425362914270063,
      "grad_norm": 0.375,
      "learning_rate": 0.000513123672382433,
      "loss": 4.6619,
      "step": 17710
    },
    {
      "epoch": 2.425499863051219,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005130831117407401,
      "loss": 4.6851,
      "step": 17711
    },
    {
      "epoch": 2.4256368118323746,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0005130425505762625,
      "loss": 4.7382,
      "step": 17712
    },
    {
      "epoch": 2.425773760613531,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005130019888893363,
      "loss": 4.6928,
      "step": 17713
    },
    {
      "epoch": 2.4259107093946866,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005129614266802974,
      "loss": 4.6419,
      "step": 17714
    },
    {
      "epoch": 2.4260476581758423,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005129208639494824,
      "loss": 4.7288,
      "step": 17715
    },
    {
      "epoch": 2.426184606956998,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000512880300697227,
      "loss": 4.7487,
      "step": 17716
    },
    {
      "epoch": 2.426321555738154,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005128397369238677,
      "loss": 4.7576,
      "step": 17717
    },
    {
      "epoch": 2.4264585045193097,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005127991726297405,
      "loss": 4.6465,
      "step": 17718
    },
    {
      "epoch": 2.4265954533004654,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005127586078151814,
      "loss": 4.7021,
      "step": 17719
    },
    {
      "epoch": 2.4267324020816217,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000512718042480527,
      "loss": 4.7041,
      "step": 17720
    },
    {
      "epoch": 2.4268693508627774,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0005126774766261132,
      "loss": 4.718,
      "step": 17721
    },
    {
      "epoch": 2.427006299643933,
      "grad_norm": 0.375,
      "learning_rate": 0.0005126369102522762,
      "loss": 4.7541,
      "step": 17722
    },
    {
      "epoch": 2.427143248425089,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005125963433593522,
      "loss": 4.7336,
      "step": 17723
    },
    {
      "epoch": 2.4272801972062448,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005125557759476772,
      "loss": 4.7786,
      "step": 17724
    },
    {
      "epoch": 2.4274171459874005,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005125152080175878,
      "loss": 4.7364,
      "step": 17725
    },
    {
      "epoch": 2.4275540947685563,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00051247463956942,
      "loss": 4.6818,
      "step": 17726
    },
    {
      "epoch": 2.4276910435497125,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005124340706035099,
      "loss": 4.7186,
      "step": 17727
    },
    {
      "epoch": 2.4278279923308683,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005123935011201937,
      "loss": 4.6731,
      "step": 17728
    },
    {
      "epoch": 2.427964941112024,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005123529311198079,
      "loss": 4.7037,
      "step": 17729
    },
    {
      "epoch": 2.42810188989318,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005123123606026882,
      "loss": 4.7617,
      "step": 17730
    },
    {
      "epoch": 2.4282388386743357,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005122717895691713,
      "loss": 4.7431,
      "step": 17731
    },
    {
      "epoch": 2.428375787455492,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005122312180195933,
      "loss": 4.699,
      "step": 17732
    },
    {
      "epoch": 2.4285127362366477,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005121906459542902,
      "loss": 4.7462,
      "step": 17733
    },
    {
      "epoch": 2.4286496850178034,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005121500733735986,
      "loss": 4.6604,
      "step": 17734
    },
    {
      "epoch": 2.428786633798959,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005121095002778544,
      "loss": 4.7716,
      "step": 17735
    },
    {
      "epoch": 2.428923582580115,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0005120689266673939,
      "loss": 4.6335,
      "step": 17736
    },
    {
      "epoch": 2.4290605313612708,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005120283525425536,
      "loss": 4.8412,
      "step": 17737
    },
    {
      "epoch": 2.4291974801424265,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005119877779036694,
      "loss": 4.7763,
      "step": 17738
    },
    {
      "epoch": 2.4293344289235828,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005119472027510779,
      "loss": 4.6667,
      "step": 17739
    },
    {
      "epoch": 2.4294713777047385,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005119066270851149,
      "loss": 4.7877,
      "step": 17740
    },
    {
      "epoch": 2.4296083264858943,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005118660509061172,
      "loss": 4.6693,
      "step": 17741
    },
    {
      "epoch": 2.42974527526705,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005118254742144207,
      "loss": 4.7189,
      "step": 17742
    },
    {
      "epoch": 2.429882224048206,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005117848970103618,
      "loss": 4.7307,
      "step": 17743
    },
    {
      "epoch": 2.4300191728293616,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005117443192942765,
      "loss": 4.7619,
      "step": 17744
    },
    {
      "epoch": 2.4301561216105174,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005117037410665015,
      "loss": 4.692,
      "step": 17745
    },
    {
      "epoch": 2.4302930703916736,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005116631623273729,
      "loss": 4.6928,
      "step": 17746
    },
    {
      "epoch": 2.4304300191728294,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000511622583077227,
      "loss": 4.6883,
      "step": 17747
    },
    {
      "epoch": 2.430566967953985,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005115820033164001,
      "loss": 4.6984,
      "step": 17748
    },
    {
      "epoch": 2.430703916735141,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005115414230452283,
      "loss": 4.704,
      "step": 17749
    },
    {
      "epoch": 2.4308408655162967,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005115008422640482,
      "loss": 4.658,
      "step": 17750
    },
    {
      "epoch": 2.430977814297453,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005114602609731959,
      "loss": 4.7551,
      "step": 17751
    },
    {
      "epoch": 2.4311147630786087,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005114196791730079,
      "loss": 4.7028,
      "step": 17752
    },
    {
      "epoch": 2.4312517118597645,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005113790968638202,
      "loss": 4.7311,
      "step": 17753
    },
    {
      "epoch": 2.4313886606409203,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005113385140459694,
      "loss": 4.6954,
      "step": 17754
    },
    {
      "epoch": 2.431525609422076,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005112979307197917,
      "loss": 4.6885,
      "step": 17755
    },
    {
      "epoch": 2.431662558203232,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005112573468856234,
      "loss": 4.759,
      "step": 17756
    },
    {
      "epoch": 2.4317995069843876,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000511216762543801,
      "loss": 4.802,
      "step": 17757
    },
    {
      "epoch": 2.431936455765544,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005111761776946605,
      "loss": 4.7861,
      "step": 17758
    },
    {
      "epoch": 2.4320734045466996,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0005111355923385386,
      "loss": 4.7065,
      "step": 17759
    },
    {
      "epoch": 2.4322103533278554,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005110950064757715,
      "loss": 4.7279,
      "step": 17760
    },
    {
      "epoch": 2.432347302109011,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005110544201066955,
      "loss": 4.7493,
      "step": 17761
    },
    {
      "epoch": 2.432484250890167,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005110138332316469,
      "loss": 4.6484,
      "step": 17762
    },
    {
      "epoch": 2.4326211996713227,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000510973245850962,
      "loss": 4.7379,
      "step": 17763
    },
    {
      "epoch": 2.432758148452479,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005109326579649776,
      "loss": 4.7186,
      "step": 17764
    },
    {
      "epoch": 2.4328950972336347,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005108920695740295,
      "loss": 4.7164,
      "step": 17765
    },
    {
      "epoch": 2.4330320460147905,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005108514806784544,
      "loss": 4.68,
      "step": 17766
    },
    {
      "epoch": 2.4331689947959463,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005108108912785885,
      "loss": 4.7601,
      "step": 17767
    },
    {
      "epoch": 2.433305943577102,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005107703013747684,
      "loss": 4.7079,
      "step": 17768
    },
    {
      "epoch": 2.433442892358258,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005107297109673302,
      "loss": 4.749,
      "step": 17769
    },
    {
      "epoch": 2.433579841139414,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005106891200566105,
      "loss": 4.7467,
      "step": 17770
    },
    {
      "epoch": 2.43371678992057,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005106485286429455,
      "loss": 4.7241,
      "step": 17771
    },
    {
      "epoch": 2.4338537387017256,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005106079367266719,
      "loss": 4.6312,
      "step": 17772
    },
    {
      "epoch": 2.4339906874828814,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005105673443081257,
      "loss": 4.758,
      "step": 17773
    },
    {
      "epoch": 2.434127636264037,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005105267513876434,
      "loss": 4.7614,
      "step": 17774
    },
    {
      "epoch": 2.434264585045193,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005104861579655617,
      "loss": 4.7074,
      "step": 17775
    },
    {
      "epoch": 2.4344015338263487,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005104455640422167,
      "loss": 4.7315,
      "step": 17776
    },
    {
      "epoch": 2.434538482607505,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000510404969617945,
      "loss": 4.6681,
      "step": 17777
    },
    {
      "epoch": 2.4346754313886607,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005103643746930827,
      "loss": 4.6878,
      "step": 17778
    },
    {
      "epoch": 2.4348123801698165,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005103237792679667,
      "loss": 4.7713,
      "step": 17779
    },
    {
      "epoch": 2.4349493289509723,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005102831833429331,
      "loss": 4.6593,
      "step": 17780
    },
    {
      "epoch": 2.435086277732128,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005102425869183183,
      "loss": 4.6955,
      "step": 17781
    },
    {
      "epoch": 2.4352232265132843,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005102019899944589,
      "loss": 4.7343,
      "step": 17782
    },
    {
      "epoch": 2.43536017529444,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005101613925716912,
      "loss": 4.6463,
      "step": 17783
    },
    {
      "epoch": 2.435497124075596,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005101207946503518,
      "loss": 4.7316,
      "step": 17784
    },
    {
      "epoch": 2.4356340728567516,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000510080196230777,
      "loss": 4.608,
      "step": 17785
    },
    {
      "epoch": 2.4357710216379074,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005100395973133032,
      "loss": 4.7423,
      "step": 17786
    },
    {
      "epoch": 2.435907970419063,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005099989978982671,
      "loss": 4.676,
      "step": 17787
    },
    {
      "epoch": 2.436044919200219,
      "grad_norm": 0.34375,
      "learning_rate": 0.000509958397986005,
      "loss": 4.7072,
      "step": 17788
    },
    {
      "epoch": 2.436181867981375,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005099177975768532,
      "loss": 4.6826,
      "step": 17789
    },
    {
      "epoch": 2.436318816762531,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005098771966711485,
      "loss": 4.6954,
      "step": 17790
    },
    {
      "epoch": 2.4364557655436867,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005098365952692271,
      "loss": 4.6509,
      "step": 17791
    },
    {
      "epoch": 2.4365927143248425,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005097959933714257,
      "loss": 4.6864,
      "step": 17792
    },
    {
      "epoch": 2.4367296631059983,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005097553909780806,
      "loss": 4.643,
      "step": 17793
    },
    {
      "epoch": 2.436866611887154,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005097147880895282,
      "loss": 4.6665,
      "step": 17794
    },
    {
      "epoch": 2.43700356066831,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005096741847061053,
      "loss": 4.6511,
      "step": 17795
    },
    {
      "epoch": 2.437140509449466,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005096335808281482,
      "loss": 4.6828,
      "step": 17796
    },
    {
      "epoch": 2.437277458230622,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005095929764559933,
      "loss": 4.6391,
      "step": 17797
    },
    {
      "epoch": 2.4374144070117776,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005095523715899772,
      "loss": 4.7779,
      "step": 17798
    },
    {
      "epoch": 2.4375513557929334,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005095117662304366,
      "loss": 4.7173,
      "step": 17799
    },
    {
      "epoch": 2.437688304574089,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005094711603777076,
      "loss": 4.6602,
      "step": 17800
    },
    {
      "epoch": 2.4378252533552454,
      "grad_norm": 0.328125,
      "learning_rate": 0.000509430554032127,
      "loss": 4.8055,
      "step": 17801
    },
    {
      "epoch": 2.437962202136401,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005093899471940313,
      "loss": 4.7243,
      "step": 17802
    },
    {
      "epoch": 2.438099150917557,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005093493398637568,
      "loss": 4.7645,
      "step": 17803
    },
    {
      "epoch": 2.4382360996987127,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005093087320416403,
      "loss": 4.8022,
      "step": 17804
    },
    {
      "epoch": 2.4383730484798685,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005092681237280182,
      "loss": 4.6241,
      "step": 17805
    },
    {
      "epoch": 2.4385099972610242,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005092275149232271,
      "loss": 4.7332,
      "step": 17806
    },
    {
      "epoch": 2.43864694604218,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005091869056276033,
      "loss": 4.7759,
      "step": 17807
    },
    {
      "epoch": 2.4387838948233362,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005091462958414837,
      "loss": 4.7162,
      "step": 17808
    },
    {
      "epoch": 2.438920843604492,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005091056855652045,
      "loss": 4.7811,
      "step": 17809
    },
    {
      "epoch": 2.439057792385648,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005090650747991023,
      "loss": 4.6512,
      "step": 17810
    },
    {
      "epoch": 2.4391947411668036,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005090244635435141,
      "loss": 4.7169,
      "step": 17811
    },
    {
      "epoch": 2.4393316899479593,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005089838517987758,
      "loss": 4.7233,
      "step": 17812
    },
    {
      "epoch": 2.439468638729115,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005089432395652245,
      "loss": 4.7552,
      "step": 17813
    },
    {
      "epoch": 2.4396055875102713,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005089026268431964,
      "loss": 4.7037,
      "step": 17814
    },
    {
      "epoch": 2.439742536291427,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005088620136330281,
      "loss": 4.655,
      "step": 17815
    },
    {
      "epoch": 2.439879485072583,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005088213999350563,
      "loss": 4.6567,
      "step": 17816
    },
    {
      "epoch": 2.4400164338537387,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005087807857496177,
      "loss": 4.7479,
      "step": 17817
    },
    {
      "epoch": 2.4401533826348945,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005087401710770486,
      "loss": 4.7145,
      "step": 17818
    },
    {
      "epoch": 2.4402903314160502,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005086995559176857,
      "loss": 4.756,
      "step": 17819
    },
    {
      "epoch": 2.4404272801972064,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005086589402718656,
      "loss": 4.6782,
      "step": 17820
    },
    {
      "epoch": 2.4405642289783622,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005086183241399249,
      "loss": 4.5823,
      "step": 17821
    },
    {
      "epoch": 2.440701177759518,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005085777075222003,
      "loss": 4.6147,
      "step": 17822
    },
    {
      "epoch": 2.440838126540674,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005085370904190282,
      "loss": 4.7294,
      "step": 17823
    },
    {
      "epoch": 2.4409750753218296,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005084964728307453,
      "loss": 4.7346,
      "step": 17824
    },
    {
      "epoch": 2.4411120241029853,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000508455854757688,
      "loss": 4.7055,
      "step": 17825
    },
    {
      "epoch": 2.441248972884141,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005084152362001933,
      "loss": 4.7153,
      "step": 17826
    },
    {
      "epoch": 2.4413859216652973,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005083746171585976,
      "loss": 4.7424,
      "step": 17827
    },
    {
      "epoch": 2.441522870446453,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005083339976332374,
      "loss": 4.7722,
      "step": 17828
    },
    {
      "epoch": 2.441659819227609,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005082933776244496,
      "loss": 4.7164,
      "step": 17829
    },
    {
      "epoch": 2.4417967680087647,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005082527571325706,
      "loss": 4.6945,
      "step": 17830
    },
    {
      "epoch": 2.4419337167899204,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005082121361579373,
      "loss": 4.6741,
      "step": 17831
    },
    {
      "epoch": 2.4420706655710767,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005081715147008859,
      "loss": 4.6937,
      "step": 17832
    },
    {
      "epoch": 2.4422076143522324,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005081308927617535,
      "loss": 4.7561,
      "step": 17833
    },
    {
      "epoch": 2.442344563133388,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005080902703408764,
      "loss": 4.6351,
      "step": 17834
    },
    {
      "epoch": 2.442481511914544,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0005080496474385916,
      "loss": 4.7012,
      "step": 17835
    },
    {
      "epoch": 2.4426184606956998,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005080090240552354,
      "loss": 4.7337,
      "step": 17836
    },
    {
      "epoch": 2.4427554094768555,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005079684001911445,
      "loss": 4.6861,
      "step": 17837
    },
    {
      "epoch": 2.4428923582580113,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005079277758466557,
      "loss": 4.7363,
      "step": 17838
    },
    {
      "epoch": 2.4430293070391675,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005078871510221057,
      "loss": 4.6897,
      "step": 17839
    },
    {
      "epoch": 2.4431662558203233,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000507846525717831,
      "loss": 4.7062,
      "step": 17840
    },
    {
      "epoch": 2.443303204601479,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005078058999341684,
      "loss": 4.8191,
      "step": 17841
    },
    {
      "epoch": 2.443440153382635,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0005077652736714545,
      "loss": 4.6504,
      "step": 17842
    },
    {
      "epoch": 2.4435771021637906,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0005077246469300261,
      "loss": 4.7106,
      "step": 17843
    },
    {
      "epoch": 2.4437140509449464,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005076840197102198,
      "loss": 4.7386,
      "step": 17844
    },
    {
      "epoch": 2.443850999726102,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005076433920123721,
      "loss": 4.7174,
      "step": 17845
    },
    {
      "epoch": 2.4439879485072584,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005076027638368201,
      "loss": 4.6631,
      "step": 17846
    },
    {
      "epoch": 2.444124897288414,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005075621351839001,
      "loss": 4.7118,
      "step": 17847
    },
    {
      "epoch": 2.44426184606957,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005075215060539491,
      "loss": 4.7696,
      "step": 17848
    },
    {
      "epoch": 2.4443987948507258,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005074808764473036,
      "loss": 4.6113,
      "step": 17849
    },
    {
      "epoch": 2.4445357436318815,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005074402463643003,
      "loss": 4.7143,
      "step": 17850
    },
    {
      "epoch": 2.4446726924130378,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005073996158052762,
      "loss": 4.7339,
      "step": 17851
    },
    {
      "epoch": 2.4448096411941935,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005073589847705677,
      "loss": 4.759,
      "step": 17852
    },
    {
      "epoch": 2.4449465899753493,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005073183532605116,
      "loss": 4.6892,
      "step": 17853
    },
    {
      "epoch": 2.445083538756505,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005072777212754446,
      "loss": 4.6685,
      "step": 17854
    },
    {
      "epoch": 2.445220487537661,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005072370888157036,
      "loss": 4.7267,
      "step": 17855
    },
    {
      "epoch": 2.4453574363188166,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000507196455881625,
      "loss": 4.7216,
      "step": 17856
    },
    {
      "epoch": 2.4454943850999724,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000507155822473546,
      "loss": 4.6744,
      "step": 17857
    },
    {
      "epoch": 2.4456313338811286,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005071151885918029,
      "loss": 4.7958,
      "step": 17858
    },
    {
      "epoch": 2.4457682826622844,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005070745542367327,
      "loss": 4.7042,
      "step": 17859
    },
    {
      "epoch": 2.44590523144344,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005070339194086721,
      "loss": 4.7541,
      "step": 17860
    },
    {
      "epoch": 2.446042180224596,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005069932841079577,
      "loss": 4.7242,
      "step": 17861
    },
    {
      "epoch": 2.4461791290057517,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005069526483349265,
      "loss": 4.698,
      "step": 17862
    },
    {
      "epoch": 2.4463160777869075,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005069120120899151,
      "loss": 4.6732,
      "step": 17863
    },
    {
      "epoch": 2.4464530265680637,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005068713753732604,
      "loss": 4.627,
      "step": 17864
    },
    {
      "epoch": 2.4465899753492195,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005068307381852988,
      "loss": 4.7581,
      "step": 17865
    },
    {
      "epoch": 2.4467269241303753,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005067901005263674,
      "loss": 4.7603,
      "step": 17866
    },
    {
      "epoch": 2.446863872911531,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000506749462396803,
      "loss": 4.7285,
      "step": 17867
    },
    {
      "epoch": 2.447000821692687,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005067088237969423,
      "loss": 4.6163,
      "step": 17868
    },
    {
      "epoch": 2.4471377704738426,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000506668184727122,
      "loss": 4.6722,
      "step": 17869
    },
    {
      "epoch": 2.447274719254999,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005066275451876789,
      "loss": 4.7813,
      "step": 17870
    },
    {
      "epoch": 2.4474116680361546,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00050658690517895,
      "loss": 4.6634,
      "step": 17871
    },
    {
      "epoch": 2.4475486168173104,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005065462647012718,
      "loss": 4.6938,
      "step": 17872
    },
    {
      "epoch": 2.447685565598466,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005065056237549813,
      "loss": 4.7434,
      "step": 17873
    },
    {
      "epoch": 2.447822514379622,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005064649823404152,
      "loss": 4.7357,
      "step": 17874
    },
    {
      "epoch": 2.4479594631607777,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005064243404579102,
      "loss": 4.717,
      "step": 17875
    },
    {
      "epoch": 2.4480964119419335,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005063836981078036,
      "loss": 4.7421,
      "step": 17876
    },
    {
      "epoch": 2.4482333607230897,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005063430552904315,
      "loss": 4.7111,
      "step": 17877
    },
    {
      "epoch": 2.4483703095042455,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005063024120061314,
      "loss": 4.6244,
      "step": 17878
    },
    {
      "epoch": 2.4485072582854013,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005062617682552395,
      "loss": 4.7284,
      "step": 17879
    },
    {
      "epoch": 2.448644207066557,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005062211240380931,
      "loss": 4.7216,
      "step": 17880
    },
    {
      "epoch": 2.448781155847713,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005061804793550287,
      "loss": 4.6998,
      "step": 17881
    },
    {
      "epoch": 2.448918104628869,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005061398342063834,
      "loss": 4.7734,
      "step": 17882
    },
    {
      "epoch": 2.449055053410025,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005060991885924938,
      "loss": 4.7323,
      "step": 17883
    },
    {
      "epoch": 2.4491920021911806,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005060585425136971,
      "loss": 4.7145,
      "step": 17884
    },
    {
      "epoch": 2.4493289509723364,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005060178959703296,
      "loss": 4.7719,
      "step": 17885
    },
    {
      "epoch": 2.449465899753492,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005059772489627285,
      "loss": 4.7181,
      "step": 17886
    },
    {
      "epoch": 2.449602848534648,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005059366014912309,
      "loss": 4.691,
      "step": 17887
    },
    {
      "epoch": 2.4497397973158037,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000505895953556173,
      "loss": 4.6849,
      "step": 17888
    },
    {
      "epoch": 2.44987674609696,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005058553051578923,
      "loss": 4.6918,
      "step": 17889
    },
    {
      "epoch": 2.4500136948781157,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005058146562967251,
      "loss": 4.7228,
      "step": 17890
    },
    {
      "epoch": 2.4501506436592715,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005057740069730087,
      "loss": 4.7555,
      "step": 17891
    },
    {
      "epoch": 2.4502875924404273,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005057333571870797,
      "loss": 4.6389,
      "step": 17892
    },
    {
      "epoch": 2.450424541221583,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005056927069392752,
      "loss": 4.7339,
      "step": 17893
    },
    {
      "epoch": 2.450561490002739,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0005056520562299319,
      "loss": 4.7032,
      "step": 17894
    },
    {
      "epoch": 2.4506984387838946,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005056114050593869,
      "loss": 4.6408,
      "step": 17895
    },
    {
      "epoch": 2.450835387565051,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005055707534279768,
      "loss": 4.7052,
      "step": 17896
    },
    {
      "epoch": 2.4509723363462066,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005055301013360386,
      "loss": 4.6557,
      "step": 17897
    },
    {
      "epoch": 2.4511092851273624,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005054894487839094,
      "loss": 4.5886,
      "step": 17898
    },
    {
      "epoch": 2.451246233908518,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005054487957719257,
      "loss": 4.6739,
      "step": 17899
    },
    {
      "epoch": 2.451383182689674,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005054081423004247,
      "loss": 4.7438,
      "step": 17900
    },
    {
      "epoch": 2.45152013147083,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0005053674883697431,
      "loss": 4.7149,
      "step": 17901
    },
    {
      "epoch": 2.451657080251986,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0005053268339802182,
      "loss": 4.6905,
      "step": 17902
    },
    {
      "epoch": 2.4517940290331417,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0005052861791321865,
      "loss": 4.6676,
      "step": 17903
    },
    {
      "epoch": 2.4519309778142975,
      "grad_norm": 0.390625,
      "learning_rate": 0.000505245523825985,
      "loss": 4.7596,
      "step": 17904
    },
    {
      "epoch": 2.4520679265954533,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0005052048680619508,
      "loss": 4.7874,
      "step": 17905
    },
    {
      "epoch": 2.452204875376609,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005051642118404205,
      "loss": 4.7503,
      "step": 17906
    },
    {
      "epoch": 2.452341824157765,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005051235551617315,
      "loss": 4.7815,
      "step": 17907
    },
    {
      "epoch": 2.452478772938921,
      "grad_norm": 0.40625,
      "learning_rate": 0.0005050828980262202,
      "loss": 4.6805,
      "step": 17908
    },
    {
      "epoch": 2.452615721720077,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000505042240434224,
      "loss": 4.6428,
      "step": 17909
    },
    {
      "epoch": 2.4527526705012326,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005050015823860795,
      "loss": 4.7953,
      "step": 17910
    },
    {
      "epoch": 2.4528896192823884,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000504960923882124,
      "loss": 4.7135,
      "step": 17911
    },
    {
      "epoch": 2.453026568063544,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000504920264922694,
      "loss": 4.63,
      "step": 17912
    },
    {
      "epoch": 2.4531635168447,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005048796055081267,
      "loss": 4.6318,
      "step": 17913
    },
    {
      "epoch": 2.4533004656258557,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005048389456387591,
      "loss": 4.6707,
      "step": 17914
    },
    {
      "epoch": 2.453437414407012,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000504798285314928,
      "loss": 4.7119,
      "step": 17915
    },
    {
      "epoch": 2.4535743631881677,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005047576245369706,
      "loss": 4.7999,
      "step": 17916
    },
    {
      "epoch": 2.4537113119693235,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005047169633052236,
      "loss": 4.7085,
      "step": 17917
    },
    {
      "epoch": 2.4538482607504792,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005046763016200242,
      "loss": 4.7726,
      "step": 17918
    },
    {
      "epoch": 2.453985209531635,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005046356394817091,
      "loss": 4.7691,
      "step": 17919
    },
    {
      "epoch": 2.4541221583127912,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005045949768906156,
      "loss": 4.6149,
      "step": 17920
    },
    {
      "epoch": 2.454259107093947,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005045543138470805,
      "loss": 4.7542,
      "step": 17921
    },
    {
      "epoch": 2.454396055875103,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005045136503514407,
      "loss": 4.6512,
      "step": 17922
    },
    {
      "epoch": 2.4545330046562586,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005044729864040334,
      "loss": 4.8165,
      "step": 17923
    },
    {
      "epoch": 2.4546699534374143,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005044323220051953,
      "loss": 4.5835,
      "step": 17924
    },
    {
      "epoch": 2.45480690221857,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005043916571552638,
      "loss": 4.6831,
      "step": 17925
    },
    {
      "epoch": 2.454943850999726,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005043509918545754,
      "loss": 4.7468,
      "step": 17926
    },
    {
      "epoch": 2.455080799780882,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0005043103261034677,
      "loss": 4.6343,
      "step": 17927
    },
    {
      "epoch": 2.455217748562038,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005042696599022773,
      "loss": 4.6378,
      "step": 17928
    },
    {
      "epoch": 2.4553546973431937,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005042289932513412,
      "loss": 4.7568,
      "step": 17929
    },
    {
      "epoch": 2.4554916461243494,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005041883261509964,
      "loss": 4.7564,
      "step": 17930
    },
    {
      "epoch": 2.4556285949055052,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005041476586015802,
      "loss": 4.7471,
      "step": 17931
    },
    {
      "epoch": 2.455765543686661,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005041069906034295,
      "loss": 4.7344,
      "step": 17932
    },
    {
      "epoch": 2.455902492467817,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005040663221568811,
      "loss": 4.7145,
      "step": 17933
    },
    {
      "epoch": 2.456039441248973,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005040256532622724,
      "loss": 4.6443,
      "step": 17934
    },
    {
      "epoch": 2.4561763900301288,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005039849839199401,
      "loss": 4.7241,
      "step": 17935
    },
    {
      "epoch": 2.4563133388112846,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005039443141302215,
      "loss": 4.7351,
      "step": 17936
    },
    {
      "epoch": 2.4564502875924403,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005039036438934533,
      "loss": 4.6442,
      "step": 17937
    },
    {
      "epoch": 2.456587236373596,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005038629732099729,
      "loss": 4.6462,
      "step": 17938
    },
    {
      "epoch": 2.4567241851547523,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005038223020801173,
      "loss": 4.8528,
      "step": 17939
    },
    {
      "epoch": 2.456861133935908,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005037816305042232,
      "loss": 4.8022,
      "step": 17940
    },
    {
      "epoch": 2.456998082717064,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005037409584826281,
      "loss": 4.7762,
      "step": 17941
    },
    {
      "epoch": 2.4571350314982197,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005037002860156688,
      "loss": 4.7443,
      "step": 17942
    },
    {
      "epoch": 2.4572719802793754,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005036596131036827,
      "loss": 4.6844,
      "step": 17943
    },
    {
      "epoch": 2.457408929060531,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005036189397470063,
      "loss": 4.6643,
      "step": 17944
    },
    {
      "epoch": 2.457545877841687,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005035782659459769,
      "loss": 4.7272,
      "step": 17945
    },
    {
      "epoch": 2.457682826622843,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0005035375917009318,
      "loss": 4.7242,
      "step": 17946
    },
    {
      "epoch": 2.457819775403999,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005034969170122079,
      "loss": 4.6015,
      "step": 17947
    },
    {
      "epoch": 2.4579567241851548,
      "grad_norm": 0.390625,
      "learning_rate": 0.0005034562418801424,
      "loss": 4.7143,
      "step": 17948
    },
    {
      "epoch": 2.4580936729663105,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005034155663050722,
      "loss": 4.6771,
      "step": 17949
    },
    {
      "epoch": 2.4582306217474663,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005033748902873344,
      "loss": 4.7962,
      "step": 17950
    },
    {
      "epoch": 2.4583675705286225,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005033342138272663,
      "loss": 4.744,
      "step": 17951
    },
    {
      "epoch": 2.4585045193097783,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005032935369252047,
      "loss": 4.7273,
      "step": 17952
    },
    {
      "epoch": 2.458641468090934,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005032528595814869,
      "loss": 4.6586,
      "step": 17953
    },
    {
      "epoch": 2.45877841687209,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00050321218179645,
      "loss": 4.7016,
      "step": 17954
    },
    {
      "epoch": 2.4589153656532456,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005031715035704312,
      "loss": 4.7557,
      "step": 17955
    },
    {
      "epoch": 2.4590523144344014,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005031308249037673,
      "loss": 4.7246,
      "step": 17956
    },
    {
      "epoch": 2.459189263215557,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005030901457967957,
      "loss": 4.6961,
      "step": 17957
    },
    {
      "epoch": 2.4593262119967134,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005030494662498533,
      "loss": 4.6617,
      "step": 17958
    },
    {
      "epoch": 2.459463160777869,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005030087862632775,
      "loss": 4.6597,
      "step": 17959
    },
    {
      "epoch": 2.459600109559025,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005029681058374052,
      "loss": 4.7782,
      "step": 17960
    },
    {
      "epoch": 2.4597370583401807,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005029274249725734,
      "loss": 4.7599,
      "step": 17961
    },
    {
      "epoch": 2.4598740071213365,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005028867436691196,
      "loss": 4.7091,
      "step": 17962
    },
    {
      "epoch": 2.4600109559024923,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005028460619273807,
      "loss": 4.7834,
      "step": 17963
    },
    {
      "epoch": 2.460147904683648,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005028053797476938,
      "loss": 4.699,
      "step": 17964
    },
    {
      "epoch": 2.4602848534648043,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005027646971303963,
      "loss": 4.6993,
      "step": 17965
    },
    {
      "epoch": 2.46042180224596,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005027240140758251,
      "loss": 4.6732,
      "step": 17966
    },
    {
      "epoch": 2.460558751027116,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005026833305843175,
      "loss": 4.7015,
      "step": 17967
    },
    {
      "epoch": 2.4606956998082716,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005026426466562105,
      "loss": 4.6948,
      "step": 17968
    },
    {
      "epoch": 2.4608326485894274,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005026019622918415,
      "loss": 4.7557,
      "step": 17969
    },
    {
      "epoch": 2.4609695973705836,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005025612774915473,
      "loss": 4.6942,
      "step": 17970
    },
    {
      "epoch": 2.4611065461517394,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0005025205922556654,
      "loss": 4.7262,
      "step": 17971
    },
    {
      "epoch": 2.461243494932895,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005024799065845327,
      "loss": 4.6917,
      "step": 17972
    },
    {
      "epoch": 2.461380443714051,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005024392204784866,
      "loss": 4.729,
      "step": 17973
    },
    {
      "epoch": 2.4615173924952067,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005023985339378642,
      "loss": 4.7722,
      "step": 17974
    },
    {
      "epoch": 2.4616543412763625,
      "grad_norm": 0.328125,
      "learning_rate": 0.0005023578469630027,
      "loss": 4.7493,
      "step": 17975
    },
    {
      "epoch": 2.4617912900575183,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005023171595542392,
      "loss": 4.7182,
      "step": 17976
    },
    {
      "epoch": 2.4619282388386745,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005022764717119108,
      "loss": 4.7518,
      "step": 17977
    },
    {
      "epoch": 2.4620651876198303,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000502235783436355,
      "loss": 4.783,
      "step": 17978
    },
    {
      "epoch": 2.462202136400986,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0005021950947279088,
      "loss": 4.6851,
      "step": 17979
    },
    {
      "epoch": 2.462339085182142,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005021544055869094,
      "loss": 4.6788,
      "step": 17980
    },
    {
      "epoch": 2.4624760339632976,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000502113716013694,
      "loss": 4.6736,
      "step": 17981
    },
    {
      "epoch": 2.4626129827444534,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005020730260085998,
      "loss": 4.6546,
      "step": 17982
    },
    {
      "epoch": 2.4627499315256096,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000502032335571964,
      "loss": 4.6566,
      "step": 17983
    },
    {
      "epoch": 2.4628868803067654,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005019916447041239,
      "loss": 4.6581,
      "step": 17984
    },
    {
      "epoch": 2.463023829087921,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005019509534054166,
      "loss": 4.7931,
      "step": 17985
    },
    {
      "epoch": 2.463160777869077,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005019102616761794,
      "loss": 4.7207,
      "step": 17986
    },
    {
      "epoch": 2.4632977266502327,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0005018695695167495,
      "loss": 4.6563,
      "step": 17987
    },
    {
      "epoch": 2.4634346754313885,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005018288769274642,
      "loss": 4.6712,
      "step": 17988
    },
    {
      "epoch": 2.4635716242125447,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005017881839086604,
      "loss": 4.6234,
      "step": 17989
    },
    {
      "epoch": 2.4637085729937005,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005017474904606757,
      "loss": 4.6333,
      "step": 17990
    },
    {
      "epoch": 2.4638455217748563,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005017067965838473,
      "loss": 4.8092,
      "step": 17991
    },
    {
      "epoch": 2.463982470556012,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005016661022785123,
      "loss": 4.7217,
      "step": 17992
    },
    {
      "epoch": 2.464119419337168,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000501625407545008,
      "loss": 4.6519,
      "step": 17993
    },
    {
      "epoch": 2.4642563681183236,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005015847123836716,
      "loss": 4.7406,
      "step": 17994
    },
    {
      "epoch": 2.4643933168994794,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0005015440167948404,
      "loss": 4.7302,
      "step": 17995
    },
    {
      "epoch": 2.4645302656806356,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005015033207788519,
      "loss": 4.7182,
      "step": 17996
    },
    {
      "epoch": 2.4646672144617914,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005014626243360428,
      "loss": 4.7613,
      "step": 17997
    },
    {
      "epoch": 2.464804163242947,
      "grad_norm": 0.34375,
      "learning_rate": 0.0005014219274667507,
      "loss": 4.6674,
      "step": 17998
    },
    {
      "epoch": 2.464941112024103,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000501381230171313,
      "loss": 4.7544,
      "step": 17999
    },
    {
      "epoch": 2.4650780608052587,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005013405324500666,
      "loss": 4.5072,
      "step": 18000
    },
    {
      "epoch": 2.465215009586415,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005012998343033491,
      "loss": 4.752,
      "step": 18001
    },
    {
      "epoch": 2.4653519583675707,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005012591357314976,
      "loss": 4.7282,
      "step": 18002
    },
    {
      "epoch": 2.4654889071487265,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005012184367348496,
      "loss": 4.6678,
      "step": 18003
    },
    {
      "epoch": 2.4656258559298823,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000501177737313742,
      "loss": 4.6884,
      "step": 18004
    },
    {
      "epoch": 2.465762804711038,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005011370374685124,
      "loss": 4.6954,
      "step": 18005
    },
    {
      "epoch": 2.465899753492194,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000501096337199498,
      "loss": 4.7862,
      "step": 18006
    },
    {
      "epoch": 2.4660367022733496,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0005010556365070359,
      "loss": 4.7385,
      "step": 18007
    },
    {
      "epoch": 2.466173651054506,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005010149353914638,
      "loss": 4.7528,
      "step": 18008
    },
    {
      "epoch": 2.4663105998356616,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0005009742338531186,
      "loss": 4.781,
      "step": 18009
    },
    {
      "epoch": 2.4664475486168174,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0005009335318923379,
      "loss": 4.7046,
      "step": 18010
    },
    {
      "epoch": 2.466584497397973,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005008928295094588,
      "loss": 4.7557,
      "step": 18011
    },
    {
      "epoch": 2.466721446179129,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0005008521267048187,
      "loss": 4.6836,
      "step": 18012
    },
    {
      "epoch": 2.4668583949602847,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0005008114234787549,
      "loss": 4.7341,
      "step": 18013
    },
    {
      "epoch": 2.4669953437414405,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0005007707198316049,
      "loss": 4.6878,
      "step": 18014
    },
    {
      "epoch": 2.4671322925225967,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0005007300157637057,
      "loss": 4.6933,
      "step": 18015
    },
    {
      "epoch": 2.4672692413037525,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0005006893112753946,
      "loss": 4.6652,
      "step": 18016
    },
    {
      "epoch": 2.4674061900849082,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0005006486063670094,
      "loss": 4.7744,
      "step": 18017
    },
    {
      "epoch": 2.467543138866064,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0005006079010388869,
      "loss": 4.7597,
      "step": 18018
    },
    {
      "epoch": 2.46768008764722,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0005005671952913649,
      "loss": 4.6554,
      "step": 18019
    },
    {
      "epoch": 2.467817036428376,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0005005264891247803,
      "loss": 4.722,
      "step": 18020
    },
    {
      "epoch": 2.467953985209532,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005004857825394708,
      "loss": 4.6953,
      "step": 18021
    },
    {
      "epoch": 2.4680909339906876,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0005004450755357735,
      "loss": 4.7288,
      "step": 18022
    },
    {
      "epoch": 2.4682278827718434,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0005004043681140258,
      "loss": 4.7042,
      "step": 18023
    },
    {
      "epoch": 2.468364831552999,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005003636602745651,
      "loss": 4.7177,
      "step": 18024
    },
    {
      "epoch": 2.468501780334155,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0005003229520177287,
      "loss": 4.745,
      "step": 18025
    },
    {
      "epoch": 2.4686387291153107,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005002822433438542,
      "loss": 4.6814,
      "step": 18026
    },
    {
      "epoch": 2.468775677896467,
      "grad_norm": 0.359375,
      "learning_rate": 0.0005002415342532787,
      "loss": 4.7082,
      "step": 18027
    },
    {
      "epoch": 2.4689126266776227,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0005002008247463396,
      "loss": 4.7364,
      "step": 18028
    },
    {
      "epoch": 2.4690495754587785,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0005001601148233743,
      "loss": 4.8,
      "step": 18029
    },
    {
      "epoch": 2.4691865242399342,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0005001194044847202,
      "loss": 4.6338,
      "step": 18030
    },
    {
      "epoch": 2.46932347302109,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005000786937307146,
      "loss": 4.6968,
      "step": 18031
    },
    {
      "epoch": 2.469460421802246,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0005000379825616949,
      "loss": 4.7375,
      "step": 18032
    },
    {
      "epoch": 2.4695973705834016,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004999972709779986,
      "loss": 4.7163,
      "step": 18033
    },
    {
      "epoch": 2.469734319364558,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000499956558979963,
      "loss": 4.6993,
      "step": 18034
    },
    {
      "epoch": 2.4698712681457136,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004999158465679255,
      "loss": 4.6907,
      "step": 18035
    },
    {
      "epoch": 2.4700082169268693,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004998751337422233,
      "loss": 4.6779,
      "step": 18036
    },
    {
      "epoch": 2.470145165708025,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004998344205031943,
      "loss": 4.8009,
      "step": 18037
    },
    {
      "epoch": 2.470282114489181,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004997937068511754,
      "loss": 4.7577,
      "step": 18038
    },
    {
      "epoch": 2.470419063270337,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004997529927865042,
      "loss": 4.736,
      "step": 18039
    },
    {
      "epoch": 2.470556012051493,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000499712278309518,
      "loss": 4.744,
      "step": 18040
    },
    {
      "epoch": 2.4706929608326487,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004996715634205546,
      "loss": 4.7257,
      "step": 18041
    },
    {
      "epoch": 2.4708299096138044,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004996308481199507,
      "loss": 4.7665,
      "step": 18042
    },
    {
      "epoch": 2.47096685839496,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004995901324080444,
      "loss": 4.7986,
      "step": 18043
    },
    {
      "epoch": 2.471103807176116,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004995494162851728,
      "loss": 4.6854,
      "step": 18044
    },
    {
      "epoch": 2.4712407559572718,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004995086997516733,
      "loss": 4.719,
      "step": 18045
    },
    {
      "epoch": 2.471377704738428,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004994679828078837,
      "loss": 4.6614,
      "step": 18046
    },
    {
      "epoch": 2.4715146535195838,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0004994272654541409,
      "loss": 4.6573,
      "step": 18047
    },
    {
      "epoch": 2.4716516023007395,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004993865476907826,
      "loss": 4.5841,
      "step": 18048
    },
    {
      "epoch": 2.4717885510818953,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0004993458295181462,
      "loss": 4.6918,
      "step": 18049
    },
    {
      "epoch": 2.471925499863051,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004993051109365692,
      "loss": 4.7617,
      "step": 18050
    },
    {
      "epoch": 2.472062448644207,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004992643919463889,
      "loss": 4.6848,
      "step": 18051
    },
    {
      "epoch": 2.472199397425363,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000499223672547943,
      "loss": 4.5802,
      "step": 18052
    },
    {
      "epoch": 2.472336346206519,
      "grad_norm": 0.375,
      "learning_rate": 0.0004991829527415687,
      "loss": 4.6708,
      "step": 18053
    },
    {
      "epoch": 2.4724732949876747,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004991422325276036,
      "loss": 4.6808,
      "step": 18054
    },
    {
      "epoch": 2.4726102437688304,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000499101511906385,
      "loss": 4.6992,
      "step": 18055
    },
    {
      "epoch": 2.472747192549986,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0004990607908782505,
      "loss": 4.6849,
      "step": 18056
    },
    {
      "epoch": 2.472884141331142,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004990200694435377,
      "loss": 4.7358,
      "step": 18057
    },
    {
      "epoch": 2.473021090112298,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004989793476025837,
      "loss": 4.6744,
      "step": 18058
    },
    {
      "epoch": 2.473158038893454,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004989386253557264,
      "loss": 4.7274,
      "step": 18059
    },
    {
      "epoch": 2.4732949876746098,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004988979027033028,
      "loss": 4.7242,
      "step": 18060
    },
    {
      "epoch": 2.4734319364557655,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004988571796456508,
      "loss": 4.6625,
      "step": 18061
    },
    {
      "epoch": 2.4735688852369213,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004988164561831076,
      "loss": 4.6359,
      "step": 18062
    },
    {
      "epoch": 2.473705834018077,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004987757323160108,
      "loss": 4.6974,
      "step": 18063
    },
    {
      "epoch": 2.473842782799233,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004987350080446979,
      "loss": 4.6213,
      "step": 18064
    },
    {
      "epoch": 2.473979731580389,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004986942833695064,
      "loss": 4.7808,
      "step": 18065
    },
    {
      "epoch": 2.474116680361545,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004986535582907738,
      "loss": 4.6027,
      "step": 18066
    },
    {
      "epoch": 2.4742536291427006,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004986128328088375,
      "loss": 4.7827,
      "step": 18067
    },
    {
      "epoch": 2.4743905779238564,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004985721069240351,
      "loss": 4.7006,
      "step": 18068
    },
    {
      "epoch": 2.474527526705012,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004985313806367041,
      "loss": 4.7248,
      "step": 18069
    },
    {
      "epoch": 2.4746644754861684,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000498490653947182,
      "loss": 4.6334,
      "step": 18070
    },
    {
      "epoch": 2.474801424267324,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004984499268558063,
      "loss": 4.757,
      "step": 18071
    },
    {
      "epoch": 2.47493837304848,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004984091993629143,
      "loss": 4.6725,
      "step": 18072
    },
    {
      "epoch": 2.4750753218296357,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004983684714688441,
      "loss": 4.7095,
      "step": 18073
    },
    {
      "epoch": 2.4752122706107915,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004983277431739327,
      "loss": 4.6874,
      "step": 18074
    },
    {
      "epoch": 2.4753492193919473,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004982870144785177,
      "loss": 4.7146,
      "step": 18075
    },
    {
      "epoch": 2.475486168173103,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004982462853829367,
      "loss": 4.6452,
      "step": 18076
    },
    {
      "epoch": 2.4756231169542593,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004982055558875272,
      "loss": 4.7818,
      "step": 18077
    },
    {
      "epoch": 2.475760065735415,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004981648259926269,
      "loss": 4.774,
      "step": 18078
    },
    {
      "epoch": 2.475897014516571,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004981240956985732,
      "loss": 4.707,
      "step": 18079
    },
    {
      "epoch": 2.4760339632977266,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004980833650057036,
      "loss": 4.653,
      "step": 18080
    },
    {
      "epoch": 2.4761709120788824,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004980426339143557,
      "loss": 4.6525,
      "step": 18081
    },
    {
      "epoch": 2.476307860860038,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000498001902424867,
      "loss": 4.7324,
      "step": 18082
    },
    {
      "epoch": 2.476444809641194,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000497961170537575,
      "loss": 4.6542,
      "step": 18083
    },
    {
      "epoch": 2.47658175842235,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004979204382528177,
      "loss": 4.6988,
      "step": 18084
    },
    {
      "epoch": 2.476718707203506,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004978797055709321,
      "loss": 4.6729,
      "step": 18085
    },
    {
      "epoch": 2.4768556559846617,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000497838972492256,
      "loss": 4.6954,
      "step": 18086
    },
    {
      "epoch": 2.4769926047658175,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004977982390171269,
      "loss": 4.7312,
      "step": 18087
    },
    {
      "epoch": 2.4771295535469733,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004977575051458824,
      "loss": 4.6811,
      "step": 18088
    },
    {
      "epoch": 2.4772665023281295,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00049771677087886,
      "loss": 4.6658,
      "step": 18089
    },
    {
      "epoch": 2.4774034511092853,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004976760362163975,
      "loss": 4.7583,
      "step": 18090
    },
    {
      "epoch": 2.477540399890441,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004976353011588323,
      "loss": 4.7787,
      "step": 18091
    },
    {
      "epoch": 2.477677348671597,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000497594565706502,
      "loss": 4.6763,
      "step": 18092
    },
    {
      "epoch": 2.4778142974527526,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004975538298597442,
      "loss": 4.6533,
      "step": 18093
    },
    {
      "epoch": 2.4779512462339084,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004975130936188964,
      "loss": 4.7199,
      "step": 18094
    },
    {
      "epoch": 2.478088195015064,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004974723569842964,
      "loss": 4.6887,
      "step": 18095
    },
    {
      "epoch": 2.4782251437962204,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004974316199562815,
      "loss": 4.7159,
      "step": 18096
    },
    {
      "epoch": 2.478362092577376,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004973908825351896,
      "loss": 4.6883,
      "step": 18097
    },
    {
      "epoch": 2.478499041358532,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004973501447213581,
      "loss": 4.6093,
      "step": 18098
    },
    {
      "epoch": 2.4786359901396877,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004973094065151247,
      "loss": 4.7464,
      "step": 18099
    },
    {
      "epoch": 2.4787729389208435,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004972686679168269,
      "loss": 4.7239,
      "step": 18100
    },
    {
      "epoch": 2.4789098877019993,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004972279289268024,
      "loss": 4.7572,
      "step": 18101
    },
    {
      "epoch": 2.4790468364831555,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000497187189545389,
      "loss": 4.6527,
      "step": 18102
    },
    {
      "epoch": 2.4791837852643113,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004971464497729238,
      "loss": 4.7268,
      "step": 18103
    },
    {
      "epoch": 2.479320734045467,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000497105709609745,
      "loss": 4.7199,
      "step": 18104
    },
    {
      "epoch": 2.479457682826623,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004970649690561898,
      "loss": 4.7025,
      "step": 18105
    },
    {
      "epoch": 2.4795946316077786,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004970242281125961,
      "loss": 4.6033,
      "step": 18106
    },
    {
      "epoch": 2.4797315803889344,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004969834867793013,
      "loss": 4.6873,
      "step": 18107
    },
    {
      "epoch": 2.4798685291700906,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004969427450566432,
      "loss": 4.7674,
      "step": 18108
    },
    {
      "epoch": 2.4800054779512464,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004969020029449594,
      "loss": 4.7559,
      "step": 18109
    },
    {
      "epoch": 2.480142426732402,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004968612604445875,
      "loss": 4.6838,
      "step": 18110
    },
    {
      "epoch": 2.480279375513558,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004968205175558653,
      "loss": 4.6944,
      "step": 18111
    },
    {
      "epoch": 2.4804163242947137,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004967797742791301,
      "loss": 4.7254,
      "step": 18112
    },
    {
      "epoch": 2.4805532730758695,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004967390306147199,
      "loss": 4.7101,
      "step": 18113
    },
    {
      "epoch": 2.4806902218570253,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004966982865629721,
      "loss": 4.6649,
      "step": 18114
    },
    {
      "epoch": 2.4808271706381815,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0004966575421242246,
      "loss": 4.7425,
      "step": 18115
    },
    {
      "epoch": 2.4809641194193373,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004966167972988149,
      "loss": 4.7057,
      "step": 18116
    },
    {
      "epoch": 2.481101068200493,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004965760520870807,
      "loss": 4.6429,
      "step": 18117
    },
    {
      "epoch": 2.481238016981649,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004965353064893596,
      "loss": 4.7038,
      "step": 18118
    },
    {
      "epoch": 2.4813749657628046,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004964945605059894,
      "loss": 4.8255,
      "step": 18119
    },
    {
      "epoch": 2.481511914543961,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004964538141373076,
      "loss": 4.6934,
      "step": 18120
    },
    {
      "epoch": 2.4816488633251166,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004964130673836521,
      "loss": 4.6719,
      "step": 18121
    },
    {
      "epoch": 2.4817858121062724,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004963723202453603,
      "loss": 4.6277,
      "step": 18122
    },
    {
      "epoch": 2.481922760887428,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004963315727227702,
      "loss": 4.7626,
      "step": 18123
    },
    {
      "epoch": 2.482059709668584,
      "grad_norm": 0.375,
      "learning_rate": 0.0004962908248162192,
      "loss": 4.6901,
      "step": 18124
    },
    {
      "epoch": 2.4821966584497397,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004962500765260451,
      "loss": 4.7117,
      "step": 18125
    },
    {
      "epoch": 2.4823336072308955,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004962093278525856,
      "loss": 4.7185,
      "step": 18126
    },
    {
      "epoch": 2.4824705560120517,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004961685787961785,
      "loss": 4.7313,
      "step": 18127
    },
    {
      "epoch": 2.4826075047932075,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0004961278293571611,
      "loss": 4.7001,
      "step": 18128
    },
    {
      "epoch": 2.4827444535743632,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004960870795358718,
      "loss": 4.7443,
      "step": 18129
    },
    {
      "epoch": 2.482881402355519,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004960463293326476,
      "loss": 4.7648,
      "step": 18130
    },
    {
      "epoch": 2.483018351136675,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004960055787478267,
      "loss": 4.6956,
      "step": 18131
    },
    {
      "epoch": 2.4831552999178306,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004959648277817464,
      "loss": 4.6638,
      "step": 18132
    },
    {
      "epoch": 2.4832922486989863,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004959240764347447,
      "loss": 4.7195,
      "step": 18133
    },
    {
      "epoch": 2.4834291974801426,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004958833247071592,
      "loss": 4.7587,
      "step": 18134
    },
    {
      "epoch": 2.4835661462612983,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004958425725993278,
      "loss": 4.758,
      "step": 18135
    },
    {
      "epoch": 2.483703095042454,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004958018201115879,
      "loss": 4.7716,
      "step": 18136
    },
    {
      "epoch": 2.48384004382361,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004957610672442775,
      "loss": 4.6442,
      "step": 18137
    },
    {
      "epoch": 2.4839769926047657,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004957203139977341,
      "loss": 4.7597,
      "step": 18138
    },
    {
      "epoch": 2.484113941385922,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004956795603722957,
      "loss": 4.6785,
      "step": 18139
    },
    {
      "epoch": 2.4842508901670777,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004956388063682999,
      "loss": 4.7893,
      "step": 18140
    },
    {
      "epoch": 2.4843878389482335,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004955980519860844,
      "loss": 4.731,
      "step": 18141
    },
    {
      "epoch": 2.4845247877293892,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000495557297225987,
      "loss": 4.6329,
      "step": 18142
    },
    {
      "epoch": 2.484661736510545,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004955165420883454,
      "loss": 4.7028,
      "step": 18143
    },
    {
      "epoch": 2.484798685291701,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004954757865734974,
      "loss": 4.771,
      "step": 18144
    },
    {
      "epoch": 2.4849356340728566,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004954350306817807,
      "loss": 4.7188,
      "step": 18145
    },
    {
      "epoch": 2.485072582854013,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000495394274413533,
      "loss": 4.7926,
      "step": 18146
    },
    {
      "epoch": 2.4852095316351686,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004953535177690923,
      "loss": 4.6921,
      "step": 18147
    },
    {
      "epoch": 2.4853464804163243,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000495312760748796,
      "loss": 4.6021,
      "step": 18148
    },
    {
      "epoch": 2.48548342919748,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004952720033529822,
      "loss": 4.7028,
      "step": 18149
    },
    {
      "epoch": 2.485620377978636,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004952312455819884,
      "loss": 4.6323,
      "step": 18150
    },
    {
      "epoch": 2.4857573267597917,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004951904874361526,
      "loss": 4.6836,
      "step": 18151
    },
    {
      "epoch": 2.4858942755409474,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004951497289158124,
      "loss": 4.7632,
      "step": 18152
    },
    {
      "epoch": 2.4860312243221037,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004951089700213058,
      "loss": 4.7745,
      "step": 18153
    },
    {
      "epoch": 2.4861681731032594,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004950682107529702,
      "loss": 4.771,
      "step": 18154
    },
    {
      "epoch": 2.486305121884415,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004950274511111435,
      "loss": 4.6861,
      "step": 18155
    },
    {
      "epoch": 2.486442070665571,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004949866910961638,
      "loss": 4.7651,
      "step": 18156
    },
    {
      "epoch": 2.4865790194467268,
      "grad_norm": 0.375,
      "learning_rate": 0.0004949459307083687,
      "loss": 4.66,
      "step": 18157
    },
    {
      "epoch": 2.486715968227883,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004949051699480959,
      "loss": 4.7189,
      "step": 18158
    },
    {
      "epoch": 2.4868529170090388,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004948644088156832,
      "loss": 4.6482,
      "step": 18159
    },
    {
      "epoch": 2.4869898657901945,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004948236473114685,
      "loss": 4.6703,
      "step": 18160
    },
    {
      "epoch": 2.4871268145713503,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004947828854357895,
      "loss": 4.6883,
      "step": 18161
    },
    {
      "epoch": 2.487263763352506,
      "grad_norm": 0.375,
      "learning_rate": 0.0004947421231889841,
      "loss": 4.661,
      "step": 18162
    },
    {
      "epoch": 2.487400712133662,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004947013605713901,
      "loss": 4.7147,
      "step": 18163
    },
    {
      "epoch": 2.4875376609148176,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004946605975833453,
      "loss": 4.7333,
      "step": 18164
    },
    {
      "epoch": 2.487674609695974,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004946198342251874,
      "loss": 4.6617,
      "step": 18165
    },
    {
      "epoch": 2.4878115584771296,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004945790704972543,
      "loss": 4.7112,
      "step": 18166
    },
    {
      "epoch": 2.4879485072582854,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004945383063998839,
      "loss": 4.7081,
      "step": 18167
    },
    {
      "epoch": 2.488085456039441,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004944975419334138,
      "loss": 4.6767,
      "step": 18168
    },
    {
      "epoch": 2.488222404820597,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004944567770981821,
      "loss": 4.6768,
      "step": 18169
    },
    {
      "epoch": 2.4883593536017528,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004944160118945265,
      "loss": 4.7248,
      "step": 18170
    },
    {
      "epoch": 2.488496302382909,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004943752463227848,
      "loss": 4.6577,
      "step": 18171
    },
    {
      "epoch": 2.4886332511640648,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004943344803832948,
      "loss": 4.6801,
      "step": 18172
    },
    {
      "epoch": 2.4887701999452205,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004942937140763944,
      "loss": 4.711,
      "step": 18173
    },
    {
      "epoch": 2.4889071487263763,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004942529474024215,
      "loss": 4.6426,
      "step": 18174
    },
    {
      "epoch": 2.489044097507532,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004942121803617138,
      "loss": 4.7203,
      "step": 18175
    },
    {
      "epoch": 2.489181046288688,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004941714129546093,
      "loss": 4.6533,
      "step": 18176
    },
    {
      "epoch": 2.489317995069844,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004941306451814458,
      "loss": 4.7853,
      "step": 18177
    },
    {
      "epoch": 2.489454943851,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004940898770425612,
      "loss": 4.7467,
      "step": 18178
    },
    {
      "epoch": 2.4895918926321556,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000494049108538293,
      "loss": 4.7051,
      "step": 18179
    },
    {
      "epoch": 2.4897288414133114,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004940083396689795,
      "loss": 4.7026,
      "step": 18180
    },
    {
      "epoch": 2.489865790194467,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004939675704349584,
      "loss": 4.7287,
      "step": 18181
    },
    {
      "epoch": 2.490002738975623,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004939268008365675,
      "loss": 4.6908,
      "step": 18182
    },
    {
      "epoch": 2.4901396877567787,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000493886030874145,
      "loss": 4.7188,
      "step": 18183
    },
    {
      "epoch": 2.490276636537935,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004938452605480282,
      "loss": 4.7538,
      "step": 18184
    },
    {
      "epoch": 2.4904135853190907,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004938044898585556,
      "loss": 4.6762,
      "step": 18185
    },
    {
      "epoch": 2.4905505341002465,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004937637188060644,
      "loss": 4.864,
      "step": 18186
    },
    {
      "epoch": 2.4906874828814023,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004937229473908929,
      "loss": 4.6934,
      "step": 18187
    },
    {
      "epoch": 2.490824431662558,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000493682175613379,
      "loss": 4.7277,
      "step": 18188
    },
    {
      "epoch": 2.4909613804437143,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004936414034738605,
      "loss": 4.6912,
      "step": 18189
    },
    {
      "epoch": 2.49109832922487,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004936006309726753,
      "loss": 4.778,
      "step": 18190
    },
    {
      "epoch": 2.491235278006026,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004935598581101613,
      "loss": 4.7277,
      "step": 18191
    },
    {
      "epoch": 2.4913722267871816,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004935190848866564,
      "loss": 4.7504,
      "step": 18192
    },
    {
      "epoch": 2.4915091755683374,
      "grad_norm": 0.375,
      "learning_rate": 0.0004934783113024983,
      "loss": 4.5882,
      "step": 18193
    },
    {
      "epoch": 2.491646124349493,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004934375373580252,
      "loss": 4.7109,
      "step": 18194
    },
    {
      "epoch": 2.491783073130649,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000493396763053575,
      "loss": 4.7676,
      "step": 18195
    },
    {
      "epoch": 2.491920021911805,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004933559883894852,
      "loss": 4.6979,
      "step": 18196
    },
    {
      "epoch": 2.492056970692961,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004933152133660942,
      "loss": 4.7474,
      "step": 18197
    },
    {
      "epoch": 2.4921939194741167,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004932744379837397,
      "loss": 4.7117,
      "step": 18198
    },
    {
      "epoch": 2.4923308682552725,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004932336622427596,
      "loss": 4.8248,
      "step": 18199
    },
    {
      "epoch": 2.4924678170364283,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004931928861434917,
      "loss": 4.6264,
      "step": 18200
    },
    {
      "epoch": 2.492604765817584,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004931521096862741,
      "loss": 4.6231,
      "step": 18201
    },
    {
      "epoch": 2.49274171459874,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004931113328714449,
      "loss": 4.7402,
      "step": 18202
    },
    {
      "epoch": 2.492878663379896,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004930705556993417,
      "loss": 4.637,
      "step": 18203
    },
    {
      "epoch": 2.493015612161052,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004930297781703025,
      "loss": 4.61,
      "step": 18204
    },
    {
      "epoch": 2.4931525609422076,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004929890002846652,
      "loss": 4.6973,
      "step": 18205
    },
    {
      "epoch": 2.4932895097233634,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004929482220427679,
      "loss": 4.7131,
      "step": 18206
    },
    {
      "epoch": 2.493426458504519,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004929074434449485,
      "loss": 4.75,
      "step": 18207
    },
    {
      "epoch": 2.4935634072856754,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004928666644915448,
      "loss": 4.7078,
      "step": 18208
    },
    {
      "epoch": 2.493700356066831,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000492825885182895,
      "loss": 4.6797,
      "step": 18209
    },
    {
      "epoch": 2.493837304847987,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004927851055193367,
      "loss": 4.7336,
      "step": 18210
    },
    {
      "epoch": 2.4939742536291427,
      "grad_norm": 0.40625,
      "learning_rate": 0.0004927443255012082,
      "loss": 4.6229,
      "step": 18211
    },
    {
      "epoch": 2.4941112024102985,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000492703545128847,
      "loss": 4.7831,
      "step": 18212
    },
    {
      "epoch": 2.4942481511914543,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004926627644025915,
      "loss": 4.7486,
      "step": 18213
    },
    {
      "epoch": 2.49438509997261,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004926219833227796,
      "loss": 4.7438,
      "step": 18214
    },
    {
      "epoch": 2.4945220487537663,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004925812018897491,
      "loss": 4.7326,
      "step": 18215
    },
    {
      "epoch": 2.494658997534922,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000492540420103838,
      "loss": 4.6969,
      "step": 18216
    },
    {
      "epoch": 2.494795946316078,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004924996379653844,
      "loss": 4.8476,
      "step": 18217
    },
    {
      "epoch": 2.4949328950972336,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004924588554747261,
      "loss": 4.6429,
      "step": 18218
    },
    {
      "epoch": 2.4950698438783894,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004924180726322011,
      "loss": 4.7383,
      "step": 18219
    },
    {
      "epoch": 2.495206792659545,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004923772894381476,
      "loss": 4.7729,
      "step": 18220
    },
    {
      "epoch": 2.4953437414407014,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004923365058929032,
      "loss": 4.6922,
      "step": 18221
    },
    {
      "epoch": 2.495480690221857,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0004922957219968061,
      "loss": 4.6177,
      "step": 18222
    },
    {
      "epoch": 2.495617639003013,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004922549377501945,
      "loss": 4.7592,
      "step": 18223
    },
    {
      "epoch": 2.4957545877841687,
      "grad_norm": 0.375,
      "learning_rate": 0.0004922141531534059,
      "loss": 4.674,
      "step": 18224
    },
    {
      "epoch": 2.4958915365653245,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0004921733682067788,
      "loss": 4.6893,
      "step": 18225
    },
    {
      "epoch": 2.4960284853464803,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004921325829106507,
      "loss": 4.7329,
      "step": 18226
    },
    {
      "epoch": 2.4961654341276365,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00049209179726536,
      "loss": 4.7147,
      "step": 18227
    },
    {
      "epoch": 2.4963023829087923,
      "grad_norm": 0.390625,
      "learning_rate": 0.0004920510112712445,
      "loss": 4.6631,
      "step": 18228
    },
    {
      "epoch": 2.496439331689948,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004920102249286423,
      "loss": 4.7351,
      "step": 18229
    },
    {
      "epoch": 2.496576280471104,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004919694382378913,
      "loss": 4.6654,
      "step": 18230
    },
    {
      "epoch": 2.4967132292522596,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004919286511993296,
      "loss": 4.6745,
      "step": 18231
    },
    {
      "epoch": 2.4968501780334154,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0004918878638132951,
      "loss": 4.5971,
      "step": 18232
    },
    {
      "epoch": 2.496987126814571,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004918470760801258,
      "loss": 4.7202,
      "step": 18233
    },
    {
      "epoch": 2.4971240755957274,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004918062880001601,
      "loss": 4.6667,
      "step": 18234
    },
    {
      "epoch": 2.497261024376883,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004917654995737356,
      "loss": 4.7156,
      "step": 18235
    },
    {
      "epoch": 2.497397973158039,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004917247108011905,
      "loss": 4.7057,
      "step": 18236
    },
    {
      "epoch": 2.4975349219391947,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004916839216828627,
      "loss": 4.7212,
      "step": 18237
    },
    {
      "epoch": 2.4976718707203505,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004916431322190903,
      "loss": 4.7208,
      "step": 18238
    },
    {
      "epoch": 2.4978088195015067,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004916023424102114,
      "loss": 4.713,
      "step": 18239
    },
    {
      "epoch": 2.4979457682826625,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004915615522565638,
      "loss": 4.708,
      "step": 18240
    },
    {
      "epoch": 2.4980827170638182,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004915207617584859,
      "loss": 4.6533,
      "step": 18241
    },
    {
      "epoch": 2.498219665844974,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004914799709163155,
      "loss": 4.7273,
      "step": 18242
    },
    {
      "epoch": 2.49835661462613,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004914391797303907,
      "loss": 4.6828,
      "step": 18243
    },
    {
      "epoch": 2.4984935634072856,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004913983882010495,
      "loss": 4.7147,
      "step": 18244
    },
    {
      "epoch": 2.4986305121884413,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004913575963286301,
      "loss": 4.7153,
      "step": 18245
    },
    {
      "epoch": 2.4987674609695976,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004913168041134703,
      "loss": 4.771,
      "step": 18246
    },
    {
      "epoch": 2.4989044097507533,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004912760115559085,
      "loss": 4.6905,
      "step": 18247
    },
    {
      "epoch": 2.499041358531909,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004912352186562823,
      "loss": 4.7408,
      "step": 18248
    },
    {
      "epoch": 2.499178307313065,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004911944254149302,
      "loss": 4.7839,
      "step": 18249
    },
    {
      "epoch": 2.4993152560942207,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004911536318321901,
      "loss": 4.6639,
      "step": 18250
    },
    {
      "epoch": 2.4994522048753764,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004911128379083999,
      "loss": 4.6592,
      "step": 18251
    },
    {
      "epoch": 2.4995891536565322,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000491072043643898,
      "loss": 4.8042,
      "step": 18252
    },
    {
      "epoch": 2.4997261024376884,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004910312490390221,
      "loss": 4.6149,
      "step": 18253
    },
    {
      "epoch": 2.4998630512188442,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004909904540941106,
      "loss": 4.6824,
      "step": 18254
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004909496588095013,
      "loss": 4.7126,
      "step": 18255
    },
    {
      "epoch": 2.5001369487811558,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004909088631855326,
      "loss": 4.6766,
      "step": 18256
    },
    {
      "epoch": 2.5002738975623116,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004908680672225423,
      "loss": 4.7064,
      "step": 18257
    },
    {
      "epoch": 2.5004108463434678,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004908272709208685,
      "loss": 4.7702,
      "step": 18258
    },
    {
      "epoch": 2.5005477951246236,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004907864742808495,
      "loss": 4.7714,
      "step": 18259
    },
    {
      "epoch": 2.5006847439057793,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004907456773028232,
      "loss": 4.7589,
      "step": 18260
    },
    {
      "epoch": 2.500821692686935,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004907048799871278,
      "loss": 4.7018,
      "step": 18261
    },
    {
      "epoch": 2.500958641468091,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004906640823341015,
      "loss": 4.7207,
      "step": 18262
    },
    {
      "epoch": 2.5010955902492467,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004906232843440821,
      "loss": 4.7795,
      "step": 18263
    },
    {
      "epoch": 2.5012325390304024,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004905824860174078,
      "loss": 4.7976,
      "step": 18264
    },
    {
      "epoch": 2.5013694878115587,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000490541687354417,
      "loss": 4.691,
      "step": 18265
    },
    {
      "epoch": 2.5015064365927144,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004905008883554473,
      "loss": 4.7076,
      "step": 18266
    },
    {
      "epoch": 2.50164338537387,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004904600890208372,
      "loss": 4.7033,
      "step": 18267
    },
    {
      "epoch": 2.501780334155026,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004904192893509248,
      "loss": 4.7032,
      "step": 18268
    },
    {
      "epoch": 2.5019172829361818,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000490378489346048,
      "loss": 4.7297,
      "step": 18269
    },
    {
      "epoch": 2.502054231717338,
      "grad_norm": 0.396484375,
      "learning_rate": 0.000490337689006545,
      "loss": 4.6924,
      "step": 18270
    },
    {
      "epoch": 2.5021911804984933,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000490296888332754,
      "loss": 4.651,
      "step": 18271
    },
    {
      "epoch": 2.5023281292796495,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004902560873250132,
      "loss": 4.6961,
      "step": 18272
    },
    {
      "epoch": 2.5024650780608053,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004902152859836605,
      "loss": 4.6414,
      "step": 18273
    },
    {
      "epoch": 2.502602026841961,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004901744843090343,
      "loss": 4.7239,
      "step": 18274
    },
    {
      "epoch": 2.502738975623117,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004901336823014723,
      "loss": 4.6925,
      "step": 18275
    },
    {
      "epoch": 2.5028759244042726,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004900928799613131,
      "loss": 4.6845,
      "step": 18276
    },
    {
      "epoch": 2.503012873185429,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004900520772888946,
      "loss": 4.6915,
      "step": 18277
    },
    {
      "epoch": 2.5031498219665846,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000490011274284555,
      "loss": 4.6161,
      "step": 18278
    },
    {
      "epoch": 2.5032867707477404,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004899704709486326,
      "loss": 4.6984,
      "step": 18279
    },
    {
      "epoch": 2.503423719528896,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004899296672814651,
      "loss": 4.7289,
      "step": 18280
    },
    {
      "epoch": 2.503560668310052,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004898888632833911,
      "loss": 4.7525,
      "step": 18281
    },
    {
      "epoch": 2.5036976170912077,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004898480589547484,
      "loss": 4.6788,
      "step": 18282
    },
    {
      "epoch": 2.5038345658723635,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004898072542958757,
      "loss": 4.705,
      "step": 18283
    },
    {
      "epoch": 2.5039715146535197,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004897664493071104,
      "loss": 4.7298,
      "step": 18284
    },
    {
      "epoch": 2.5041084634346755,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004897256439887913,
      "loss": 4.7493,
      "step": 18285
    },
    {
      "epoch": 2.5042454122158313,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004896848383412562,
      "loss": 4.6303,
      "step": 18286
    },
    {
      "epoch": 2.504382360996987,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004896440323648435,
      "loss": 4.7229,
      "step": 18287
    },
    {
      "epoch": 2.504519309778143,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004896032260598911,
      "loss": 4.7377,
      "step": 18288
    },
    {
      "epoch": 2.504656258559299,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004895624194267375,
      "loss": 4.6961,
      "step": 18289
    },
    {
      "epoch": 2.5047932073404544,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004895216124657206,
      "loss": 4.726,
      "step": 18290
    },
    {
      "epoch": 2.5049301561216106,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004894808051771788,
      "loss": 4.6964,
      "step": 18291
    },
    {
      "epoch": 2.5050671049027664,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004894399975614502,
      "loss": 4.6823,
      "step": 18292
    },
    {
      "epoch": 2.505204053683922,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004893991896188726,
      "loss": 4.7033,
      "step": 18293
    },
    {
      "epoch": 2.505341002465078,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004893583813497848,
      "loss": 4.6635,
      "step": 18294
    },
    {
      "epoch": 2.5054779512462337,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004893175727545247,
      "loss": 4.7613,
      "step": 18295
    },
    {
      "epoch": 2.50561490002739,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004892767638334304,
      "loss": 4.721,
      "step": 18296
    },
    {
      "epoch": 2.5057518488085457,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004892359545868403,
      "loss": 4.7248,
      "step": 18297
    },
    {
      "epoch": 2.5058887975897015,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004891951450150924,
      "loss": 4.6827,
      "step": 18298
    },
    {
      "epoch": 2.5060257463708573,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004891543351185252,
      "loss": 4.6506,
      "step": 18299
    },
    {
      "epoch": 2.506162695152013,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004891135248974764,
      "loss": 4.7027,
      "step": 18300
    },
    {
      "epoch": 2.506299643933169,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004890727143522847,
      "loss": 4.7476,
      "step": 18301
    },
    {
      "epoch": 2.5064365927143246,
      "grad_norm": 0.384765625,
      "learning_rate": 0.000489031903483288,
      "loss": 4.6535,
      "step": 18302
    },
    {
      "epoch": 2.506573541495481,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004889910922908248,
      "loss": 4.7142,
      "step": 18303
    },
    {
      "epoch": 2.5067104902766366,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000488950280775233,
      "loss": 4.7415,
      "step": 18304
    },
    {
      "epoch": 2.5068474390577924,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004889094689368508,
      "loss": 4.8215,
      "step": 18305
    },
    {
      "epoch": 2.506984387838948,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004888686567760167,
      "loss": 4.7016,
      "step": 18306
    },
    {
      "epoch": 2.507121336620104,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004888278442930687,
      "loss": 4.7152,
      "step": 18307
    },
    {
      "epoch": 2.50725828540126,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004887870314883453,
      "loss": 4.698,
      "step": 18308
    },
    {
      "epoch": 2.5073952341824155,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004887462183621843,
      "loss": 4.7532,
      "step": 18309
    },
    {
      "epoch": 2.5075321829635717,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004887054049149244,
      "loss": 4.6296,
      "step": 18310
    },
    {
      "epoch": 2.5076691317447275,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004886645911469033,
      "loss": 4.7341,
      "step": 18311
    },
    {
      "epoch": 2.5078060805258833,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004886237770584598,
      "loss": 4.7069,
      "step": 18312
    },
    {
      "epoch": 2.507943029307039,
      "grad_norm": 0.375,
      "learning_rate": 0.0004885829626499317,
      "loss": 4.693,
      "step": 18313
    },
    {
      "epoch": 2.508079978088195,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004885421479216574,
      "loss": 4.6698,
      "step": 18314
    },
    {
      "epoch": 2.508216926869351,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004885013328739751,
      "loss": 4.6983,
      "step": 18315
    },
    {
      "epoch": 2.508353875650507,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004884605175072231,
      "loss": 4.6695,
      "step": 18316
    },
    {
      "epoch": 2.5084908244316626,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004884197018217397,
      "loss": 4.6958,
      "step": 18317
    },
    {
      "epoch": 2.5086277732128184,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004883788858178631,
      "loss": 4.6521,
      "step": 18318
    },
    {
      "epoch": 2.508764721993974,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004883380694959314,
      "loss": 4.7246,
      "step": 18319
    },
    {
      "epoch": 2.5089016707751304,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000488297252856283,
      "loss": 4.6806,
      "step": 18320
    },
    {
      "epoch": 2.5090386195562857,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004882564358992563,
      "loss": 4.7127,
      "step": 18321
    },
    {
      "epoch": 2.509175568337442,
      "grad_norm": 0.359375,
      "learning_rate": 0.00048821561862518923,
      "loss": 4.6432,
      "step": 18322
    },
    {
      "epoch": 2.5093125171185977,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004881748010344203,
      "loss": 4.8404,
      "step": 18323
    },
    {
      "epoch": 2.5094494658997535,
      "grad_norm": 0.34375,
      "learning_rate": 0.00048813398312728767,
      "loss": 4.7526,
      "step": 18324
    },
    {
      "epoch": 2.5095864146809093,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048809316490412957,
      "loss": 4.712,
      "step": 18325
    },
    {
      "epoch": 2.509723363462065,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00048805234636528443,
      "loss": 4.6199,
      "step": 18326
    },
    {
      "epoch": 2.5098603122432213,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004880115275110904,
      "loss": 4.7615,
      "step": 18327
    },
    {
      "epoch": 2.509997261024377,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00048797070834188586,
      "loss": 4.7462,
      "step": 18328
    },
    {
      "epoch": 2.510134209805533,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00048792988885800885,
      "loss": 4.6739,
      "step": 18329
    },
    {
      "epoch": 2.5102711585866886,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000487889069059798,
      "loss": 4.782,
      "step": 18330
    },
    {
      "epoch": 2.5104081073678444,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00048784824894759126,
      "loss": 4.7054,
      "step": 18331
    },
    {
      "epoch": 2.510545056149,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00048780742852172715,
      "loss": 4.7793,
      "step": 18332
    },
    {
      "epoch": 2.510682004930156,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004877666077825438,
      "loss": 4.8037,
      "step": 18333
    },
    {
      "epoch": 2.510818953711312,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00048772578673037957,
      "loss": 4.6859,
      "step": 18334
    },
    {
      "epoch": 2.510955902492468,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00048768496536557286,
      "loss": 4.6809,
      "step": 18335
    },
    {
      "epoch": 2.5110928512736237,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004876441436884617,
      "loss": 4.7075,
      "step": 18336
    },
    {
      "epoch": 2.5112298000547795,
      "grad_norm": 0.375,
      "learning_rate": 0.00048760332169938466,
      "loss": 4.7279,
      "step": 18337
    },
    {
      "epoch": 2.5113667488359352,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004875624993986799,
      "loss": 4.6383,
      "step": 18338
    },
    {
      "epoch": 2.5115036976170915,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00048752167678668587,
      "loss": 4.7423,
      "step": 18339
    },
    {
      "epoch": 2.511640646398247,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00048748085386374066,
      "loss": 4.712,
      "step": 18340
    },
    {
      "epoch": 2.511777595179403,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00048744003063018264,
      "loss": 4.6481,
      "step": 18341
    },
    {
      "epoch": 2.511914543960559,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00048739920708635014,
      "loss": 4.7869,
      "step": 18342
    },
    {
      "epoch": 2.5120514927417146,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004873583832325816,
      "loss": 4.7718,
      "step": 18343
    },
    {
      "epoch": 2.5121884415228704,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00048731755906921526,
      "loss": 4.7545,
      "step": 18344
    },
    {
      "epoch": 2.512325390304026,
      "grad_norm": 0.375,
      "learning_rate": 0.0004872767345965893,
      "loss": 4.6659,
      "step": 18345
    },
    {
      "epoch": 2.5124623390851824,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004872359098150423,
      "loss": 4.727,
      "step": 18346
    },
    {
      "epoch": 2.512599287866338,
      "grad_norm": 0.375,
      "learning_rate": 0.00048719508472491236,
      "loss": 4.6757,
      "step": 18347
    },
    {
      "epoch": 2.512736236647494,
      "grad_norm": 0.359375,
      "learning_rate": 0.00048715425932653795,
      "loss": 4.7296,
      "step": 18348
    },
    {
      "epoch": 2.5128731854286497,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004871134336202572,
      "loss": 4.7285,
      "step": 18349
    },
    {
      "epoch": 2.5130101342098055,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004870726076064087,
      "loss": 4.7174,
      "step": 18350
    },
    {
      "epoch": 2.5131470829909612,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004870317812853306,
      "loss": 4.6848,
      "step": 18351
    },
    {
      "epoch": 2.513284031772117,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048699095465736127,
      "loss": 4.6599,
      "step": 18352
    },
    {
      "epoch": 2.5134209805532732,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048695012772283924,
      "loss": 4.7556,
      "step": 18353
    },
    {
      "epoch": 2.513557929334429,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004869093004821026,
      "loss": 4.695,
      "step": 18354
    },
    {
      "epoch": 2.513694878115585,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048686847293548976,
      "loss": 4.7419,
      "step": 18355
    },
    {
      "epoch": 2.5138318268967406,
      "grad_norm": 0.375,
      "learning_rate": 0.0004868276450833392,
      "loss": 4.6917,
      "step": 18356
    },
    {
      "epoch": 2.5139687756778963,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004867868169259891,
      "loss": 4.7674,
      "step": 18357
    },
    {
      "epoch": 2.5141057244590526,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00048674598846377786,
      "loss": 4.7222,
      "step": 18358
    },
    {
      "epoch": 2.514242673240208,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00048670515969704393,
      "loss": 4.723,
      "step": 18359
    },
    {
      "epoch": 2.514379622021364,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00048666433062612553,
      "loss": 4.5424,
      "step": 18360
    },
    {
      "epoch": 2.51451657080252,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000486623501251361,
      "loss": 4.6896,
      "step": 18361
    },
    {
      "epoch": 2.5146535195836757,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00048658267157308897,
      "loss": 4.7544,
      "step": 18362
    },
    {
      "epoch": 2.5147904683648314,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00048654184159164756,
      "loss": 4.7624,
      "step": 18363
    },
    {
      "epoch": 2.514927417145987,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00048650101130737525,
      "loss": 4.6335,
      "step": 18364
    },
    {
      "epoch": 2.5150643659271434,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004864601807206102,
      "loss": 4.6682,
      "step": 18365
    },
    {
      "epoch": 2.515201314708299,
      "grad_norm": 0.34375,
      "learning_rate": 0.00048641934983169116,
      "loss": 4.6855,
      "step": 18366
    },
    {
      "epoch": 2.515338263489455,
      "grad_norm": 0.34375,
      "learning_rate": 0.00048637851864095607,
      "loss": 4.7217,
      "step": 18367
    },
    {
      "epoch": 2.5154752122706108,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004863376871487437,
      "loss": 4.7751,
      "step": 18368
    },
    {
      "epoch": 2.5156121610517665,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004862968553553922,
      "loss": 4.7368,
      "step": 18369
    },
    {
      "epoch": 2.5157491098329223,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00048625602326123995,
      "loss": 4.7243,
      "step": 18370
    },
    {
      "epoch": 2.515886058614078,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00048621519086662553,
      "loss": 4.7768,
      "step": 18371
    },
    {
      "epoch": 2.5160230073952343,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00048617435817188703,
      "loss": 4.7209,
      "step": 18372
    },
    {
      "epoch": 2.51615995617639,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00048613352517736316,
      "loss": 4.7338,
      "step": 18373
    },
    {
      "epoch": 2.516296904957546,
      "grad_norm": 0.34375,
      "learning_rate": 0.000486092691883392,
      "loss": 4.6793,
      "step": 18374
    },
    {
      "epoch": 2.5164338537387017,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004860518582903123,
      "loss": 4.7435,
      "step": 18375
    },
    {
      "epoch": 2.5165708025198574,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00048601102439846204,
      "loss": 4.7272,
      "step": 18376
    },
    {
      "epoch": 2.5167077513010137,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00048597019020818006,
      "loss": 4.7217,
      "step": 18377
    },
    {
      "epoch": 2.5168447000821694,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004859293557198044,
      "loss": 4.7605,
      "step": 18378
    },
    {
      "epoch": 2.516981648863325,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0004858885209336736,
      "loss": 4.7271,
      "step": 18379
    },
    {
      "epoch": 2.517118597644481,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004858476858501261,
      "loss": 4.7558,
      "step": 18380
    },
    {
      "epoch": 2.5172555464256368,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00048580685046950025,
      "loss": 4.6394,
      "step": 18381
    },
    {
      "epoch": 2.5173924952067925,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004857660147921346,
      "loss": 4.6321,
      "step": 18382
    },
    {
      "epoch": 2.5175294439879483,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00048572517881836734,
      "loss": 4.728,
      "step": 18383
    },
    {
      "epoch": 2.5176663927691045,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004856843425485372,
      "loss": 4.7142,
      "step": 18384
    },
    {
      "epoch": 2.5178033415502603,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004856435059829822,
      "loss": 4.7631,
      "step": 18385
    },
    {
      "epoch": 2.517940290331416,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004856026691220411,
      "loss": 4.6601,
      "step": 18386
    },
    {
      "epoch": 2.518077239112572,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004855618319660521,
      "loss": 4.6997,
      "step": 18387
    },
    {
      "epoch": 2.5182141878937276,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048552099451535383,
      "loss": 4.7959,
      "step": 18388
    },
    {
      "epoch": 2.518351136674884,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004854801567702846,
      "loss": 4.6736,
      "step": 18389
    },
    {
      "epoch": 2.518488085456039,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004854393187311828,
      "loss": 4.6881,
      "step": 18390
    },
    {
      "epoch": 2.5186250342371954,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004853984803983869,
      "loss": 4.6221,
      "step": 18391
    },
    {
      "epoch": 2.518761983018351,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00048535764177223535,
      "loss": 4.7611,
      "step": 18392
    },
    {
      "epoch": 2.518898931799507,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00048531680285306676,
      "loss": 4.6734,
      "step": 18393
    },
    {
      "epoch": 2.5190358805806627,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004852759636412193,
      "loss": 4.7608,
      "step": 18394
    },
    {
      "epoch": 2.5191728293618185,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00048523512413703153,
      "loss": 4.6815,
      "step": 18395
    },
    {
      "epoch": 2.5193097781429747,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004851942843408418,
      "loss": 4.8272,
      "step": 18396
    },
    {
      "epoch": 2.5194467269241305,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004851534442529889,
      "loss": 4.6742,
      "step": 18397
    },
    {
      "epoch": 2.5195836757052863,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00048511260387381076,
      "loss": 4.7446,
      "step": 18398
    },
    {
      "epoch": 2.519720624486442,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004850717632036461,
      "loss": 4.7125,
      "step": 18399
    },
    {
      "epoch": 2.519857573267598,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048503092224283354,
      "loss": 4.7505,
      "step": 18400
    },
    {
      "epoch": 2.5199945220487536,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00048499008099171136,
      "loss": 4.685,
      "step": 18401
    },
    {
      "epoch": 2.5201314708299094,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00048494923945061806,
      "loss": 4.6949,
      "step": 18402
    },
    {
      "epoch": 2.5202684196110656,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00048490839761989197,
      "loss": 4.648,
      "step": 18403
    },
    {
      "epoch": 2.5204053683922214,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0004848675554998718,
      "loss": 4.6755,
      "step": 18404
    },
    {
      "epoch": 2.520542317173377,
      "grad_norm": 0.34375,
      "learning_rate": 0.00048482671309089574,
      "loss": 4.7618,
      "step": 18405
    },
    {
      "epoch": 2.520679265954533,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00048478587039330246,
      "loss": 4.6997,
      "step": 18406
    },
    {
      "epoch": 2.5208162147356887,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004847450274074305,
      "loss": 4.6907,
      "step": 18407
    },
    {
      "epoch": 2.520953163516845,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00048470418413361805,
      "loss": 4.753,
      "step": 18408
    },
    {
      "epoch": 2.5210901122980003,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004846633405722039,
      "loss": 4.7025,
      "step": 18409
    },
    {
      "epoch": 2.5212270610791565,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004846224967235263,
      "loss": 4.6404,
      "step": 18410
    },
    {
      "epoch": 2.5213640098603123,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004845816525879239,
      "loss": 4.6775,
      "step": 18411
    },
    {
      "epoch": 2.521500958641468,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048454080816573496,
      "loss": 4.651,
      "step": 18412
    },
    {
      "epoch": 2.521637907422624,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004844999634572982,
      "loss": 4.6814,
      "step": 18413
    },
    {
      "epoch": 2.5217748562037796,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004844591184629521,
      "loss": 4.6273,
      "step": 18414
    },
    {
      "epoch": 2.521911804984936,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004844182731830349,
      "loss": 4.6951,
      "step": 18415
    },
    {
      "epoch": 2.5220487537660916,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00048437742761788547,
      "loss": 4.6816,
      "step": 18416
    },
    {
      "epoch": 2.5221857025472474,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000484336581767842,
      "loss": 4.7073,
      "step": 18417
    },
    {
      "epoch": 2.522322651328403,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00048429573563324316,
      "loss": 4.6873,
      "step": 18418
    },
    {
      "epoch": 2.522459600109559,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004842548892144273,
      "loss": 4.6475,
      "step": 18419
    },
    {
      "epoch": 2.5225965488907147,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00048421404251173297,
      "loss": 4.7914,
      "step": 18420
    },
    {
      "epoch": 2.5227334976718705,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00048417319552549886,
      "loss": 4.6804,
      "step": 18421
    },
    {
      "epoch": 2.5228704464530267,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00048413234825606317,
      "loss": 4.7109,
      "step": 18422
    },
    {
      "epoch": 2.5230073952341825,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00048409150070376477,
      "loss": 4.7943,
      "step": 18423
    },
    {
      "epoch": 2.5231443440153383,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004840506528689417,
      "loss": 4.7357,
      "step": 18424
    },
    {
      "epoch": 2.523281292796494,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00048400980475193304,
      "loss": 4.7572,
      "step": 18425
    },
    {
      "epoch": 2.52341824157765,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004839689563530769,
      "loss": 4.8049,
      "step": 18426
    },
    {
      "epoch": 2.523555190358806,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004839281076727119,
      "loss": 4.7506,
      "step": 18427
    },
    {
      "epoch": 2.523692139139962,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004838872587111766,
      "loss": 4.7583,
      "step": 18428
    },
    {
      "epoch": 2.5238290879211176,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00048384640946880957,
      "loss": 4.6851,
      "step": 18429
    },
    {
      "epoch": 2.5239660367022734,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00048380555994594933,
      "loss": 4.6675,
      "step": 18430
    },
    {
      "epoch": 2.524102985483429,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048376471014293423,
      "loss": 4.7038,
      "step": 18431
    },
    {
      "epoch": 2.524239934264585,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00048372386006010303,
      "loss": 4.7611,
      "step": 18432
    },
    {
      "epoch": 2.5243768830457407,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000483683009697794,
      "loss": 4.7197,
      "step": 18433
    },
    {
      "epoch": 2.524513831826897,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00048364215905634595,
      "loss": 4.7017,
      "step": 18434
    },
    {
      "epoch": 2.5246507806080527,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00048360130813609743,
      "loss": 4.6969,
      "step": 18435
    },
    {
      "epoch": 2.5247877293892085,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00048356045693738665,
      "loss": 4.6699,
      "step": 18436
    },
    {
      "epoch": 2.5249246781703643,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00048351960546055253,
      "loss": 4.6618,
      "step": 18437
    },
    {
      "epoch": 2.52506162695152,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00048347875370593335,
      "loss": 4.7369,
      "step": 18438
    },
    {
      "epoch": 2.5251985757326763,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00048343790167386787,
      "loss": 4.7086,
      "step": 18439
    },
    {
      "epoch": 2.5253355245138316,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004833970493646944,
      "loss": 4.798,
      "step": 18440
    },
    {
      "epoch": 2.525472473294988,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00048335619677875166,
      "loss": 4.7045,
      "step": 18441
    },
    {
      "epoch": 2.5256094220761436,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004833153439163782,
      "loss": 4.6302,
      "step": 18442
    },
    {
      "epoch": 2.5257463708572994,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00048327449077791257,
      "loss": 4.6093,
      "step": 18443
    },
    {
      "epoch": 2.525883319638455,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004832336373636933,
      "loss": 4.6896,
      "step": 18444
    },
    {
      "epoch": 2.526020268419611,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004831927836740589,
      "loss": 4.7272,
      "step": 18445
    },
    {
      "epoch": 2.526157217200767,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00048315192970934805,
      "loss": 4.7108,
      "step": 18446
    },
    {
      "epoch": 2.526294165981923,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004831110754698992,
      "loss": 4.7198,
      "step": 18447
    },
    {
      "epoch": 2.5264311147630787,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000483070220956051,
      "loss": 4.7053,
      "step": 18448
    },
    {
      "epoch": 2.5265680635442345,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00048302936616814203,
      "loss": 4.6591,
      "step": 18449
    },
    {
      "epoch": 2.5267050123253902,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004829885111065108,
      "loss": 4.7619,
      "step": 18450
    },
    {
      "epoch": 2.526841961106546,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048294765577149603,
      "loss": 4.7034,
      "step": 18451
    },
    {
      "epoch": 2.526978909887702,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00048290680016343607,
      "loss": 4.6111,
      "step": 18452
    },
    {
      "epoch": 2.527115858668858,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004828659442826696,
      "loss": 4.661,
      "step": 18453
    },
    {
      "epoch": 2.527252807450014,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004828250881295353,
      "loss": 4.632,
      "step": 18454
    },
    {
      "epoch": 2.5273897562311696,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0004827842317043715,
      "loss": 4.7527,
      "step": 18455
    },
    {
      "epoch": 2.5275267050123253,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004827433750075171,
      "loss": 4.7118,
      "step": 18456
    },
    {
      "epoch": 2.527663653793481,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00048270251803931056,
      "loss": 4.5962,
      "step": 18457
    },
    {
      "epoch": 2.5278006025746373,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004826616608000905,
      "loss": 4.6248,
      "step": 18458
    },
    {
      "epoch": 2.5279375513557927,
      "grad_norm": 0.34375,
      "learning_rate": 0.00048262080329019533,
      "loss": 4.664,
      "step": 18459
    },
    {
      "epoch": 2.528074500136949,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004825799455099639,
      "loss": 4.6923,
      "step": 18460
    },
    {
      "epoch": 2.5282114489181047,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004825390874597347,
      "loss": 4.753,
      "step": 18461
    },
    {
      "epoch": 2.5283483976992605,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00048249822913984624,
      "loss": 4.6179,
      "step": 18462
    },
    {
      "epoch": 2.5284853464804162,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00048245737055063726,
      "loss": 4.6754,
      "step": 18463
    },
    {
      "epoch": 2.528622295261572,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00048241651169244633,
      "loss": 4.7387,
      "step": 18464
    },
    {
      "epoch": 2.5287592440427282,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004823756525656121,
      "loss": 4.6578,
      "step": 18465
    },
    {
      "epoch": 2.528896192823884,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000482334793170473,
      "loss": 4.716,
      "step": 18466
    },
    {
      "epoch": 2.52903314160504,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004822939335073678,
      "loss": 4.7267,
      "step": 18467
    },
    {
      "epoch": 2.5291700903861956,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00048225307357663516,
      "loss": 4.6216,
      "step": 18468
    },
    {
      "epoch": 2.5293070391673513,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048221221337861354,
      "loss": 4.6771,
      "step": 18469
    },
    {
      "epoch": 2.529443987948507,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004821713529136416,
      "loss": 4.6907,
      "step": 18470
    },
    {
      "epoch": 2.529580936729663,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004821304921820582,
      "loss": 4.7178,
      "step": 18471
    },
    {
      "epoch": 2.529717885510819,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004820896311842015,
      "loss": 4.6702,
      "step": 18472
    },
    {
      "epoch": 2.529854834291975,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004820487699204104,
      "loss": 4.6656,
      "step": 18473
    },
    {
      "epoch": 2.5299917830731307,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00048200790839102365,
      "loss": 4.7437,
      "step": 18474
    },
    {
      "epoch": 2.5301287318542864,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004819670465963796,
      "loss": 4.7459,
      "step": 18475
    },
    {
      "epoch": 2.530265680635442,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004819261845368171,
      "loss": 4.7155,
      "step": 18476
    },
    {
      "epoch": 2.5304026294165984,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00048188532221267466,
      "loss": 4.7251,
      "step": 18477
    },
    {
      "epoch": 2.5305395781977538,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004818444596242911,
      "loss": 4.7867,
      "step": 18478
    },
    {
      "epoch": 2.53067652697891,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004818035967720047,
      "loss": 4.6369,
      "step": 18479
    },
    {
      "epoch": 2.5308134757600658,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00048176273365615444,
      "loss": 4.6515,
      "step": 18480
    },
    {
      "epoch": 2.5309504245412215,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00048172187027707876,
      "loss": 4.7169,
      "step": 18481
    },
    {
      "epoch": 2.5310873733223773,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048168100663511647,
      "loss": 4.6834,
      "step": 18482
    },
    {
      "epoch": 2.531224322103533,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048164014273060606,
      "loss": 4.6484,
      "step": 18483
    },
    {
      "epoch": 2.5313612708846893,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00048159927856388635,
      "loss": 4.7759,
      "step": 18484
    },
    {
      "epoch": 2.531498219665845,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048155841413529576,
      "loss": 4.7513,
      "step": 18485
    },
    {
      "epoch": 2.531635168447001,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004815175494451731,
      "loss": 4.6543,
      "step": 18486
    },
    {
      "epoch": 2.5317721172281566,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00048147668449385715,
      "loss": 4.6823,
      "step": 18487
    },
    {
      "epoch": 2.5319090660093124,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048143581928168617,
      "loss": 4.7475,
      "step": 18488
    },
    {
      "epoch": 2.532046014790468,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004813949538089993,
      "loss": 4.7659,
      "step": 18489
    },
    {
      "epoch": 2.532182963571624,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00048135408807613474,
      "loss": 4.7075,
      "step": 18490
    },
    {
      "epoch": 2.53231991235278,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048131322208343154,
      "loss": 4.6816,
      "step": 18491
    },
    {
      "epoch": 2.532456861133936,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004812723558312282,
      "loss": 4.7501,
      "step": 18492
    },
    {
      "epoch": 2.5325938099150918,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00048123148931986326,
      "loss": 4.8321,
      "step": 18493
    },
    {
      "epoch": 2.5327307586962475,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004811906225496757,
      "loss": 4.6658,
      "step": 18494
    },
    {
      "epoch": 2.5328677074774033,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004811497555210039,
      "loss": 4.7578,
      "step": 18495
    },
    {
      "epoch": 2.5330046562585595,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004811088882341867,
      "loss": 4.687,
      "step": 18496
    },
    {
      "epoch": 2.5331416050397153,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048106802068956277,
      "loss": 4.7832,
      "step": 18497
    },
    {
      "epoch": 2.533278553820871,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004810271528874707,
      "loss": 4.7581,
      "step": 18498
    },
    {
      "epoch": 2.533415502602027,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00048098628482824924,
      "loss": 4.6469,
      "step": 18499
    },
    {
      "epoch": 2.5335524513831826,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00048094541651223706,
      "loss": 4.6631,
      "step": 18500
    },
    {
      "epoch": 2.5336894001643384,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048090454793977275,
      "loss": 4.7545,
      "step": 18501
    },
    {
      "epoch": 2.533826348945494,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004808636791111951,
      "loss": 4.7033,
      "step": 18502
    },
    {
      "epoch": 2.5339632977266504,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00048082281002684296,
      "loss": 4.7103,
      "step": 18503
    },
    {
      "epoch": 2.534100246507806,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004807819406870547,
      "loss": 4.677,
      "step": 18504
    },
    {
      "epoch": 2.534237195288962,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004807410710921692,
      "loss": 4.6982,
      "step": 18505
    },
    {
      "epoch": 2.5343741440701177,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004807002012425251,
      "loss": 4.7306,
      "step": 18506
    },
    {
      "epoch": 2.5345110928512735,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00048065933113846114,
      "loss": 4.7857,
      "step": 18507
    },
    {
      "epoch": 2.5346480416324297,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048061846078031596,
      "loss": 4.7206,
      "step": 18508
    },
    {
      "epoch": 2.534784990413585,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004805775901684284,
      "loss": 4.6688,
      "step": 18509
    },
    {
      "epoch": 2.5349219391947413,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00048053671930313696,
      "loss": 4.6844,
      "step": 18510
    },
    {
      "epoch": 2.535058887975897,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00048049584818478044,
      "loss": 4.6839,
      "step": 18511
    },
    {
      "epoch": 2.535195836757053,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004804549768136976,
      "loss": 4.6713,
      "step": 18512
    },
    {
      "epoch": 2.5353327855382086,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004804141051902271,
      "loss": 4.7191,
      "step": 18513
    },
    {
      "epoch": 2.5354697343193644,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0004803732333147077,
      "loss": 4.7205,
      "step": 18514
    },
    {
      "epoch": 2.5356066831005206,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00048033236118747797,
      "loss": 4.703,
      "step": 18515
    },
    {
      "epoch": 2.5357436318816764,
      "grad_norm": 0.41796875,
      "learning_rate": 0.000480291488808877,
      "loss": 4.7202,
      "step": 18516
    },
    {
      "epoch": 2.535880580662832,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00048025061617924293,
      "loss": 4.7503,
      "step": 18517
    },
    {
      "epoch": 2.536017529443988,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000480209743298915,
      "loss": 4.7767,
      "step": 18518
    },
    {
      "epoch": 2.5361544782251437,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004801688701682315,
      "loss": 4.6865,
      "step": 18519
    },
    {
      "epoch": 2.5362914270062995,
      "grad_norm": 0.34375,
      "learning_rate": 0.00048012799678753164,
      "loss": 4.6503,
      "step": 18520
    },
    {
      "epoch": 2.5364283757874553,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004800871231571538,
      "loss": 4.6828,
      "step": 18521
    },
    {
      "epoch": 2.5365653245686115,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004800462492774368,
      "loss": 4.7479,
      "step": 18522
    },
    {
      "epoch": 2.5367022733497673,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004800053751487194,
      "loss": 4.6971,
      "step": 18523
    },
    {
      "epoch": 2.536839222130923,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00047996450077134024,
      "loss": 4.7851,
      "step": 18524
    },
    {
      "epoch": 2.536976170912079,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004799236261456381,
      "loss": 4.7543,
      "step": 18525
    },
    {
      "epoch": 2.5371131196932346,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00047988275127195176,
      "loss": 4.6897,
      "step": 18526
    },
    {
      "epoch": 2.537250068474391,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004798418761506201,
      "loss": 4.6936,
      "step": 18527
    },
    {
      "epoch": 2.537387017255546,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00047980100078198144,
      "loss": 4.6783,
      "step": 18528
    },
    {
      "epoch": 2.5375239660367024,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00047976012516637496,
      "loss": 4.6078,
      "step": 18529
    },
    {
      "epoch": 2.537660914817858,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047971924930413916,
      "loss": 4.7228,
      "step": 18530
    },
    {
      "epoch": 2.537797863599014,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0004796783731956129,
      "loss": 4.6109,
      "step": 18531
    },
    {
      "epoch": 2.5379348123801697,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047963749684113495,
      "loss": 4.7817,
      "step": 18532
    },
    {
      "epoch": 2.5380717611613255,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00047959662024104385,
      "loss": 4.7081,
      "step": 18533
    },
    {
      "epoch": 2.5382087099424817,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00047955574339567867,
      "loss": 4.7029,
      "step": 18534
    },
    {
      "epoch": 2.5383456587236375,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004795148663053778,
      "loss": 4.7494,
      "step": 18535
    },
    {
      "epoch": 2.5384826075047933,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00047947398897048044,
      "loss": 4.7556,
      "step": 18536
    },
    {
      "epoch": 2.538619556285949,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004794331113913249,
      "loss": 4.654,
      "step": 18537
    },
    {
      "epoch": 2.538756505067105,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004793922335682502,
      "loss": 4.6517,
      "step": 18538
    },
    {
      "epoch": 2.5388934538482606,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047935135550159516,
      "loss": 4.7028,
      "step": 18539
    },
    {
      "epoch": 2.5390304026294164,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004793104771916983,
      "loss": 4.6837,
      "step": 18540
    },
    {
      "epoch": 2.5391673514105726,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047926959863889876,
      "loss": 4.7078,
      "step": 18541
    },
    {
      "epoch": 2.5393043001917284,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004792287198435348,
      "loss": 4.6282,
      "step": 18542
    },
    {
      "epoch": 2.539441248972884,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004791878408059457,
      "loss": 4.6687,
      "step": 18543
    },
    {
      "epoch": 2.53957819775404,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047914696152646984,
      "loss": 4.7223,
      "step": 18544
    },
    {
      "epoch": 2.5397151465351957,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047910608200544636,
      "loss": 4.7064,
      "step": 18545
    },
    {
      "epoch": 2.539852095316352,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047906520224321364,
      "loss": 4.6957,
      "step": 18546
    },
    {
      "epoch": 2.5399890440975077,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047902432224011067,
      "loss": 4.7642,
      "step": 18547
    },
    {
      "epoch": 2.5401259928786635,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004789834419964764,
      "loss": 4.7023,
      "step": 18548
    },
    {
      "epoch": 2.5402629416598193,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004789425615126493,
      "loss": 4.7427,
      "step": 18549
    },
    {
      "epoch": 2.540399890440975,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004789016807889684,
      "loss": 4.7389,
      "step": 18550
    },
    {
      "epoch": 2.540536839222131,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047886079982577224,
      "loss": 4.6801,
      "step": 18551
    },
    {
      "epoch": 2.5406737880032866,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004788199186233999,
      "loss": 4.7429,
      "step": 18552
    },
    {
      "epoch": 2.540810736784443,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004787790371821899,
      "loss": 4.6419,
      "step": 18553
    },
    {
      "epoch": 2.5409476855655986,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004787381555024813,
      "loss": 4.6849,
      "step": 18554
    },
    {
      "epoch": 2.5410846343467544,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004786972735846126,
      "loss": 4.663,
      "step": 18555
    },
    {
      "epoch": 2.54122158312791,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00047865639142892296,
      "loss": 4.6506,
      "step": 18556
    },
    {
      "epoch": 2.541358531909066,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004786155090357508,
      "loss": 4.6887,
      "step": 18557
    },
    {
      "epoch": 2.541495480690222,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047857462640543507,
      "loss": 4.7011,
      "step": 18558
    },
    {
      "epoch": 2.5416324294713775,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004785337435383148,
      "loss": 4.6904,
      "step": 18559
    },
    {
      "epoch": 2.5417693782525337,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047849286043472836,
      "loss": 4.6754,
      "step": 18560
    },
    {
      "epoch": 2.5419063270336895,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000478451977095015,
      "loss": 4.7285,
      "step": 18561
    },
    {
      "epoch": 2.5420432758148452,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004784110935195132,
      "loss": 4.5787,
      "step": 18562
    },
    {
      "epoch": 2.542180224596001,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00047837020970856203,
      "loss": 4.7444,
      "step": 18563
    },
    {
      "epoch": 2.542317173377157,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004783293256625,
      "loss": 4.6701,
      "step": 18564
    },
    {
      "epoch": 2.542454122158313,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004782884413816662,
      "loss": 4.7026,
      "step": 18565
    },
    {
      "epoch": 2.542591070939469,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004782475568663993,
      "loss": 4.6737,
      "step": 18566
    },
    {
      "epoch": 2.5427280197206246,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00047820667211703814,
      "loss": 4.6253,
      "step": 18567
    },
    {
      "epoch": 2.5428649685017803,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004781657871339217,
      "loss": 4.7656,
      "step": 18568
    },
    {
      "epoch": 2.543001917282936,
      "grad_norm": 0.375,
      "learning_rate": 0.00047812490191738856,
      "loss": 4.689,
      "step": 18569
    },
    {
      "epoch": 2.543138866064092,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00047808401646777765,
      "loss": 4.7071,
      "step": 18570
    },
    {
      "epoch": 2.5432758148452477,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047804313078542783,
      "loss": 4.7141,
      "step": 18571
    },
    {
      "epoch": 2.543412763626404,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00047800224487067793,
      "loss": 4.7155,
      "step": 18572
    },
    {
      "epoch": 2.5435497124075597,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00047796135872386675,
      "loss": 4.6688,
      "step": 18573
    },
    {
      "epoch": 2.5436866611887154,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004779204723453331,
      "loss": 4.6448,
      "step": 18574
    },
    {
      "epoch": 2.5438236099698712,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047787958573541585,
      "loss": 4.6409,
      "step": 18575
    },
    {
      "epoch": 2.543960558751027,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004778386988944538,
      "loss": 4.736,
      "step": 18576
    },
    {
      "epoch": 2.5440975075321832,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047779781182278594,
      "loss": 4.7495,
      "step": 18577
    },
    {
      "epoch": 2.5442344563133386,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047775692452075083,
      "loss": 4.7333,
      "step": 18578
    },
    {
      "epoch": 2.5443714050944948,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004777160369886876,
      "loss": 4.6717,
      "step": 18579
    },
    {
      "epoch": 2.5445083538756506,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047767514922693484,
      "loss": 4.7109,
      "step": 18580
    },
    {
      "epoch": 2.5446453026568063,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047763426123583163,
      "loss": 4.7273,
      "step": 18581
    },
    {
      "epoch": 2.544782251437962,
      "grad_norm": 0.359375,
      "learning_rate": 0.00047759337301571663,
      "loss": 4.6686,
      "step": 18582
    },
    {
      "epoch": 2.544919200219118,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004775524845669289,
      "loss": 4.6418,
      "step": 18583
    },
    {
      "epoch": 2.545056149000274,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004775115958898071,
      "loss": 4.7467,
      "step": 18584
    },
    {
      "epoch": 2.54519309778143,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047747070698469,
      "loss": 4.6766,
      "step": 18585
    },
    {
      "epoch": 2.5453300465625857,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004774298178519168,
      "loss": 4.7483,
      "step": 18586
    },
    {
      "epoch": 2.5454669953437414,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004773889284918261,
      "loss": 4.7063,
      "step": 18587
    },
    {
      "epoch": 2.545603944124897,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004773480389047569,
      "loss": 4.7127,
      "step": 18588
    },
    {
      "epoch": 2.545740892906053,
      "grad_norm": 0.375,
      "learning_rate": 0.00047730714909104783,
      "loss": 4.7663,
      "step": 18589
    },
    {
      "epoch": 2.5458778416872088,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047726625905103806,
      "loss": 4.6024,
      "step": 18590
    },
    {
      "epoch": 2.546014790468365,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004772253687850662,
      "loss": 4.7151,
      "step": 18591
    },
    {
      "epoch": 2.5461517392495208,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00047718447829347125,
      "loss": 4.7235,
      "step": 18592
    },
    {
      "epoch": 2.5462886880306765,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000477143587576592,
      "loss": 4.6612,
      "step": 18593
    },
    {
      "epoch": 2.5464256368118323,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004771026966347675,
      "loss": 4.7176,
      "step": 18594
    },
    {
      "epoch": 2.546562585592988,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047706180546833643,
      "loss": 4.6975,
      "step": 18595
    },
    {
      "epoch": 2.5466995343741443,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004770209140776376,
      "loss": 4.7388,
      "step": 18596
    },
    {
      "epoch": 2.5468364831552996,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00047698002246301024,
      "loss": 4.6896,
      "step": 18597
    },
    {
      "epoch": 2.546973431936456,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004769391306247929,
      "loss": 4.6912,
      "step": 18598
    },
    {
      "epoch": 2.5471103807176116,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00047689823856332454,
      "loss": 4.7052,
      "step": 18599
    },
    {
      "epoch": 2.5472473294987674,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047685734627894407,
      "loss": 4.7044,
      "step": 18600
    },
    {
      "epoch": 2.547384278279923,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004768164537719905,
      "loss": 4.6827,
      "step": 18601
    },
    {
      "epoch": 2.547521227061079,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00047677556104280247,
      "loss": 4.6648,
      "step": 18602
    },
    {
      "epoch": 2.547658175842235,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004767346680917189,
      "loss": 4.6599,
      "step": 18603
    },
    {
      "epoch": 2.547795124623391,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004766937749190789,
      "loss": 4.7478,
      "step": 18604
    },
    {
      "epoch": 2.5479320734045467,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047665288152522114,
      "loss": 4.6325,
      "step": 18605
    },
    {
      "epoch": 2.5480690221857025,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004766119879104848,
      "loss": 4.7358,
      "step": 18606
    },
    {
      "epoch": 2.5482059709668583,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004765710940752083,
      "loss": 4.6924,
      "step": 18607
    },
    {
      "epoch": 2.548342919748014,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000476530200019731,
      "loss": 4.623,
      "step": 18608
    },
    {
      "epoch": 2.54847986852917,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00047648930574439147,
      "loss": 4.7344,
      "step": 18609
    },
    {
      "epoch": 2.548616817310326,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004764484112495289,
      "loss": 4.7043,
      "step": 18610
    },
    {
      "epoch": 2.548753766091482,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004764075165354819,
      "loss": 4.6247,
      "step": 18611
    },
    {
      "epoch": 2.5488907148726376,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00047636662160258964,
      "loss": 4.6533,
      "step": 18612
    },
    {
      "epoch": 2.5490276636537934,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004763257264511909,
      "loss": 4.7302,
      "step": 18613
    },
    {
      "epoch": 2.549164612434949,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004762848310816244,
      "loss": 4.6519,
      "step": 18614
    },
    {
      "epoch": 2.5493015612161054,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047624393549422955,
      "loss": 4.6336,
      "step": 18615
    },
    {
      "epoch": 2.549438509997261,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004762030396893447,
      "loss": 4.6639,
      "step": 18616
    },
    {
      "epoch": 2.549575458778417,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00047616214366730914,
      "loss": 4.6851,
      "step": 18617
    },
    {
      "epoch": 2.5497124075595727,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00047612124742846165,
      "loss": 4.7071,
      "step": 18618
    },
    {
      "epoch": 2.5498493563407285,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004760803509731411,
      "loss": 4.7045,
      "step": 18619
    },
    {
      "epoch": 2.5499863051218843,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004760394543016865,
      "loss": 4.7003,
      "step": 18620
    },
    {
      "epoch": 2.55012325390304,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00047599855741443687,
      "loss": 4.6845,
      "step": 18621
    },
    {
      "epoch": 2.5502602026841963,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047595766031173075,
      "loss": 4.7361,
      "step": 18622
    },
    {
      "epoch": 2.550397151465352,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004759167629939075,
      "loss": 4.7102,
      "step": 18623
    },
    {
      "epoch": 2.550534100246508,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00047587586546130576,
      "loss": 4.7283,
      "step": 18624
    },
    {
      "epoch": 2.5506710490276636,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004758349677142646,
      "loss": 4.7624,
      "step": 18625
    },
    {
      "epoch": 2.5508079978088194,
      "grad_norm": 0.390625,
      "learning_rate": 0.0004757940697531229,
      "loss": 4.6492,
      "step": 18626
    },
    {
      "epoch": 2.5509449465899756,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00047575317157821955,
      "loss": 4.6279,
      "step": 18627
    },
    {
      "epoch": 2.551081895371131,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00047571227318989366,
      "loss": 4.6736,
      "step": 18628
    },
    {
      "epoch": 2.551218844152287,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004756713745884839,
      "loss": 4.7335,
      "step": 18629
    },
    {
      "epoch": 2.551355792933443,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047563047577432944,
      "loss": 4.6761,
      "step": 18630
    },
    {
      "epoch": 2.5514927417145987,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000475589576747769,
      "loss": 4.6353,
      "step": 18631
    },
    {
      "epoch": 2.5516296904957545,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004755486775091417,
      "loss": 4.6511,
      "step": 18632
    },
    {
      "epoch": 2.5517666392769103,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004755077780587865,
      "loss": 4.778,
      "step": 18633
    },
    {
      "epoch": 2.5519035880580665,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00047546687839704213,
      "loss": 4.693,
      "step": 18634
    },
    {
      "epoch": 2.5520405368392223,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004754259785242478,
      "loss": 4.7435,
      "step": 18635
    },
    {
      "epoch": 2.552177485620378,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004753850784407422,
      "loss": 4.7037,
      "step": 18636
    },
    {
      "epoch": 2.552314434401534,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004753441781468645,
      "loss": 4.6541,
      "step": 18637
    },
    {
      "epoch": 2.5524513831826896,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004753032776429535,
      "loss": 4.7342,
      "step": 18638
    },
    {
      "epoch": 2.5525883319638454,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047526237692934836,
      "loss": 4.7079,
      "step": 18639
    },
    {
      "epoch": 2.552725280745001,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00047522147600638764,
      "loss": 4.6603,
      "step": 18640
    },
    {
      "epoch": 2.5528622295261574,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004751805748744107,
      "loss": 4.6772,
      "step": 18641
    },
    {
      "epoch": 2.552999178307313,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004751396735337563,
      "loss": 4.6977,
      "step": 18642
    },
    {
      "epoch": 2.553136127088469,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047509877198476336,
      "loss": 4.6491,
      "step": 18643
    },
    {
      "epoch": 2.5532730758696247,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004750578702277711,
      "loss": 4.6381,
      "step": 18644
    },
    {
      "epoch": 2.5534100246507805,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004750169682631181,
      "loss": 4.6709,
      "step": 18645
    },
    {
      "epoch": 2.5535469734319367,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004749760660911437,
      "loss": 4.6837,
      "step": 18646
    },
    {
      "epoch": 2.553683922213092,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004749351637121865,
      "loss": 4.6445,
      "step": 18647
    },
    {
      "epoch": 2.5538208709942483,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047489426112658575,
      "loss": 4.7281,
      "step": 18648
    },
    {
      "epoch": 2.553957819775404,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00047485335833468036,
      "loss": 4.7076,
      "step": 18649
    },
    {
      "epoch": 2.55409476855656,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00047481245533680924,
      "loss": 4.6737,
      "step": 18650
    },
    {
      "epoch": 2.5542317173377156,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004747715521333114,
      "loss": 4.7124,
      "step": 18651
    },
    {
      "epoch": 2.5543686661188714,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047473064872452573,
      "loss": 4.5924,
      "step": 18652
    },
    {
      "epoch": 2.5545056149000276,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004746897451107914,
      "loss": 4.6658,
      "step": 18653
    },
    {
      "epoch": 2.5546425636811834,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004746488412924472,
      "loss": 4.6823,
      "step": 18654
    },
    {
      "epoch": 2.554779512462339,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00047460793726983205,
      "loss": 4.689,
      "step": 18655
    },
    {
      "epoch": 2.554916461243495,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00047456703304328535,
      "loss": 4.6911,
      "step": 18656
    },
    {
      "epoch": 2.5550534100246507,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00047452612861314554,
      "loss": 4.6057,
      "step": 18657
    },
    {
      "epoch": 2.5551903588058065,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004744852239797521,
      "loss": 4.7393,
      "step": 18658
    },
    {
      "epoch": 2.5553273075869622,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004744443191434435,
      "loss": 4.7679,
      "step": 18659
    },
    {
      "epoch": 2.5554642563681185,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004744034141045592,
      "loss": 4.6912,
      "step": 18660
    },
    {
      "epoch": 2.5556012051492742,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004743625088634379,
      "loss": 4.7534,
      "step": 18661
    },
    {
      "epoch": 2.55573815393043,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00047432160342041865,
      "loss": 4.7026,
      "step": 18662
    },
    {
      "epoch": 2.555875102711586,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00047428069777584057,
      "loss": 4.6389,
      "step": 18663
    },
    {
      "epoch": 2.5560120514927416,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00047423979193004254,
      "loss": 4.6786,
      "step": 18664
    },
    {
      "epoch": 2.556149000273898,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00047419888588336363,
      "loss": 4.6191,
      "step": 18665
    },
    {
      "epoch": 2.5562859490550536,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00047415797963614267,
      "loss": 4.7271,
      "step": 18666
    },
    {
      "epoch": 2.5564228978362094,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004741170731887189,
      "loss": 4.8147,
      "step": 18667
    },
    {
      "epoch": 2.556559846617365,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00047407616654143107,
      "loss": 4.7074,
      "step": 18668
    },
    {
      "epoch": 2.556696795398521,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004740352596946184,
      "loss": 4.5799,
      "step": 18669
    },
    {
      "epoch": 2.5568337441796767,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004739943526486198,
      "loss": 4.7055,
      "step": 18670
    },
    {
      "epoch": 2.5569706929608325,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004739534454037742,
      "loss": 4.6783,
      "step": 18671
    },
    {
      "epoch": 2.5571076417419887,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004739125379604208,
      "loss": 4.6572,
      "step": 18672
    },
    {
      "epoch": 2.5572445905231445,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047387163031889837,
      "loss": 4.6807,
      "step": 18673
    },
    {
      "epoch": 2.5573815393043002,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004738307224795462,
      "loss": 4.6385,
      "step": 18674
    },
    {
      "epoch": 2.557518488085456,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000473789814442703,
      "loss": 4.7218,
      "step": 18675
    },
    {
      "epoch": 2.557655436866612,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004737489062087081,
      "loss": 4.7401,
      "step": 18676
    },
    {
      "epoch": 2.557792385647768,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004737079977779002,
      "loss": 4.708,
      "step": 18677
    },
    {
      "epoch": 2.5579293344289233,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047366708915061857,
      "loss": 4.6801,
      "step": 18678
    },
    {
      "epoch": 2.5580662832100796,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047362618032720217,
      "loss": 4.665,
      "step": 18679
    },
    {
      "epoch": 2.5582032319912353,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004735852713079899,
      "loss": 4.7014,
      "step": 18680
    },
    {
      "epoch": 2.558340180772391,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004735443620933209,
      "loss": 4.7364,
      "step": 18681
    },
    {
      "epoch": 2.558477129553547,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00047350345268353417,
      "loss": 4.6562,
      "step": 18682
    },
    {
      "epoch": 2.5586140783347027,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004734625430789686,
      "loss": 4.655,
      "step": 18683
    },
    {
      "epoch": 2.558751027115859,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00047342163327996354,
      "loss": 4.7209,
      "step": 18684
    },
    {
      "epoch": 2.5588879758970147,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004733807232868577,
      "loss": 4.6408,
      "step": 18685
    },
    {
      "epoch": 2.5590249246781704,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00047333981309999036,
      "loss": 4.7113,
      "step": 18686
    },
    {
      "epoch": 2.559161873459326,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047329890271970035,
      "loss": 4.6708,
      "step": 18687
    },
    {
      "epoch": 2.559298822240482,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00047325799214632667,
      "loss": 4.6854,
      "step": 18688
    },
    {
      "epoch": 2.5594357710216378,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047321708138020857,
      "loss": 4.7518,
      "step": 18689
    },
    {
      "epoch": 2.5595727198027936,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000473176170421685,
      "loss": 4.7589,
      "step": 18690
    },
    {
      "epoch": 2.5597096685839498,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00047313525927109506,
      "loss": 4.6863,
      "step": 18691
    },
    {
      "epoch": 2.5598466173651055,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0004730943479287775,
      "loss": 4.6653,
      "step": 18692
    },
    {
      "epoch": 2.5599835661462613,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004730534363950718,
      "loss": 4.5801,
      "step": 18693
    },
    {
      "epoch": 2.560120514927417,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047301252467031666,
      "loss": 4.7042,
      "step": 18694
    },
    {
      "epoch": 2.560257463708573,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047297161275485124,
      "loss": 4.6564,
      "step": 18695
    },
    {
      "epoch": 2.560394412489729,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00047293070064901465,
      "loss": 4.6849,
      "step": 18696
    },
    {
      "epoch": 2.5605313612708844,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004728897883531457,
      "loss": 4.6347,
      "step": 18697
    },
    {
      "epoch": 2.5606683100520407,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047284887586758386,
      "loss": 4.6957,
      "step": 18698
    },
    {
      "epoch": 2.5608052588331964,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004728079631926678,
      "loss": 4.7329,
      "step": 18699
    },
    {
      "epoch": 2.560942207614352,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00047276705032873686,
      "loss": 4.7207,
      "step": 18700
    },
    {
      "epoch": 2.561079156395508,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004727261372761297,
      "loss": 4.6851,
      "step": 18701
    },
    {
      "epoch": 2.5612161051766638,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00047268522403518594,
      "loss": 4.7001,
      "step": 18702
    },
    {
      "epoch": 2.56135305395782,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00047264431060624405,
      "loss": 4.7613,
      "step": 18703
    },
    {
      "epoch": 2.5614900027389758,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00047260339698964357,
      "loss": 4.7194,
      "step": 18704
    },
    {
      "epoch": 2.5616269515201315,
      "grad_norm": 0.328125,
      "learning_rate": 0.00047256248318572314,
      "loss": 4.7896,
      "step": 18705
    },
    {
      "epoch": 2.5617639003012873,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047252156919482213,
      "loss": 4.7337,
      "step": 18706
    },
    {
      "epoch": 2.561900849082443,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00047248065501727957,
      "loss": 4.7082,
      "step": 18707
    },
    {
      "epoch": 2.562037797863599,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047243974065343437,
      "loss": 4.6215,
      "step": 18708
    },
    {
      "epoch": 2.5621747466447546,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047239882610362584,
      "loss": 4.6779,
      "step": 18709
    },
    {
      "epoch": 2.562311695425911,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004723579113681927,
      "loss": 4.6799,
      "step": 18710
    },
    {
      "epoch": 2.5624486442070666,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004723169964474744,
      "loss": 4.6941,
      "step": 18711
    },
    {
      "epoch": 2.5625855929882224,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047227608134180974,
      "loss": 4.7033,
      "step": 18712
    },
    {
      "epoch": 2.562722541769378,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004722351660515379,
      "loss": 4.6563,
      "step": 18713
    },
    {
      "epoch": 2.562859490550534,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00047219425057699795,
      "loss": 4.7472,
      "step": 18714
    },
    {
      "epoch": 2.56299643933169,
      "grad_norm": 0.34375,
      "learning_rate": 0.00047215333491852884,
      "loss": 4.717,
      "step": 18715
    },
    {
      "epoch": 2.5631333881128455,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00047211241907647,
      "loss": 4.7125,
      "step": 18716
    },
    {
      "epoch": 2.5632703368940017,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004720715030511601,
      "loss": 4.6986,
      "step": 18717
    },
    {
      "epoch": 2.5634072856751575,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00047203058684293847,
      "loss": 4.6519,
      "step": 18718
    },
    {
      "epoch": 2.5635442344563133,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004719896704521441,
      "loss": 4.7814,
      "step": 18719
    },
    {
      "epoch": 2.563681183237469,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004719487538791161,
      "loss": 4.6818,
      "step": 18720
    },
    {
      "epoch": 2.563818132018625,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00047190783712419346,
      "loss": 4.6701,
      "step": 18721
    },
    {
      "epoch": 2.563955080799781,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00047186692018771545,
      "loss": 4.6643,
      "step": 18722
    },
    {
      "epoch": 2.564092029580937,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000471826003070021,
      "loss": 4.5563,
      "step": 18723
    },
    {
      "epoch": 2.5642289783620926,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00047178508577144936,
      "loss": 4.5861,
      "step": 18724
    },
    {
      "epoch": 2.5643659271432484,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004717441682923394,
      "loss": 4.6764,
      "step": 18725
    },
    {
      "epoch": 2.564502875924404,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004717032506330304,
      "loss": 4.717,
      "step": 18726
    },
    {
      "epoch": 2.56463982470556,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004716623327938614,
      "loss": 4.761,
      "step": 18727
    },
    {
      "epoch": 2.5647767734867157,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00047162141477517143,
      "loss": 4.6575,
      "step": 18728
    },
    {
      "epoch": 2.564913722267872,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00047158049657729977,
      "loss": 4.728,
      "step": 18729
    },
    {
      "epoch": 2.5650506710490277,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004715395782005852,
      "loss": 4.6675,
      "step": 18730
    },
    {
      "epoch": 2.5651876198301835,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004714986596453673,
      "loss": 4.6667,
      "step": 18731
    },
    {
      "epoch": 2.5653245686113393,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00047145774091198465,
      "loss": 4.7097,
      "step": 18732
    },
    {
      "epoch": 2.565461517392495,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00047141682200077664,
      "loss": 4.7011,
      "step": 18733
    },
    {
      "epoch": 2.5655984661736513,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047137590291208224,
      "loss": 4.728,
      "step": 18734
    },
    {
      "epoch": 2.565735414954807,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0004713349836462407,
      "loss": 4.6898,
      "step": 18735
    },
    {
      "epoch": 2.565872363735963,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00047129406420359115,
      "loss": 4.6667,
      "step": 18736
    },
    {
      "epoch": 2.5660093125171186,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004712531445844725,
      "loss": 4.651,
      "step": 18737
    },
    {
      "epoch": 2.5661462612982744,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004712122247892241,
      "loss": 4.7003,
      "step": 18738
    },
    {
      "epoch": 2.56628321007943,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004711713048181848,
      "loss": 4.6827,
      "step": 18739
    },
    {
      "epoch": 2.566420158860586,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000471130384671694,
      "loss": 4.7872,
      "step": 18740
    },
    {
      "epoch": 2.566557107641742,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047108946435009055,
      "loss": 4.6963,
      "step": 18741
    },
    {
      "epoch": 2.566694056422898,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00047104854385371375,
      "loss": 4.7198,
      "step": 18742
    },
    {
      "epoch": 2.5668310052040537,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004710076231829026,
      "loss": 4.6312,
      "step": 18743
    },
    {
      "epoch": 2.5669679539852095,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004709667023379963,
      "loss": 4.7388,
      "step": 18744
    },
    {
      "epoch": 2.5671049027663653,
      "grad_norm": 0.34375,
      "learning_rate": 0.000470925781319334,
      "loss": 4.6635,
      "step": 18745
    },
    {
      "epoch": 2.5672418515475215,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00047088486012725466,
      "loss": 4.667,
      "step": 18746
    },
    {
      "epoch": 2.567378800328677,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0004708439387620976,
      "loss": 4.7447,
      "step": 18747
    },
    {
      "epoch": 2.567515749109833,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00047080301722420177,
      "loss": 4.7999,
      "step": 18748
    },
    {
      "epoch": 2.567652697890989,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004707620955139064,
      "loss": 4.7325,
      "step": 18749
    },
    {
      "epoch": 2.5677896466721446,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004707211736315506,
      "loss": 4.6822,
      "step": 18750
    },
    {
      "epoch": 2.5679265954533004,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00047068025157747355,
      "loss": 4.7302,
      "step": 18751
    },
    {
      "epoch": 2.568063544234456,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00047063932935201427,
      "loss": 4.6774,
      "step": 18752
    },
    {
      "epoch": 2.5682004930156124,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004705984069555119,
      "loss": 4.6871,
      "step": 18753
    },
    {
      "epoch": 2.568337441796768,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00047055748438830575,
      "loss": 4.7325,
      "step": 18754
    },
    {
      "epoch": 2.568474390577924,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00047051656165073474,
      "loss": 4.6763,
      "step": 18755
    },
    {
      "epoch": 2.5686113393590797,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004704756387431381,
      "loss": 4.718,
      "step": 18756
    },
    {
      "epoch": 2.5687482881402355,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00047043471566585496,
      "loss": 4.7468,
      "step": 18757
    },
    {
      "epoch": 2.5688852369213913,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004703937924192245,
      "loss": 4.6449,
      "step": 18758
    },
    {
      "epoch": 2.569022185702547,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00047035286900358583,
      "loss": 4.7593,
      "step": 18759
    },
    {
      "epoch": 2.5691591344837033,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00047031194541927813,
      "loss": 4.7551,
      "step": 18760
    },
    {
      "epoch": 2.569296083264859,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00047027102166664035,
      "loss": 4.6632,
      "step": 18761
    },
    {
      "epoch": 2.569433032046015,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00047023009774601183,
      "loss": 4.7215,
      "step": 18762
    },
    {
      "epoch": 2.5695699808271706,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00047018917365773173,
      "loss": 4.7072,
      "step": 18763
    },
    {
      "epoch": 2.5697069296083264,
      "grad_norm": 0.32421875,
      "learning_rate": 0.000470148249402139,
      "loss": 4.7318,
      "step": 18764
    },
    {
      "epoch": 2.5698438783894826,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004701073249795732,
      "loss": 4.6026,
      "step": 18765
    },
    {
      "epoch": 2.569980827170638,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004700664003903729,
      "loss": 4.6904,
      "step": 18766
    },
    {
      "epoch": 2.570117775951794,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0004700254756348778,
      "loss": 4.699,
      "step": 18767
    },
    {
      "epoch": 2.57025472473295,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004699845507134267,
      "loss": 4.7739,
      "step": 18768
    },
    {
      "epoch": 2.5703916735141057,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046994362562635883,
      "loss": 4.6897,
      "step": 18769
    },
    {
      "epoch": 2.5705286222952615,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00046990270037401347,
      "loss": 4.7497,
      "step": 18770
    },
    {
      "epoch": 2.5706655710764172,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004698617749567296,
      "loss": 4.6691,
      "step": 18771
    },
    {
      "epoch": 2.5708025198575735,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004698208493748466,
      "loss": 4.7233,
      "step": 18772
    },
    {
      "epoch": 2.5709394686387292,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004697799236287033,
      "loss": 4.7145,
      "step": 18773
    },
    {
      "epoch": 2.571076417419885,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004697389977186393,
      "loss": 4.6398,
      "step": 18774
    },
    {
      "epoch": 2.571213366201041,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00046969807164499327,
      "loss": 4.6583,
      "step": 18775
    },
    {
      "epoch": 2.5713503149821966,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00046965714540810483,
      "loss": 4.6405,
      "step": 18776
    },
    {
      "epoch": 2.5714872637633523,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004696162190083129,
      "loss": 4.6714,
      "step": 18777
    },
    {
      "epoch": 2.571624212544508,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004695752924459567,
      "loss": 4.7914,
      "step": 18778
    },
    {
      "epoch": 2.5717611613256643,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00046953436572137524,
      "loss": 4.716,
      "step": 18779
    },
    {
      "epoch": 2.57189811010682,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000469493438834908,
      "loss": 4.7192,
      "step": 18780
    },
    {
      "epoch": 2.572035058887976,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004694525117868939,
      "loss": 4.6662,
      "step": 18781
    },
    {
      "epoch": 2.5721720076691317,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004694115845776721,
      "loss": 4.6701,
      "step": 18782
    },
    {
      "epoch": 2.5723089564502875,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004693706572075821,
      "loss": 4.6905,
      "step": 18783
    },
    {
      "epoch": 2.5724459052314437,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00046932972967696267,
      "loss": 4.7107,
      "step": 18784
    },
    {
      "epoch": 2.5725828540125995,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004692888019861533,
      "loss": 4.7117,
      "step": 18785
    },
    {
      "epoch": 2.5727198027937552,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000469247874135493,
      "loss": 4.701,
      "step": 18786
    },
    {
      "epoch": 2.572856751574911,
      "grad_norm": 0.37109375,
      "learning_rate": 0.000469206946125321,
      "loss": 4.681,
      "step": 18787
    },
    {
      "epoch": 2.572993700356067,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046916601795597636,
      "loss": 4.6639,
      "step": 18788
    },
    {
      "epoch": 2.5731306491372226,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004691250896277985,
      "loss": 4.7214,
      "step": 18789
    },
    {
      "epoch": 2.5732675979183783,
      "grad_norm": 0.359375,
      "learning_rate": 0.00046908416114112625,
      "loss": 4.7329,
      "step": 18790
    },
    {
      "epoch": 2.5734045466995346,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00046904323249629917,
      "loss": 4.6955,
      "step": 18791
    },
    {
      "epoch": 2.5735414954806903,
      "grad_norm": 0.375,
      "learning_rate": 0.0004690023036936563,
      "loss": 4.7228,
      "step": 18792
    },
    {
      "epoch": 2.573678444261846,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004689613747335367,
      "loss": 4.6656,
      "step": 18793
    },
    {
      "epoch": 2.573815393043002,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00046892044561627983,
      "loss": 4.6786,
      "step": 18794
    },
    {
      "epoch": 2.5739523418241577,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004688795163422245,
      "loss": 4.7284,
      "step": 18795
    },
    {
      "epoch": 2.574089290605314,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004688385869117103,
      "loss": 4.7232,
      "step": 18796
    },
    {
      "epoch": 2.574226239386469,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004687976573250762,
      "loss": 4.6442,
      "step": 18797
    },
    {
      "epoch": 2.5743631881676254,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046875672758266146,
      "loss": 4.618,
      "step": 18798
    },
    {
      "epoch": 2.574500136948781,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004687157976848051,
      "loss": 4.6942,
      "step": 18799
    },
    {
      "epoch": 2.574637085729937,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004686748676318466,
      "loss": 4.6923,
      "step": 18800
    },
    {
      "epoch": 2.5747740345110928,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00046863393742412496,
      "loss": 4.6267,
      "step": 18801
    },
    {
      "epoch": 2.5749109832922485,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004685930070619795,
      "loss": 4.7767,
      "step": 18802
    },
    {
      "epoch": 2.5750479320734048,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00046855207654574935,
      "loss": 4.7141,
      "step": 18803
    },
    {
      "epoch": 2.5751848808545605,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004685111458757737,
      "loss": 4.6805,
      "step": 18804
    },
    {
      "epoch": 2.5753218296357163,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00046847021505239177,
      "loss": 4.6904,
      "step": 18805
    },
    {
      "epoch": 2.575458778416872,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004684292840759428,
      "loss": 4.6796,
      "step": 18806
    },
    {
      "epoch": 2.575595727198028,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00046838835294676597,
      "loss": 4.8089,
      "step": 18807
    },
    {
      "epoch": 2.5757326759791837,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00046834742166520046,
      "loss": 4.7464,
      "step": 18808
    },
    {
      "epoch": 2.5758696247603394,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046830649023158535,
      "loss": 4.6908,
      "step": 18809
    },
    {
      "epoch": 2.5760065735414956,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004682655586462602,
      "loss": 4.6599,
      "step": 18810
    },
    {
      "epoch": 2.5761435223226514,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004682246269095639,
      "loss": 4.627,
      "step": 18811
    },
    {
      "epoch": 2.576280471103807,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004681836950218359,
      "loss": 4.7283,
      "step": 18812
    },
    {
      "epoch": 2.576417419884963,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046814276298341506,
      "loss": 4.6701,
      "step": 18813
    },
    {
      "epoch": 2.5765543686661188,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046810183079464105,
      "loss": 4.6378,
      "step": 18814
    },
    {
      "epoch": 2.576691317447275,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0004680608984558527,
      "loss": 4.728,
      "step": 18815
    },
    {
      "epoch": 2.5768282662284303,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004680199659673895,
      "loss": 4.6806,
      "step": 18816
    },
    {
      "epoch": 2.5769652150095865,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00046797903332959043,
      "loss": 4.7647,
      "step": 18817
    },
    {
      "epoch": 2.5771021637907423,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004679381005427949,
      "loss": 4.6829,
      "step": 18818
    },
    {
      "epoch": 2.577239112571898,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000467897167607342,
      "loss": 4.7791,
      "step": 18819
    },
    {
      "epoch": 2.577376061353054,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046785623452357103,
      "loss": 4.6673,
      "step": 18820
    },
    {
      "epoch": 2.5775130101342096,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004678153012918213,
      "loss": 4.6016,
      "step": 18821
    },
    {
      "epoch": 2.577649958915366,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00046777436791243176,
      "loss": 4.6728,
      "step": 18822
    },
    {
      "epoch": 2.5777869076965216,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004677334343857419,
      "loss": 4.6639,
      "step": 18823
    },
    {
      "epoch": 2.5779238564776774,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004676925007120908,
      "loss": 4.788,
      "step": 18824
    },
    {
      "epoch": 2.578060805258833,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004676515668918178,
      "loss": 4.7291,
      "step": 18825
    },
    {
      "epoch": 2.578197754039989,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046761063292526193,
      "loss": 4.5848,
      "step": 18826
    },
    {
      "epoch": 2.5783347028211447,
      "grad_norm": 0.328125,
      "learning_rate": 0.00046756969881276263,
      "loss": 4.6617,
      "step": 18827
    },
    {
      "epoch": 2.5784716516023005,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004675287645546591,
      "loss": 4.6741,
      "step": 18828
    },
    {
      "epoch": 2.5786086003834567,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004674878301512905,
      "loss": 4.7128,
      "step": 18829
    },
    {
      "epoch": 2.5787455491646125,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00046744689560299606,
      "loss": 4.7751,
      "step": 18830
    },
    {
      "epoch": 2.5788824979457683,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000467405960910115,
      "loss": 4.7517,
      "step": 18831
    },
    {
      "epoch": 2.579019446726924,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004673650260729866,
      "loss": 4.6969,
      "step": 18832
    },
    {
      "epoch": 2.57915639550808,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00046732409109195004,
      "loss": 4.7267,
      "step": 18833
    },
    {
      "epoch": 2.579293344289236,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00046728315596734476,
      "loss": 4.7245,
      "step": 18834
    },
    {
      "epoch": 2.5794302930703914,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004672422206995097,
      "loss": 4.6485,
      "step": 18835
    },
    {
      "epoch": 2.5795672418515476,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00046720128528878434,
      "loss": 4.693,
      "step": 18836
    },
    {
      "epoch": 2.5797041906327034,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00046716034973550785,
      "loss": 4.6422,
      "step": 18837
    },
    {
      "epoch": 2.579841139413859,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004671194140400193,
      "loss": 4.7437,
      "step": 18838
    },
    {
      "epoch": 2.579978088195015,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00046707847820265836,
      "loss": 4.7305,
      "step": 18839
    },
    {
      "epoch": 2.5801150369761707,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00046703754222376367,
      "loss": 4.717,
      "step": 18840
    },
    {
      "epoch": 2.580251985757327,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004669966061036751,
      "loss": 4.6734,
      "step": 18841
    },
    {
      "epoch": 2.5803889345384827,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00046695566984273143,
      "loss": 4.7019,
      "step": 18842
    },
    {
      "epoch": 2.5805258833196385,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004669147334412722,
      "loss": 4.7381,
      "step": 18843
    },
    {
      "epoch": 2.5806628321007943,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00046687379689963637,
      "loss": 4.6679,
      "step": 18844
    },
    {
      "epoch": 2.58079978088195,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00046683286021816355,
      "loss": 4.7382,
      "step": 18845
    },
    {
      "epoch": 2.580936729663106,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004667919233971927,
      "loss": 4.6839,
      "step": 18846
    },
    {
      "epoch": 2.5810736784442616,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046675098643706315,
      "loss": 4.7018,
      "step": 18847
    },
    {
      "epoch": 2.581210627225418,
      "grad_norm": 0.328125,
      "learning_rate": 0.00046671004933811433,
      "loss": 4.758,
      "step": 18848
    },
    {
      "epoch": 2.5813475760065736,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004666691121006852,
      "loss": 4.681,
      "step": 18849
    },
    {
      "epoch": 2.5814845247877294,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004666281747251153,
      "loss": 4.7579,
      "step": 18850
    },
    {
      "epoch": 2.581621473568885,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004665872372117436,
      "loss": 4.6982,
      "step": 18851
    },
    {
      "epoch": 2.581758422350041,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00046654629956090964,
      "loss": 4.7223,
      "step": 18852
    },
    {
      "epoch": 2.581895371131197,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004665053617729525,
      "loss": 4.7424,
      "step": 18853
    },
    {
      "epoch": 2.582032319912353,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046646442384821156,
      "loss": 4.7165,
      "step": 18854
    },
    {
      "epoch": 2.5821692686935087,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004664234857870259,
      "loss": 4.6876,
      "step": 18855
    },
    {
      "epoch": 2.5823062174746645,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046638254758973484,
      "loss": 4.6596,
      "step": 18856
    },
    {
      "epoch": 2.5824431662558203,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046634160925667793,
      "loss": 4.6946,
      "step": 18857
    },
    {
      "epoch": 2.582580115036976,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00046630067078819404,
      "loss": 4.7492,
      "step": 18858
    },
    {
      "epoch": 2.582717063818132,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00046625973218462273,
      "loss": 4.6459,
      "step": 18859
    },
    {
      "epoch": 2.582854012599288,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004662187934463031,
      "loss": 4.7161,
      "step": 18860
    },
    {
      "epoch": 2.582990961380444,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00046617785457357447,
      "loss": 4.7096,
      "step": 18861
    },
    {
      "epoch": 2.5831279101615996,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00046613691556677596,
      "loss": 4.77,
      "step": 18862
    },
    {
      "epoch": 2.5832648589427554,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004660959764262472,
      "loss": 4.7056,
      "step": 18863
    },
    {
      "epoch": 2.583401807723911,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046605503715232705,
      "loss": 4.7044,
      "step": 18864
    },
    {
      "epoch": 2.5835387565050674,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004660140977453551,
      "loss": 4.7095,
      "step": 18865
    },
    {
      "epoch": 2.5836757052862227,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004659731582056706,
      "loss": 4.6913,
      "step": 18866
    },
    {
      "epoch": 2.583812654067379,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00046593221853361255,
      "loss": 4.7591,
      "step": 18867
    },
    {
      "epoch": 2.5839496028485347,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00046589127872952055,
      "loss": 4.7238,
      "step": 18868
    },
    {
      "epoch": 2.5840865516296905,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004658503387937336,
      "loss": 4.7757,
      "step": 18869
    },
    {
      "epoch": 2.5842235004108463,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00046580939872659116,
      "loss": 4.6593,
      "step": 18870
    },
    {
      "epoch": 2.584360449192002,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004657684585284325,
      "loss": 4.681,
      "step": 18871
    },
    {
      "epoch": 2.5844973979731583,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00046572751819959693,
      "loss": 4.7105,
      "step": 18872
    },
    {
      "epoch": 2.584634346754314,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004656865777404235,
      "loss": 4.741,
      "step": 18873
    },
    {
      "epoch": 2.58477129553547,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004656456371512517,
      "loss": 4.7832,
      "step": 18874
    },
    {
      "epoch": 2.5849082443166256,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046560469643242095,
      "loss": 4.6809,
      "step": 18875
    },
    {
      "epoch": 2.5850451930977814,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004655637555842702,
      "loss": 4.6008,
      "step": 18876
    },
    {
      "epoch": 2.585182141878937,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000465522814607139,
      "loss": 4.6929,
      "step": 18877
    },
    {
      "epoch": 2.585319090660093,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004654818735013664,
      "loss": 4.6969,
      "step": 18878
    },
    {
      "epoch": 2.585456039441249,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00046544093226729197,
      "loss": 4.7649,
      "step": 18879
    },
    {
      "epoch": 2.585592988222405,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004653999909052548,
      "loss": 4.6083,
      "step": 18880
    },
    {
      "epoch": 2.5857299370035607,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004653590494155941,
      "loss": 4.7123,
      "step": 18881
    },
    {
      "epoch": 2.5858668857847165,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004653181077986495,
      "loss": 4.71,
      "step": 18882
    },
    {
      "epoch": 2.5860038345658722,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046527716605476,
      "loss": 4.6159,
      "step": 18883
    },
    {
      "epoch": 2.5861407833470285,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000465236224184265,
      "loss": 4.7342,
      "step": 18884
    },
    {
      "epoch": 2.586277732128184,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004651952821875038,
      "loss": 4.7133,
      "step": 18885
    },
    {
      "epoch": 2.58641468090934,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0004651543400648157,
      "loss": 4.7212,
      "step": 18886
    },
    {
      "epoch": 2.586551629690496,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00046511339781653987,
      "loss": 4.6848,
      "step": 18887
    },
    {
      "epoch": 2.5866885784716516,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004650724554430157,
      "loss": 4.6682,
      "step": 18888
    },
    {
      "epoch": 2.5868255272528073,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00046503151294458267,
      "loss": 4.6778,
      "step": 18889
    },
    {
      "epoch": 2.586962476033963,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004649905703215797,
      "loss": 4.7011,
      "step": 18890
    },
    {
      "epoch": 2.5870994248151193,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004649496275743465,
      "loss": 4.6569,
      "step": 18891
    },
    {
      "epoch": 2.587236373596275,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046490868470322206,
      "loss": 4.6055,
      "step": 18892
    },
    {
      "epoch": 2.587373322377431,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046486774170854587,
      "loss": 4.6743,
      "step": 18893
    },
    {
      "epoch": 2.5875102711585867,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0004648267985906571,
      "loss": 4.7384,
      "step": 18894
    },
    {
      "epoch": 2.5876472199397424,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046478585534989503,
      "loss": 4.687,
      "step": 18895
    },
    {
      "epoch": 2.5877841687208982,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004647449119865993,
      "loss": 4.7528,
      "step": 18896
    },
    {
      "epoch": 2.587921117502054,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004647039685011087,
      "loss": 4.7149,
      "step": 18897
    },
    {
      "epoch": 2.5880580662832102,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000464663024893763,
      "loss": 4.6735,
      "step": 18898
    },
    {
      "epoch": 2.588195015064366,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004646220811649013,
      "loss": 4.7133,
      "step": 18899
    },
    {
      "epoch": 2.5883319638455218,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00046458113731486287,
      "loss": 4.7196,
      "step": 18900
    },
    {
      "epoch": 2.5884689126266776,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00046454019334398703,
      "loss": 4.7098,
      "step": 18901
    },
    {
      "epoch": 2.5886058614078333,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004644992492526132,
      "loss": 4.6556,
      "step": 18902
    },
    {
      "epoch": 2.5887428101889896,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00046445830504108067,
      "loss": 4.7235,
      "step": 18903
    },
    {
      "epoch": 2.5888797589701453,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004644173607097286,
      "loss": 4.7265,
      "step": 18904
    },
    {
      "epoch": 2.589016707751301,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004643764162588966,
      "loss": 4.7534,
      "step": 18905
    },
    {
      "epoch": 2.589153656532457,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00046433547168892367,
      "loss": 4.6701,
      "step": 18906
    },
    {
      "epoch": 2.5892906053136127,
      "grad_norm": 0.359375,
      "learning_rate": 0.00046429452700014944,
      "loss": 4.73,
      "step": 18907
    },
    {
      "epoch": 2.5894275540947684,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00046425358219291294,
      "loss": 4.7178,
      "step": 18908
    },
    {
      "epoch": 2.589564502875924,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004642126372675535,
      "loss": 4.6509,
      "step": 18909
    },
    {
      "epoch": 2.5897014516570804,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004641716922244108,
      "loss": 4.6541,
      "step": 18910
    },
    {
      "epoch": 2.589838400438236,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004641307470638236,
      "loss": 4.6358,
      "step": 18911
    },
    {
      "epoch": 2.589975349219392,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046408980178613184,
      "loss": 4.629,
      "step": 18912
    },
    {
      "epoch": 2.5901122980005478,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046404885639167437,
      "loss": 4.7206,
      "step": 18913
    },
    {
      "epoch": 2.5902492467817035,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004640079108807907,
      "loss": 4.6125,
      "step": 18914
    },
    {
      "epoch": 2.5903861955628598,
      "grad_norm": 0.328125,
      "learning_rate": 0.00046396696525382013,
      "loss": 4.7368,
      "step": 18915
    },
    {
      "epoch": 2.590523144344015,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000463926019511102,
      "loss": 4.6494,
      "step": 18916
    },
    {
      "epoch": 2.5906600931251713,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046388507365297573,
      "loss": 4.6947,
      "step": 18917
    },
    {
      "epoch": 2.590797041906327,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004638441276797804,
      "loss": 4.6801,
      "step": 18918
    },
    {
      "epoch": 2.590933990687483,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00046380318159185553,
      "loss": 4.6742,
      "step": 18919
    },
    {
      "epoch": 2.5910709394686386,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004637622353895404,
      "loss": 4.6949,
      "step": 18920
    },
    {
      "epoch": 2.5912078882497944,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00046372128907317447,
      "loss": 4.6948,
      "step": 18921
    },
    {
      "epoch": 2.5913448370309506,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004636803426430968,
      "loss": 4.7023,
      "step": 18922
    },
    {
      "epoch": 2.5914817858121064,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004636393960996468,
      "loss": 4.6701,
      "step": 18923
    },
    {
      "epoch": 2.591618734593262,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000463598449443164,
      "loss": 4.6636,
      "step": 18924
    },
    {
      "epoch": 2.591755683374418,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046355750267398763,
      "loss": 4.7356,
      "step": 18925
    },
    {
      "epoch": 2.5918926321555738,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00046351655579245695,
      "loss": 4.6701,
      "step": 18926
    },
    {
      "epoch": 2.5920295809367295,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004634756087989113,
      "loss": 4.7399,
      "step": 18927
    },
    {
      "epoch": 2.5921665297178853,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004634346616936902,
      "loss": 4.6916,
      "step": 18928
    },
    {
      "epoch": 2.5923034784990415,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00046339371447713277,
      "loss": 4.6467,
      "step": 18929
    },
    {
      "epoch": 2.5924404272801973,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004633527671495784,
      "loss": 4.7312,
      "step": 18930
    },
    {
      "epoch": 2.592577376061353,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046331181971136665,
      "loss": 4.7089,
      "step": 18931
    },
    {
      "epoch": 2.592714324842509,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004632708721628364,
      "loss": 4.7843,
      "step": 18932
    },
    {
      "epoch": 2.5928512736236646,
      "grad_norm": 0.328125,
      "learning_rate": 0.00046322992450432754,
      "loss": 4.6575,
      "step": 18933
    },
    {
      "epoch": 2.592988222404821,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000463188976736179,
      "loss": 4.7383,
      "step": 18934
    },
    {
      "epoch": 2.593125171185976,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004631480288587304,
      "loss": 4.7295,
      "step": 18935
    },
    {
      "epoch": 2.5932621199671324,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00046310708087232076,
      "loss": 4.7031,
      "step": 18936
    },
    {
      "epoch": 2.593399068748288,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004630661327772897,
      "loss": 4.65,
      "step": 18937
    },
    {
      "epoch": 2.593536017529444,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004630251845739765,
      "loss": 4.774,
      "step": 18938
    },
    {
      "epoch": 2.5936729663105997,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004629842362627204,
      "loss": 4.7717,
      "step": 18939
    },
    {
      "epoch": 2.5938099150917555,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000462943287843861,
      "loss": 4.6986,
      "step": 18940
    },
    {
      "epoch": 2.5939468638729117,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004629023393177374,
      "loss": 4.6545,
      "step": 18941
    },
    {
      "epoch": 2.5940838126540675,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004628613906846891,
      "loss": 4.6271,
      "step": 18942
    },
    {
      "epoch": 2.5942207614352233,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004628204419450553,
      "loss": 4.6847,
      "step": 18943
    },
    {
      "epoch": 2.594357710216379,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00046277949309917545,
      "loss": 4.7564,
      "step": 18944
    },
    {
      "epoch": 2.594494658997535,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00046273854414738896,
      "loss": 4.6732,
      "step": 18945
    },
    {
      "epoch": 2.5946316077786906,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00046269759509003506,
      "loss": 4.6053,
      "step": 18946
    },
    {
      "epoch": 2.5947685565598464,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00046265664592745316,
      "loss": 4.7045,
      "step": 18947
    },
    {
      "epoch": 2.5949055053410026,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004626156966599828,
      "loss": 4.7227,
      "step": 18948
    },
    {
      "epoch": 2.5950424541221584,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00046257474728796295,
      "loss": 4.6714,
      "step": 18949
    },
    {
      "epoch": 2.595179402903314,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004625337978117331,
      "loss": 4.749,
      "step": 18950
    },
    {
      "epoch": 2.59531635168447,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046249284823163294,
      "loss": 4.72,
      "step": 18951
    },
    {
      "epoch": 2.5954533004656257,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004624518985480013,
      "loss": 4.6391,
      "step": 18952
    },
    {
      "epoch": 2.595590249246782,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004624109487611781,
      "loss": 4.673,
      "step": 18953
    },
    {
      "epoch": 2.5957271980279373,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004623699988715022,
      "loss": 4.6531,
      "step": 18954
    },
    {
      "epoch": 2.5958641468090935,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00046232904887931334,
      "loss": 4.7161,
      "step": 18955
    },
    {
      "epoch": 2.5960010955902493,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004622880987849505,
      "loss": 4.6979,
      "step": 18956
    },
    {
      "epoch": 2.596138044371405,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004622471485887534,
      "loss": 4.7322,
      "step": 18957
    },
    {
      "epoch": 2.596274993152561,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004622061982910612,
      "loss": 4.7313,
      "step": 18958
    },
    {
      "epoch": 2.5964119419337166,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004621652478922133,
      "loss": 4.6815,
      "step": 18959
    },
    {
      "epoch": 2.596548890714873,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004621242973925492,
      "loss": 4.7391,
      "step": 18960
    },
    {
      "epoch": 2.5966858394960286,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000462083346792408,
      "loss": 4.8276,
      "step": 18961
    },
    {
      "epoch": 2.5968227882771844,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046204239609212944,
      "loss": 4.6197,
      "step": 18962
    },
    {
      "epoch": 2.59695973705834,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004620014452920525,
      "loss": 4.7052,
      "step": 18963
    },
    {
      "epoch": 2.597096685839496,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004619604943925168,
      "loss": 4.6819,
      "step": 18964
    },
    {
      "epoch": 2.5972336346206517,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004619195433938616,
      "loss": 4.7179,
      "step": 18965
    },
    {
      "epoch": 2.5973705834018075,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046187859229642635,
      "loss": 4.6447,
      "step": 18966
    },
    {
      "epoch": 2.5975075321829637,
      "grad_norm": 0.328125,
      "learning_rate": 0.00046183764110055035,
      "loss": 4.6823,
      "step": 18967
    },
    {
      "epoch": 2.5976444809641195,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00046179668980657297,
      "loss": 4.721,
      "step": 18968
    },
    {
      "epoch": 2.5977814297452753,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004617557384148336,
      "loss": 4.696,
      "step": 18969
    },
    {
      "epoch": 2.597918378526431,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004617147869256717,
      "loss": 4.7643,
      "step": 18970
    },
    {
      "epoch": 2.598055327307587,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00046167383533942655,
      "loss": 4.7044,
      "step": 18971
    },
    {
      "epoch": 2.598192276088743,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004616328836564375,
      "loss": 4.6974,
      "step": 18972
    },
    {
      "epoch": 2.598329224869899,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00046159193187704403,
      "loss": 4.6574,
      "step": 18973
    },
    {
      "epoch": 2.5984661736510546,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004615509800015854,
      "loss": 4.7279,
      "step": 18974
    },
    {
      "epoch": 2.5986031224322104,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046151002803040107,
      "loss": 4.6855,
      "step": 18975
    },
    {
      "epoch": 2.598740071213366,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004614690759638304,
      "loss": 4.7082,
      "step": 18976
    },
    {
      "epoch": 2.598877019994522,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004614281238022127,
      "loss": 4.6658,
      "step": 18977
    },
    {
      "epoch": 2.5990139687756777,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004613871715458875,
      "loss": 4.6138,
      "step": 18978
    },
    {
      "epoch": 2.599150917556834,
      "grad_norm": 0.359375,
      "learning_rate": 0.00046134621919519407,
      "loss": 4.698,
      "step": 18979
    },
    {
      "epoch": 2.5992878663379897,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004613052667504718,
      "loss": 4.665,
      "step": 18980
    },
    {
      "epoch": 2.5994248151191455,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004612643142120601,
      "loss": 4.763,
      "step": 18981
    },
    {
      "epoch": 2.5995617639003012,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004612233615802984,
      "loss": 4.5971,
      "step": 18982
    },
    {
      "epoch": 2.599698712681457,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046118240885552596,
      "loss": 4.7113,
      "step": 18983
    },
    {
      "epoch": 2.5998356614626132,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046114145603808233,
      "loss": 4.7549,
      "step": 18984
    },
    {
      "epoch": 2.5999726102437686,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004611005031283066,
      "loss": 4.6327,
      "step": 18985
    },
    {
      "epoch": 2.600109559024925,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00046105955012653844,
      "loss": 4.6451,
      "step": 18986
    },
    {
      "epoch": 2.6002465078060806,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004610185970331173,
      "loss": 4.6931,
      "step": 18987
    },
    {
      "epoch": 2.6003834565872364,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00046097764384838227,
      "loss": 4.7027,
      "step": 18988
    },
    {
      "epoch": 2.600520405368392,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00046093669057267295,
      "loss": 4.6914,
      "step": 18989
    },
    {
      "epoch": 2.600657354149548,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004608957372063286,
      "loss": 4.7075,
      "step": 18990
    },
    {
      "epoch": 2.600794302930704,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004608547837496888,
      "loss": 4.6283,
      "step": 18991
    },
    {
      "epoch": 2.60093125171186,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046081383020309265,
      "loss": 4.7408,
      "step": 18992
    },
    {
      "epoch": 2.6010682004930157,
      "grad_norm": 0.328125,
      "learning_rate": 0.00046077287656687987,
      "loss": 4.7183,
      "step": 18993
    },
    {
      "epoch": 2.6012051492741715,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046073192284138965,
      "loss": 4.706,
      "step": 18994
    },
    {
      "epoch": 2.6013420980553272,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00046069096902696136,
      "loss": 4.7336,
      "step": 18995
    },
    {
      "epoch": 2.601479046836483,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004606500151239345,
      "loss": 4.7753,
      "step": 18996
    },
    {
      "epoch": 2.601615995617639,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004606090611326485,
      "loss": 4.7148,
      "step": 18997
    },
    {
      "epoch": 2.601752944398795,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00046056810705344257,
      "loss": 4.7911,
      "step": 18998
    },
    {
      "epoch": 2.601889893179951,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00046052715288665626,
      "loss": 4.7725,
      "step": 18999
    },
    {
      "epoch": 2.6020268419611066,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000460486198632629,
      "loss": 4.7116,
      "step": 19000
    },
    {
      "epoch": 2.6021637907422623,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004604452442917,
      "loss": 4.6314,
      "step": 19001
    },
    {
      "epoch": 2.602300739523418,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00046040428986420895,
      "loss": 4.7531,
      "step": 19002
    },
    {
      "epoch": 2.6024376883045743,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004603633353504949,
      "loss": 4.691,
      "step": 19003
    },
    {
      "epoch": 2.6025746370857297,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004603223807508975,
      "loss": 4.6922,
      "step": 19004
    },
    {
      "epoch": 2.602711585866886,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00046028142606575597,
      "loss": 4.6667,
      "step": 19005
    },
    {
      "epoch": 2.6028485346480417,
      "grad_norm": 0.375,
      "learning_rate": 0.0004602404712954098,
      "loss": 4.7099,
      "step": 19006
    },
    {
      "epoch": 2.6029854834291974,
      "grad_norm": 0.34375,
      "learning_rate": 0.00046019951644019855,
      "loss": 4.7407,
      "step": 19007
    },
    {
      "epoch": 2.603122432210353,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004601585615004614,
      "loss": 4.6402,
      "step": 19008
    },
    {
      "epoch": 2.603259380991509,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004601176064765379,
      "loss": 4.6709,
      "step": 19009
    },
    {
      "epoch": 2.603396329772665,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00046007665136876726,
      "loss": 4.7665,
      "step": 19010
    },
    {
      "epoch": 2.603533278553821,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00046003569617748916,
      "loss": 4.754,
      "step": 19011
    },
    {
      "epoch": 2.6036702273349768,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0004599947409030427,
      "loss": 4.6965,
      "step": 19012
    },
    {
      "epoch": 2.6038071761161325,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004599537855457676,
      "loss": 4.6627,
      "step": 19013
    },
    {
      "epoch": 2.6039441248972883,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0004599128301060029,
      "loss": 4.6442,
      "step": 19014
    },
    {
      "epoch": 2.604081073678444,
      "grad_norm": 0.359375,
      "learning_rate": 0.00045987187458408835,
      "loss": 4.6618,
      "step": 19015
    },
    {
      "epoch": 2.6042180224596,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00045983091898036326,
      "loss": 4.6732,
      "step": 19016
    },
    {
      "epoch": 2.604354971240756,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004597899632951669,
      "loss": 4.6616,
      "step": 19017
    },
    {
      "epoch": 2.604491920021912,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00045974900752883885,
      "loss": 4.8147,
      "step": 19018
    },
    {
      "epoch": 2.6046288688030677,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00045970805168171834,
      "loss": 4.7266,
      "step": 19019
    },
    {
      "epoch": 2.6047658175842234,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004596670957541451,
      "loss": 4.7509,
      "step": 19020
    },
    {
      "epoch": 2.604902766365379,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004596261397464581,
      "loss": 4.6947,
      "step": 19021
    },
    {
      "epoch": 2.6050397151465354,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00045958518365899723,
      "loss": 4.5724,
      "step": 19022
    },
    {
      "epoch": 2.605176663927691,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004595442274921014,
      "loss": 4.7285,
      "step": 19023
    },
    {
      "epoch": 2.605313612708847,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004595032712461105,
      "loss": 4.6492,
      "step": 19024
    },
    {
      "epoch": 2.6054505614900028,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004594623149213636,
      "loss": 4.6904,
      "step": 19025
    },
    {
      "epoch": 2.6055875102711585,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045942135851820027,
      "loss": 4.738,
      "step": 19026
    },
    {
      "epoch": 2.6057244590523143,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004593804020369599,
      "loss": 4.698,
      "step": 19027
    },
    {
      "epoch": 2.60586140783347,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00045933944547798194,
      "loss": 4.7723,
      "step": 19028
    },
    {
      "epoch": 2.6059983566146263,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004592984888416057,
      "loss": 4.7618,
      "step": 19029
    },
    {
      "epoch": 2.606135305395782,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004592575321281707,
      "loss": 4.687,
      "step": 19030
    },
    {
      "epoch": 2.606272254176938,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00045921657533801646,
      "loss": 4.6425,
      "step": 19031
    },
    {
      "epoch": 2.6064092029580936,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00045917561847148207,
      "loss": 4.6532,
      "step": 19032
    },
    {
      "epoch": 2.6065461517392494,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004591346615289071,
      "loss": 4.734,
      "step": 19033
    },
    {
      "epoch": 2.6066831005204056,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00045909370451063125,
      "loss": 4.7598,
      "step": 19034
    },
    {
      "epoch": 2.606820049301561,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004590527474169935,
      "loss": 4.8143,
      "step": 19035
    },
    {
      "epoch": 2.606956998082717,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004590117902483336,
      "loss": 4.7231,
      "step": 19036
    },
    {
      "epoch": 2.607093946863873,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004589708330049908,
      "loss": 4.6402,
      "step": 19037
    },
    {
      "epoch": 2.6072308956450287,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004589298756873046,
      "loss": 4.6703,
      "step": 19038
    },
    {
      "epoch": 2.6073678444261845,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004588889182956143,
      "loss": 4.7301,
      "step": 19039
    },
    {
      "epoch": 2.6075047932073403,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00045884796083025956,
      "loss": 4.7201,
      "step": 19040
    },
    {
      "epoch": 2.6076417419884965,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00045880700329157956,
      "loss": 4.6837,
      "step": 19041
    },
    {
      "epoch": 2.6077786907696523,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00045876604567991373,
      "loss": 4.742,
      "step": 19042
    },
    {
      "epoch": 2.607915639550808,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004587250879956018,
      "loss": 4.6773,
      "step": 19043
    },
    {
      "epoch": 2.608052588331964,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00045868413023898285,
      "loss": 4.7174,
      "step": 19044
    },
    {
      "epoch": 2.6081895371131196,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004586431724103965,
      "loss": 4.7105,
      "step": 19045
    },
    {
      "epoch": 2.6083264858942754,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000458602214510182,
      "loss": 4.6393,
      "step": 19046
    },
    {
      "epoch": 2.608463434675431,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004585612565386791,
      "loss": 4.6284,
      "step": 19047
    },
    {
      "epoch": 2.6086003834565874,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004585202984962269,
      "loss": 4.6532,
      "step": 19048
    },
    {
      "epoch": 2.608737332237743,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00045847934038316504,
      "loss": 4.6683,
      "step": 19049
    },
    {
      "epoch": 2.608874281018899,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004584383821998328,
      "loss": 4.6396,
      "step": 19050
    },
    {
      "epoch": 2.6090112298000547,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004583974239465696,
      "loss": 4.7329,
      "step": 19051
    },
    {
      "epoch": 2.6091481785812105,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004583564656237151,
      "loss": 4.6937,
      "step": 19052
    },
    {
      "epoch": 2.6092851273623667,
      "grad_norm": 0.359375,
      "learning_rate": 0.00045831550723160843,
      "loss": 4.6832,
      "step": 19053
    },
    {
      "epoch": 2.609422076143522,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004582745487705893,
      "loss": 4.6984,
      "step": 19054
    },
    {
      "epoch": 2.6095590249246783,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004582335902409969,
      "loss": 4.7159,
      "step": 19055
    },
    {
      "epoch": 2.609695973705834,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000458192631643171,
      "loss": 4.6485,
      "step": 19056
    },
    {
      "epoch": 2.60983292248699,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00045815167297745047,
      "loss": 4.7204,
      "step": 19057
    },
    {
      "epoch": 2.6099698712681456,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004581107142441754,
      "loss": 4.7153,
      "step": 19058
    },
    {
      "epoch": 2.6101068200493014,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004580697554436847,
      "loss": 4.7096,
      "step": 19059
    },
    {
      "epoch": 2.6102437688304576,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004580287965763181,
      "loss": 4.7416,
      "step": 19060
    },
    {
      "epoch": 2.6103807176116134,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000457987837642415,
      "loss": 4.7376,
      "step": 19061
    },
    {
      "epoch": 2.610517666392769,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004579468786423146,
      "loss": 4.6936,
      "step": 19062
    },
    {
      "epoch": 2.610654615173925,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00045790591957635667,
      "loss": 4.6973,
      "step": 19063
    },
    {
      "epoch": 2.6107915639550807,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00045786496044488047,
      "loss": 4.6665,
      "step": 19064
    },
    {
      "epoch": 2.6109285127362365,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004578240012482255,
      "loss": 4.7374,
      "step": 19065
    },
    {
      "epoch": 2.6110654615173923,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004577830419867311,
      "loss": 4.7195,
      "step": 19066
    },
    {
      "epoch": 2.6112024102985485,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004577420826607369,
      "loss": 4.7106,
      "step": 19067
    },
    {
      "epoch": 2.6113393590797043,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00045770112327058214,
      "loss": 4.5887,
      "step": 19068
    },
    {
      "epoch": 2.61147630786086,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004576601638166063,
      "loss": 4.6988,
      "step": 19069
    },
    {
      "epoch": 2.611613256642016,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004576192042991488,
      "loss": 4.7613,
      "step": 19070
    },
    {
      "epoch": 2.6117502054231716,
      "grad_norm": 0.375,
      "learning_rate": 0.0004575782447185493,
      "loss": 4.6519,
      "step": 19071
    },
    {
      "epoch": 2.611887154204328,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00045753728507514697,
      "loss": 4.6048,
      "step": 19072
    },
    {
      "epoch": 2.612024102985483,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004574963253692814,
      "loss": 4.6661,
      "step": 19073
    },
    {
      "epoch": 2.6121610517666394,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045745536560129203,
      "loss": 4.6858,
      "step": 19074
    },
    {
      "epoch": 2.612298000547795,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004574144057715182,
      "loss": 4.68,
      "step": 19075
    },
    {
      "epoch": 2.612434949328951,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004573734458802995,
      "loss": 4.7889,
      "step": 19076
    },
    {
      "epoch": 2.6125718981101067,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00045733248592797523,
      "loss": 4.6699,
      "step": 19077
    },
    {
      "epoch": 2.6127088468912625,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004572915259148849,
      "loss": 4.6625,
      "step": 19078
    },
    {
      "epoch": 2.6128457956724187,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000457250565841368,
      "loss": 4.648,
      "step": 19079
    },
    {
      "epoch": 2.6129827444535745,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00045720960570776387,
      "loss": 4.7544,
      "step": 19080
    },
    {
      "epoch": 2.6131196932347303,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004571686455144121,
      "loss": 4.6864,
      "step": 19081
    },
    {
      "epoch": 2.613256642015886,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00045712768526165204,
      "loss": 4.701,
      "step": 19082
    },
    {
      "epoch": 2.613393590797042,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0004570867249498231,
      "loss": 4.7099,
      "step": 19083
    },
    {
      "epoch": 2.613530539578198,
      "grad_norm": 0.34375,
      "learning_rate": 0.00045704576457926475,
      "loss": 4.6935,
      "step": 19084
    },
    {
      "epoch": 2.6136674883593534,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00045700480415031665,
      "loss": 4.6992,
      "step": 19085
    },
    {
      "epoch": 2.6138044371405096,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004569638436633179,
      "loss": 4.7782,
      "step": 19086
    },
    {
      "epoch": 2.6139413859216654,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004569228831186082,
      "loss": 4.6348,
      "step": 19087
    },
    {
      "epoch": 2.614078334702821,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004568819225165269,
      "loss": 4.7905,
      "step": 19088
    },
    {
      "epoch": 2.614215283483977,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00045684096185741335,
      "loss": 4.6465,
      "step": 19089
    },
    {
      "epoch": 2.6143522322651327,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004568000011416073,
      "loss": 4.589,
      "step": 19090
    },
    {
      "epoch": 2.614489181046289,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004567590403694479,
      "loss": 4.6077,
      "step": 19091
    },
    {
      "epoch": 2.6146261298274447,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004567180795412749,
      "loss": 4.6751,
      "step": 19092
    },
    {
      "epoch": 2.6147630786086005,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045667711865742737,
      "loss": 4.6286,
      "step": 19093
    },
    {
      "epoch": 2.6149000273897562,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00045663615771824505,
      "loss": 4.624,
      "step": 19094
    },
    {
      "epoch": 2.615036976170912,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004565951967240673,
      "loss": 4.7327,
      "step": 19095
    },
    {
      "epoch": 2.615173924952068,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004565542356752336,
      "loss": 4.6309,
      "step": 19096
    },
    {
      "epoch": 2.6153108737332236,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00045651327457208336,
      "loss": 4.7506,
      "step": 19097
    },
    {
      "epoch": 2.61544782251438,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000456472313414956,
      "loss": 4.6331,
      "step": 19098
    },
    {
      "epoch": 2.6155847712955356,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00045643135220419114,
      "loss": 4.6942,
      "step": 19099
    },
    {
      "epoch": 2.6157217200766913,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004563903909401281,
      "loss": 4.6531,
      "step": 19100
    },
    {
      "epoch": 2.615858668857847,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004563494296231064,
      "loss": 4.6906,
      "step": 19101
    },
    {
      "epoch": 2.615995617639003,
      "grad_norm": 0.328125,
      "learning_rate": 0.00045630846825346535,
      "loss": 4.7758,
      "step": 19102
    },
    {
      "epoch": 2.616132566420159,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004562675068315447,
      "loss": 4.6705,
      "step": 19103
    },
    {
      "epoch": 2.6162695152013145,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004562265453576836,
      "loss": 4.7646,
      "step": 19104
    },
    {
      "epoch": 2.6164064639824707,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00045618558383222173,
      "loss": 4.6426,
      "step": 19105
    },
    {
      "epoch": 2.6165434127636265,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004561446222554983,
      "loss": 4.7562,
      "step": 19106
    },
    {
      "epoch": 2.6166803615447822,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000456103660627853,
      "loss": 4.7302,
      "step": 19107
    },
    {
      "epoch": 2.616817310325938,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0004560626989496252,
      "loss": 4.7122,
      "step": 19108
    },
    {
      "epoch": 2.616954259107094,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004560217372211544,
      "loss": 4.6991,
      "step": 19109
    },
    {
      "epoch": 2.61709120788825,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045598077544278006,
      "loss": 4.6988,
      "step": 19110
    },
    {
      "epoch": 2.617228156669406,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004559398136148415,
      "loss": 4.7696,
      "step": 19111
    },
    {
      "epoch": 2.6173651054505616,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004558988517376784,
      "loss": 4.6594,
      "step": 19112
    },
    {
      "epoch": 2.6175020542317173,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004558578898116301,
      "loss": 4.6149,
      "step": 19113
    },
    {
      "epoch": 2.617639003012873,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00045581692783703604,
      "loss": 4.6933,
      "step": 19114
    },
    {
      "epoch": 2.617775951794029,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004557759658142356,
      "loss": 4.6718,
      "step": 19115
    },
    {
      "epoch": 2.6179129005751847,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00045573500374356847,
      "loss": 4.7669,
      "step": 19116
    },
    {
      "epoch": 2.618049849356341,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004556940416253741,
      "loss": 4.6674,
      "step": 19117
    },
    {
      "epoch": 2.6181867981374967,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00045565307945999163,
      "loss": 4.7013,
      "step": 19118
    },
    {
      "epoch": 2.6183237469186524,
      "grad_norm": 0.34375,
      "learning_rate": 0.00045561211724776086,
      "loss": 4.7138,
      "step": 19119
    },
    {
      "epoch": 2.618460695699808,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004555711549890211,
      "loss": 4.6746,
      "step": 19120
    },
    {
      "epoch": 2.618597644480964,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000455530192684112,
      "loss": 4.705,
      "step": 19121
    },
    {
      "epoch": 2.61873459326212,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00045548923033337277,
      "loss": 4.7036,
      "step": 19122
    },
    {
      "epoch": 2.6188715420432755,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00045544826793714286,
      "loss": 4.634,
      "step": 19123
    },
    {
      "epoch": 2.6190084908244318,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000455407305495762,
      "loss": 4.7328,
      "step": 19124
    },
    {
      "epoch": 2.6191454396055875,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00045536634300956944,
      "loss": 4.7597,
      "step": 19125
    },
    {
      "epoch": 2.6192823883867433,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004553253804789048,
      "loss": 4.7638,
      "step": 19126
    },
    {
      "epoch": 2.619419337167899,
      "grad_norm": 0.328125,
      "learning_rate": 0.00045528441790410734,
      "loss": 4.6631,
      "step": 19127
    },
    {
      "epoch": 2.619556285949055,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045524345528551684,
      "loss": 4.7118,
      "step": 19128
    },
    {
      "epoch": 2.619693234730211,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00045520249262347243,
      "loss": 4.667,
      "step": 19129
    },
    {
      "epoch": 2.619830183511367,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00045516152991831364,
      "loss": 4.7063,
      "step": 19130
    },
    {
      "epoch": 2.6199671322925226,
      "grad_norm": 0.375,
      "learning_rate": 0.0004551205671703802,
      "loss": 4.742,
      "step": 19131
    },
    {
      "epoch": 2.6201040810736784,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004550796043800113,
      "loss": 4.6884,
      "step": 19132
    },
    {
      "epoch": 2.620241029854834,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004550386415475466,
      "loss": 4.6095,
      "step": 19133
    },
    {
      "epoch": 2.62037797863599,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004549976786733254,
      "loss": 4.7728,
      "step": 19134
    },
    {
      "epoch": 2.6205149274171458,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004549567157576873,
      "loss": 4.6899,
      "step": 19135
    },
    {
      "epoch": 2.620651876198302,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004549157528009717,
      "loss": 4.7333,
      "step": 19136
    },
    {
      "epoch": 2.6207888249794578,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004548747898035179,
      "loss": 4.7719,
      "step": 19137
    },
    {
      "epoch": 2.6209257737606135,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00045483382676566583,
      "loss": 4.6474,
      "step": 19138
    },
    {
      "epoch": 2.6210627225417693,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00045479286368775454,
      "loss": 4.8025,
      "step": 19139
    },
    {
      "epoch": 2.621199671322925,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004547519005701237,
      "loss": 4.7104,
      "step": 19140
    },
    {
      "epoch": 2.6213366201040813,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004547109374131127,
      "loss": 4.7579,
      "step": 19141
    },
    {
      "epoch": 2.621473568885237,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00045466997421706113,
      "loss": 4.6885,
      "step": 19142
    },
    {
      "epoch": 2.621610517666393,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004546290109823082,
      "loss": 4.6603,
      "step": 19143
    },
    {
      "epoch": 2.6217474664475486,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00045458804770919366,
      "loss": 4.6605,
      "step": 19144
    },
    {
      "epoch": 2.6218844152287044,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004545470843980568,
      "loss": 4.6608,
      "step": 19145
    },
    {
      "epoch": 2.62202136400986,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004545061210492373,
      "loss": 4.711,
      "step": 19146
    },
    {
      "epoch": 2.622158312791016,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00045446515766307447,
      "loss": 4.7256,
      "step": 19147
    },
    {
      "epoch": 2.622295261572172,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004544241942399077,
      "loss": 4.6796,
      "step": 19148
    },
    {
      "epoch": 2.622432210353328,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004543832307800767,
      "loss": 4.5968,
      "step": 19149
    },
    {
      "epoch": 2.6225691591344837,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045434226728392074,
      "loss": 4.6613,
      "step": 19150
    },
    {
      "epoch": 2.6227061079156395,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004543013037517793,
      "loss": 4.6396,
      "step": 19151
    },
    {
      "epoch": 2.6228430566967953,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004542603401839922,
      "loss": 4.6993,
      "step": 19152
    },
    {
      "epoch": 2.6229800054779515,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004542193765808984,
      "loss": 4.7265,
      "step": 19153
    },
    {
      "epoch": 2.623116954259107,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004541784129428378,
      "loss": 4.603,
      "step": 19154
    },
    {
      "epoch": 2.623253903040263,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004541374492701495,
      "loss": 4.7231,
      "step": 19155
    },
    {
      "epoch": 2.623390851821419,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004540964855631732,
      "loss": 4.6307,
      "step": 19156
    },
    {
      "epoch": 2.6235278006025746,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00045405552182224847,
      "loss": 4.6799,
      "step": 19157
    },
    {
      "epoch": 2.6236647493837304,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004540145580477146,
      "loss": 4.6249,
      "step": 19158
    },
    {
      "epoch": 2.623801698164886,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004539735942399112,
      "loss": 4.7462,
      "step": 19159
    },
    {
      "epoch": 2.6239386469460424,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004539326303991776,
      "loss": 4.7053,
      "step": 19160
    },
    {
      "epoch": 2.624075595727198,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004538916665258535,
      "loss": 4.6943,
      "step": 19161
    },
    {
      "epoch": 2.624212544508354,
      "grad_norm": 0.359375,
      "learning_rate": 0.000453850702620278,
      "loss": 4.6493,
      "step": 19162
    },
    {
      "epoch": 2.6243494932895097,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004538097386827909,
      "loss": 4.6924,
      "step": 19163
    },
    {
      "epoch": 2.6244864420706655,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00045376877471373165,
      "loss": 4.7294,
      "step": 19164
    },
    {
      "epoch": 2.6246233908518213,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004537278107134396,
      "loss": 4.728,
      "step": 19165
    },
    {
      "epoch": 2.624760339632977,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004536868466822544,
      "loss": 4.7019,
      "step": 19166
    },
    {
      "epoch": 2.6248972884141333,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004536458826205153,
      "loss": 4.6853,
      "step": 19167
    },
    {
      "epoch": 2.625034237195289,
      "grad_norm": 0.34375,
      "learning_rate": 0.00045360491852856205,
      "loss": 4.664,
      "step": 19168
    },
    {
      "epoch": 2.625171185976445,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004535639544067338,
      "loss": 4.6983,
      "step": 19169
    },
    {
      "epoch": 2.6253081347576006,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00045352299025537026,
      "loss": 4.6851,
      "step": 19170
    },
    {
      "epoch": 2.6254450835387564,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045348202607481094,
      "loss": 4.7007,
      "step": 19171
    },
    {
      "epoch": 2.6255820323199126,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00045344106186539514,
      "loss": 4.6665,
      "step": 19172
    },
    {
      "epoch": 2.625718981101068,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004534000976274625,
      "loss": 4.71,
      "step": 19173
    },
    {
      "epoch": 2.625855929882224,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045335913336135236,
      "loss": 4.6212,
      "step": 19174
    },
    {
      "epoch": 2.62599287866338,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004533181690674044,
      "loss": 4.7299,
      "step": 19175
    },
    {
      "epoch": 2.6261298274445357,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045327720474595796,
      "loss": 4.6769,
      "step": 19176
    },
    {
      "epoch": 2.6262667762256915,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004532362403973525,
      "loss": 4.624,
      "step": 19177
    },
    {
      "epoch": 2.6264037250068473,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0004531952760219275,
      "loss": 4.639,
      "step": 19178
    },
    {
      "epoch": 2.6265406737880035,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00045315431162002264,
      "loss": 4.7275,
      "step": 19179
    },
    {
      "epoch": 2.6266776225691593,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00045311334719197715,
      "loss": 4.6754,
      "step": 19180
    },
    {
      "epoch": 2.626814571350315,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00045307238273813063,
      "loss": 4.5988,
      "step": 19181
    },
    {
      "epoch": 2.626951520131471,
      "grad_norm": 0.359375,
      "learning_rate": 0.00045303141825882255,
      "loss": 4.6649,
      "step": 19182
    },
    {
      "epoch": 2.6270884689126266,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045299045375439225,
      "loss": 4.7599,
      "step": 19183
    },
    {
      "epoch": 2.6272254176937824,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004529494892251795,
      "loss": 4.6673,
      "step": 19184
    },
    {
      "epoch": 2.627362366474938,
      "grad_norm": 0.390625,
      "learning_rate": 0.0004529085246715236,
      "loss": 4.5981,
      "step": 19185
    },
    {
      "epoch": 2.6274993152560944,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00045286756009376404,
      "loss": 4.7319,
      "step": 19186
    },
    {
      "epoch": 2.62763626403725,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045282659549224035,
      "loss": 4.6487,
      "step": 19187
    },
    {
      "epoch": 2.627773212818406,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00045278563086729204,
      "loss": 4.7191,
      "step": 19188
    },
    {
      "epoch": 2.6279101615995617,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00045274466621925846,
      "loss": 4.6684,
      "step": 19189
    },
    {
      "epoch": 2.6280471103807175,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045270370154847924,
      "loss": 4.7323,
      "step": 19190
    },
    {
      "epoch": 2.6281840591618737,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00045266273685529374,
      "loss": 4.6037,
      "step": 19191
    },
    {
      "epoch": 2.628321007943029,
      "grad_norm": 0.375,
      "learning_rate": 0.00045262177214004136,
      "loss": 4.7083,
      "step": 19192
    },
    {
      "epoch": 2.6284579567241853,
      "grad_norm": 0.359375,
      "learning_rate": 0.000452580807403062,
      "loss": 4.641,
      "step": 19193
    },
    {
      "epoch": 2.628594905505341,
      "grad_norm": 0.34375,
      "learning_rate": 0.00045253984264469476,
      "loss": 4.7411,
      "step": 19194
    },
    {
      "epoch": 2.628731854286497,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00045249887786527924,
      "loss": 4.6625,
      "step": 19195
    },
    {
      "epoch": 2.6288688030676526,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045245791306515486,
      "loss": 4.7425,
      "step": 19196
    },
    {
      "epoch": 2.6290057518488084,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004524169482446613,
      "loss": 4.7376,
      "step": 19197
    },
    {
      "epoch": 2.6291427006299646,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004523759834041379,
      "loss": 4.7884,
      "step": 19198
    },
    {
      "epoch": 2.6292796494111204,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004523350185439241,
      "loss": 4.734,
      "step": 19199
    },
    {
      "epoch": 2.629416598192276,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00045229405366435947,
      "loss": 4.6135,
      "step": 19200
    },
    {
      "epoch": 2.629553546973432,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00045225308876578336,
      "loss": 4.7615,
      "step": 19201
    },
    {
      "epoch": 2.6296904957545877,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004522121238485355,
      "loss": 4.6081,
      "step": 19202
    },
    {
      "epoch": 2.629827444535744,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00045217115891295524,
      "loss": 4.7068,
      "step": 19203
    },
    {
      "epoch": 2.6299643933168992,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000452130193959382,
      "loss": 4.6902,
      "step": 19204
    },
    {
      "epoch": 2.6301013420980555,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004520892289881554,
      "loss": 4.7411,
      "step": 19205
    },
    {
      "epoch": 2.6302382908792112,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004520482639996149,
      "loss": 4.7014,
      "step": 19206
    },
    {
      "epoch": 2.630375239660367,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004520072989940998,
      "loss": 4.6917,
      "step": 19207
    },
    {
      "epoch": 2.630512188441523,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00045196633397194986,
      "loss": 4.6659,
      "step": 19208
    },
    {
      "epoch": 2.6306491372226786,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00045192536893350435,
      "loss": 4.713,
      "step": 19209
    },
    {
      "epoch": 2.630786086003835,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004518844038791029,
      "loss": 4.6381,
      "step": 19210
    },
    {
      "epoch": 2.6309230347849906,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00045184343880908494,
      "loss": 4.7976,
      "step": 19211
    },
    {
      "epoch": 2.6310599835661463,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004518024737237899,
      "loss": 4.7627,
      "step": 19212
    },
    {
      "epoch": 2.631196932347302,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0004517615086235575,
      "loss": 4.5611,
      "step": 19213
    },
    {
      "epoch": 2.631333881128458,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00045172054350872687,
      "loss": 4.7106,
      "step": 19214
    },
    {
      "epoch": 2.6314708299096137,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004516795783796379,
      "loss": 4.6739,
      "step": 19215
    },
    {
      "epoch": 2.6316077786907695,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004516386132366296,
      "loss": 4.6933,
      "step": 19216
    },
    {
      "epoch": 2.6317447274719257,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000451597648080042,
      "loss": 4.6405,
      "step": 19217
    },
    {
      "epoch": 2.6318816762530814,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004515566829102141,
      "loss": 4.849,
      "step": 19218
    },
    {
      "epoch": 2.6320186250342372,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00045151571772748556,
      "loss": 4.6823,
      "step": 19219
    },
    {
      "epoch": 2.632155573815393,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004514747525321961,
      "loss": 4.7346,
      "step": 19220
    },
    {
      "epoch": 2.632292522596549,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00045143378732468494,
      "loss": 4.5588,
      "step": 19221
    },
    {
      "epoch": 2.632429471377705,
      "grad_norm": 0.3125,
      "learning_rate": 0.00045139282210529167,
      "loss": 4.7572,
      "step": 19222
    },
    {
      "epoch": 2.6325664201588603,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004513518568743556,
      "loss": 4.739,
      "step": 19223
    },
    {
      "epoch": 2.6327033689400166,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00045131089163221653,
      "loss": 4.7329,
      "step": 19224
    },
    {
      "epoch": 2.6328403177211723,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00045126992637921376,
      "loss": 4.6726,
      "step": 19225
    },
    {
      "epoch": 2.632977266502328,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004512289611156868,
      "loss": 4.6791,
      "step": 19226
    },
    {
      "epoch": 2.633114215283484,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00045118799584197505,
      "loss": 4.6889,
      "step": 19227
    },
    {
      "epoch": 2.6332511640646397,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004511470305584183,
      "loss": 4.6907,
      "step": 19228
    },
    {
      "epoch": 2.633388112845796,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004511060652653557,
      "loss": 4.634,
      "step": 19229
    },
    {
      "epoch": 2.6335250616269517,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00045106509996312676,
      "loss": 4.6211,
      "step": 19230
    },
    {
      "epoch": 2.6336620104081074,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004510241346520713,
      "loss": 4.721,
      "step": 19231
    },
    {
      "epoch": 2.633798959189263,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00045098316933252843,
      "loss": 4.713,
      "step": 19232
    },
    {
      "epoch": 2.633935907970419,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004509422040048379,
      "loss": 4.7398,
      "step": 19233
    },
    {
      "epoch": 2.6340728567515748,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004509012386693391,
      "loss": 4.711,
      "step": 19234
    },
    {
      "epoch": 2.6342098055327305,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00045086027332637144,
      "loss": 4.7936,
      "step": 19235
    },
    {
      "epoch": 2.6343467543138868,
      "grad_norm": 0.328125,
      "learning_rate": 0.00045081930797627456,
      "loss": 4.7546,
      "step": 19236
    },
    {
      "epoch": 2.6344837030950425,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004507783426193879,
      "loss": 4.6037,
      "step": 19237
    },
    {
      "epoch": 2.6346206518761983,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004507373772560508,
      "loss": 4.6765,
      "step": 19238
    },
    {
      "epoch": 2.634757600657354,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000450696411886603,
      "loss": 4.6699,
      "step": 19239
    },
    {
      "epoch": 2.63489454943851,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00045065544651138386,
      "loss": 4.6505,
      "step": 19240
    },
    {
      "epoch": 2.635031498219666,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00045061448113073275,
      "loss": 4.6505,
      "step": 19241
    },
    {
      "epoch": 2.6351684470008214,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004505735157449896,
      "loss": 4.6629,
      "step": 19242
    },
    {
      "epoch": 2.6353053957819776,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004505325503544933,
      "loss": 4.557,
      "step": 19243
    },
    {
      "epoch": 2.6354423445631334,
      "grad_norm": 0.34375,
      "learning_rate": 0.00045049158495958367,
      "loss": 4.772,
      "step": 19244
    },
    {
      "epoch": 2.635579293344289,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00045045061956060017,
      "loss": 4.725,
      "step": 19245
    },
    {
      "epoch": 2.635716242125445,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004504096541578824,
      "loss": 4.6674,
      "step": 19246
    },
    {
      "epoch": 2.6358531909066008,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00045036868875176963,
      "loss": 4.6277,
      "step": 19247
    },
    {
      "epoch": 2.635990139687757,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00045032772334260135,
      "loss": 4.7692,
      "step": 19248
    },
    {
      "epoch": 2.6361270884689127,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00045028675793071737,
      "loss": 4.6039,
      "step": 19249
    },
    {
      "epoch": 2.6362640372500685,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00045024579251645675,
      "loss": 4.6785,
      "step": 19250
    },
    {
      "epoch": 2.6364009860312243,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004502048271001594,
      "loss": 4.686,
      "step": 19251
    },
    {
      "epoch": 2.63653793481238,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004501638616821644,
      "loss": 4.7119,
      "step": 19252
    },
    {
      "epoch": 2.636674883593536,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00045012289626281154,
      "loss": 4.6556,
      "step": 19253
    },
    {
      "epoch": 2.6368118323746916,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00045008193084244014,
      "loss": 4.7124,
      "step": 19254
    },
    {
      "epoch": 2.636948781155848,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004500409654213899,
      "loss": 4.6073,
      "step": 19255
    },
    {
      "epoch": 2.6370857299370036,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00045,
      "loss": 4.723,
      "step": 19256
    },
    {
      "epoch": 2.6372226787181594,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00044995903457861026,
      "loss": 4.7205,
      "step": 19257
    },
    {
      "epoch": 2.637359627499315,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004499180691575598,
      "loss": 4.629,
      "step": 19258
    },
    {
      "epoch": 2.637496576280471,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004498771037371885,
      "loss": 4.726,
      "step": 19259
    },
    {
      "epoch": 2.637633525061627,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00044983613831783565,
      "loss": 4.655,
      "step": 19260
    },
    {
      "epoch": 2.637770473842783,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004497951728998408,
      "loss": 4.684,
      "step": 19261
    },
    {
      "epoch": 2.6379074226239387,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00044975420748354317,
      "loss": 4.6413,
      "step": 19262
    },
    {
      "epoch": 2.6380443714050945,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004497132420692827,
      "loss": 4.6942,
      "step": 19263
    },
    {
      "epoch": 2.6381813201862503,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004496722766573987,
      "loss": 4.6025,
      "step": 19264
    },
    {
      "epoch": 2.638318268967406,
      "grad_norm": 0.359375,
      "learning_rate": 0.00044963131124823056,
      "loss": 4.7178,
      "step": 19265
    },
    {
      "epoch": 2.638455217748562,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004495903458421176,
      "loss": 4.716,
      "step": 19266
    },
    {
      "epoch": 2.638592166529718,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004495493804393998,
      "loss": 4.6798,
      "step": 19267
    },
    {
      "epoch": 2.638729115310874,
      "grad_norm": 0.390625,
      "learning_rate": 0.0004495084150404164,
      "loss": 4.6643,
      "step": 19268
    },
    {
      "epoch": 2.6388660640920296,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00044946744964550664,
      "loss": 4.6594,
      "step": 19269
    },
    {
      "epoch": 2.6390030128731854,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004494264842550105,
      "loss": 4.6289,
      "step": 19270
    },
    {
      "epoch": 2.639139961654341,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004493855188692672,
      "loss": 4.751,
      "step": 19271
    },
    {
      "epoch": 2.6392769104354974,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004493445534886163,
      "loss": 4.66,
      "step": 19272
    },
    {
      "epoch": 2.6394138592166527,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000449303588113397,
      "loss": 4.7105,
      "step": 19273
    },
    {
      "epoch": 2.639550807997809,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004492626227439492,
      "loss": 4.7127,
      "step": 19274
    },
    {
      "epoch": 2.6396877567789647,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004492216573806122,
      "loss": 4.6806,
      "step": 19275
    },
    {
      "epoch": 2.6398247055601205,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004491806920237254,
      "loss": 4.7192,
      "step": 19276
    },
    {
      "epoch": 2.6399616543412763,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004491397266736286,
      "loss": 4.6189,
      "step": 19277
    },
    {
      "epoch": 2.640098603122432,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000449098761330661,
      "loss": 4.646,
      "step": 19278
    },
    {
      "epoch": 2.6402355519035883,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044905779599516225,
      "loss": 4.6941,
      "step": 19279
    },
    {
      "epoch": 2.640372500684744,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004490168306674715,
      "loss": 4.6402,
      "step": 19280
    },
    {
      "epoch": 2.6405094494659,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004489758653479288,
      "loss": 4.6248,
      "step": 19281
    },
    {
      "epoch": 2.6406463982470556,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00044893490003687327,
      "loss": 4.6215,
      "step": 19282
    },
    {
      "epoch": 2.6407833470282114,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004488939347346445,
      "loss": 4.6968,
      "step": 19283
    },
    {
      "epoch": 2.640920295809367,
      "grad_norm": 0.34375,
      "learning_rate": 0.00044885296944158177,
      "loss": 4.6451,
      "step": 19284
    },
    {
      "epoch": 2.641057244590523,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000448812004158025,
      "loss": 4.7322,
      "step": 19285
    },
    {
      "epoch": 2.641194193371679,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00044877103888431335,
      "loss": 4.6855,
      "step": 19286
    },
    {
      "epoch": 2.641331142152835,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004487300736207862,
      "loss": 4.6976,
      "step": 19287
    },
    {
      "epoch": 2.6414680909339907,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004486891083677835,
      "loss": 4.6315,
      "step": 19288
    },
    {
      "epoch": 2.6416050397151465,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004486481431256444,
      "loss": 4.7747,
      "step": 19289
    },
    {
      "epoch": 2.6417419884963023,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004486071778947085,
      "loss": 4.6868,
      "step": 19290
    },
    {
      "epoch": 2.6418789372774585,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00044856621267531504,
      "loss": 4.5673,
      "step": 19291
    },
    {
      "epoch": 2.642015886058614,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004485252474678039,
      "loss": 4.6333,
      "step": 19292
    },
    {
      "epoch": 2.64215283483977,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004484842822725144,
      "loss": 4.702,
      "step": 19293
    },
    {
      "epoch": 2.642289783620926,
      "grad_norm": 0.40625,
      "learning_rate": 0.0004484433170897859,
      "loss": 4.6382,
      "step": 19294
    },
    {
      "epoch": 2.6424267324020816,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004484023519199581,
      "loss": 4.7359,
      "step": 19295
    },
    {
      "epoch": 2.6425636811832374,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004483613867633704,
      "loss": 4.713,
      "step": 19296
    },
    {
      "epoch": 2.642700629964393,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00044832042162036226,
      "loss": 4.6323,
      "step": 19297
    },
    {
      "epoch": 2.6428375787455494,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044827945649127305,
      "loss": 4.7336,
      "step": 19298
    },
    {
      "epoch": 2.642974527526705,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0004482384913764425,
      "loss": 4.6104,
      "step": 19299
    },
    {
      "epoch": 2.643111476307861,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00044819752627621006,
      "loss": 4.6571,
      "step": 19300
    },
    {
      "epoch": 2.6432484250890167,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004481565611909152,
      "loss": 4.7303,
      "step": 19301
    },
    {
      "epoch": 2.6433853738701725,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044811559612089707,
      "loss": 4.668,
      "step": 19302
    },
    {
      "epoch": 2.6435223226513282,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004480746310664957,
      "loss": 4.7192,
      "step": 19303
    },
    {
      "epoch": 2.643659271432484,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00044803366602805023,
      "loss": 4.6585,
      "step": 19304
    },
    {
      "epoch": 2.6437962202136402,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00044799270100590017,
      "loss": 4.6786,
      "step": 19305
    },
    {
      "epoch": 2.643933168994796,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00044795173600038513,
      "loss": 4.743,
      "step": 19306
    },
    {
      "epoch": 2.644070117775952,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00044791077101184463,
      "loss": 4.6764,
      "step": 19307
    },
    {
      "epoch": 2.6442070665571076,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004478698060406181,
      "loss": 4.7425,
      "step": 19308
    },
    {
      "epoch": 2.6443440153382634,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044782884108704473,
      "loss": 4.6858,
      "step": 19309
    },
    {
      "epoch": 2.6444809641194196,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00044778787615146453,
      "loss": 4.7966,
      "step": 19310
    },
    {
      "epoch": 2.644617912900575,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044774691123421667,
      "loss": 4.5816,
      "step": 19311
    },
    {
      "epoch": 2.644754861681731,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004477059463356405,
      "loss": 4.7174,
      "step": 19312
    },
    {
      "epoch": 2.644891810462887,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004476649814560759,
      "loss": 4.6955,
      "step": 19313
    },
    {
      "epoch": 2.6450287592440427,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004476240165958622,
      "loss": 4.7514,
      "step": 19314
    },
    {
      "epoch": 2.6451657080251985,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004475830517553388,
      "loss": 4.7166,
      "step": 19315
    },
    {
      "epoch": 2.6453026568063542,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000447542086934845,
      "loss": 4.6959,
      "step": 19316
    },
    {
      "epoch": 2.6454396055875105,
      "grad_norm": 0.34375,
      "learning_rate": 0.00044750112213472074,
      "loss": 4.7435,
      "step": 19317
    },
    {
      "epoch": 2.6455765543686662,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00044746015735530533,
      "loss": 4.7207,
      "step": 19318
    },
    {
      "epoch": 2.645713503149822,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00044741919259693813,
      "loss": 4.6727,
      "step": 19319
    },
    {
      "epoch": 2.645850451930978,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004473782278599585,
      "loss": 4.7522,
      "step": 19320
    },
    {
      "epoch": 2.6459874007121336,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00044733726314470635,
      "loss": 4.7587,
      "step": 19321
    },
    {
      "epoch": 2.64612434949329,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004472962984515209,
      "loss": 4.726,
      "step": 19322
    },
    {
      "epoch": 2.646261298274445,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004472553337807415,
      "loss": 4.7151,
      "step": 19323
    },
    {
      "epoch": 2.6463982470556013,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00044721436913270805,
      "loss": 4.6218,
      "step": 19324
    },
    {
      "epoch": 2.646535195836757,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004471734045077597,
      "loss": 4.6756,
      "step": 19325
    },
    {
      "epoch": 2.646672144617913,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044713243990623605,
      "loss": 4.5993,
      "step": 19326
    },
    {
      "epoch": 2.6468090933990687,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00044709147532847633,
      "loss": 4.6866,
      "step": 19327
    },
    {
      "epoch": 2.6469460421802244,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004470505107748205,
      "loss": 4.6782,
      "step": 19328
    },
    {
      "epoch": 2.6470829909613807,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004470095462456078,
      "loss": 4.6836,
      "step": 19329
    },
    {
      "epoch": 2.6472199397425364,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00044696858174117754,
      "loss": 4.7662,
      "step": 19330
    },
    {
      "epoch": 2.647356888523692,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004469276172618694,
      "loss": 4.6932,
      "step": 19331
    },
    {
      "epoch": 2.647493837304848,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044688665280802293,
      "loss": 4.7028,
      "step": 19332
    },
    {
      "epoch": 2.6476307860860038,
      "grad_norm": 0.359375,
      "learning_rate": 0.00044684568837997745,
      "loss": 4.6619,
      "step": 19333
    },
    {
      "epoch": 2.6477677348671596,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004468047239780724,
      "loss": 4.7375,
      "step": 19334
    },
    {
      "epoch": 2.6479046836483153,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00044676375960264747,
      "loss": 4.7158,
      "step": 19335
    },
    {
      "epoch": 2.6480416324294715,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004467227952540421,
      "loss": 4.7022,
      "step": 19336
    },
    {
      "epoch": 2.6481785812106273,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004466818309325956,
      "loss": 4.7582,
      "step": 19337
    },
    {
      "epoch": 2.648315529991783,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00044664086663864756,
      "loss": 4.6981,
      "step": 19338
    },
    {
      "epoch": 2.648452478772939,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00044659990237253753,
      "loss": 4.637,
      "step": 19339
    },
    {
      "epoch": 2.6485894275540947,
      "grad_norm": 0.359375,
      "learning_rate": 0.00044655893813460494,
      "loss": 4.6862,
      "step": 19340
    },
    {
      "epoch": 2.648726376335251,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00044651797392518904,
      "loss": 4.7772,
      "step": 19341
    },
    {
      "epoch": 2.648863325116406,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004464770097446297,
      "loss": 4.6407,
      "step": 19342
    },
    {
      "epoch": 2.6490002738975624,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004464360455932663,
      "loss": 4.6474,
      "step": 19343
    },
    {
      "epoch": 2.649137222678718,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044639508147143803,
      "loss": 4.7022,
      "step": 19344
    },
    {
      "epoch": 2.649274171459874,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044635411737948465,
      "loss": 4.6833,
      "step": 19345
    },
    {
      "epoch": 2.6494111202410298,
      "grad_norm": 0.359375,
      "learning_rate": 0.00044631315331774565,
      "loss": 4.7723,
      "step": 19346
    },
    {
      "epoch": 2.6495480690221855,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044627218928656044,
      "loss": 4.6727,
      "step": 19347
    },
    {
      "epoch": 2.6496850178033418,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004462312252862684,
      "loss": 4.6989,
      "step": 19348
    },
    {
      "epoch": 2.6498219665844975,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00044619026131720904,
      "loss": 4.6368,
      "step": 19349
    },
    {
      "epoch": 2.6499589153656533,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000446149297379722,
      "loss": 4.6621,
      "step": 19350
    },
    {
      "epoch": 2.650095864146809,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004461083334741466,
      "loss": 4.7827,
      "step": 19351
    },
    {
      "epoch": 2.650232812927965,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004460673696008223,
      "loss": 4.6734,
      "step": 19352
    },
    {
      "epoch": 2.6503697617091206,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004460264057600888,
      "loss": 4.6578,
      "step": 19353
    },
    {
      "epoch": 2.6505067104902764,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004459854419522854,
      "loss": 4.7257,
      "step": 19354
    },
    {
      "epoch": 2.6506436592714326,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004459444781777515,
      "loss": 4.6723,
      "step": 19355
    },
    {
      "epoch": 2.6507806080525884,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004459035144368267,
      "loss": 4.6249,
      "step": 19356
    },
    {
      "epoch": 2.650917556833744,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004458625507298505,
      "loss": 4.7211,
      "step": 19357
    },
    {
      "epoch": 2.6510545056149,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00044582158705716227,
      "loss": 4.7646,
      "step": 19358
    },
    {
      "epoch": 2.6511914543960557,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004457806234191015,
      "loss": 4.6993,
      "step": 19359
    },
    {
      "epoch": 2.651328403177212,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004457396598160079,
      "loss": 4.7674,
      "step": 19360
    },
    {
      "epoch": 2.6514653519583673,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044569869624822066,
      "loss": 4.7194,
      "step": 19361
    },
    {
      "epoch": 2.6516023007395235,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004456577327160793,
      "loss": 4.5798,
      "step": 19362
    },
    {
      "epoch": 2.6517392495206793,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004456167692199233,
      "loss": 4.6271,
      "step": 19363
    },
    {
      "epoch": 2.651876198301835,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004455758057600923,
      "loss": 4.7298,
      "step": 19364
    },
    {
      "epoch": 2.652013147082991,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00044553484233692556,
      "loss": 4.6077,
      "step": 19365
    },
    {
      "epoch": 2.6521500958641466,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044549387895076274,
      "loss": 4.717,
      "step": 19366
    },
    {
      "epoch": 2.652287044645303,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044545291560194316,
      "loss": 4.6021,
      "step": 19367
    },
    {
      "epoch": 2.6524239934264586,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004454119522908064,
      "loss": 4.642,
      "step": 19368
    },
    {
      "epoch": 2.6525609422076144,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004453709890176918,
      "loss": 4.6608,
      "step": 19369
    },
    {
      "epoch": 2.65269789098877,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004453300257829389,
      "loss": 4.6491,
      "step": 19370
    },
    {
      "epoch": 2.652834839769926,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004452890625868873,
      "loss": 4.6202,
      "step": 19371
    },
    {
      "epoch": 2.6529717885510817,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004452480994298763,
      "loss": 4.7196,
      "step": 19372
    },
    {
      "epoch": 2.6531087373322375,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004452071363122455,
      "loss": 4.5817,
      "step": 19373
    },
    {
      "epoch": 2.6532456861133937,
      "grad_norm": 0.34375,
      "learning_rate": 0.00044516617323433414,
      "loss": 4.6804,
      "step": 19374
    },
    {
      "epoch": 2.6533826348945495,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004451252101964821,
      "loss": 4.6776,
      "step": 19375
    },
    {
      "epoch": 2.6535195836757053,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004450842471990284,
      "loss": 4.6364,
      "step": 19376
    },
    {
      "epoch": 2.653656532456861,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00044504328424231285,
      "loss": 4.7561,
      "step": 19377
    },
    {
      "epoch": 2.653793481238017,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00044500232132667464,
      "loss": 4.7383,
      "step": 19378
    },
    {
      "epoch": 2.653930430019173,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00044496135845245344,
      "loss": 4.5925,
      "step": 19379
    },
    {
      "epoch": 2.654067378800329,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00044492039561998877,
      "loss": 4.6732,
      "step": 19380
    },
    {
      "epoch": 2.6542043275814846,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004448794328296198,
      "loss": 4.6764,
      "step": 19381
    },
    {
      "epoch": 2.6543412763626404,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004448384700816864,
      "loss": 4.8146,
      "step": 19382
    },
    {
      "epoch": 2.654478225143796,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00044479750737652765,
      "loss": 4.6506,
      "step": 19383
    },
    {
      "epoch": 2.654615173924952,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004447565447144833,
      "loss": 4.6365,
      "step": 19384
    },
    {
      "epoch": 2.6547521227061077,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004447155820958927,
      "loss": 4.6608,
      "step": 19385
    },
    {
      "epoch": 2.654889071487264,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004446746195210952,
      "loss": 4.6989,
      "step": 19386
    },
    {
      "epoch": 2.6550260202684197,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004446336569904306,
      "loss": 4.6461,
      "step": 19387
    },
    {
      "epoch": 2.6551629690495755,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000444592694504238,
      "loss": 4.7189,
      "step": 19388
    },
    {
      "epoch": 2.6552999178307313,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00044455173206285717,
      "loss": 4.6873,
      "step": 19389
    },
    {
      "epoch": 2.655436866611887,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004445107696666273,
      "loss": 4.7199,
      "step": 19390
    },
    {
      "epoch": 2.6555738153930433,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00044446980731588815,
      "loss": 4.6834,
      "step": 19391
    },
    {
      "epoch": 2.6557107641741986,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004444288450109789,
      "loss": 4.6243,
      "step": 19392
    },
    {
      "epoch": 2.655847712955355,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004443878827522391,
      "loss": 4.7133,
      "step": 19393
    },
    {
      "epoch": 2.6559846617365106,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004443469205400084,
      "loss": 4.6991,
      "step": 19394
    },
    {
      "epoch": 2.6561216105176664,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00044430595837462606,
      "loss": 4.6701,
      "step": 19395
    },
    {
      "epoch": 2.656258559298822,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004442649962564315,
      "loss": 4.6748,
      "step": 19396
    },
    {
      "epoch": 2.656395508079978,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044422403418576436,
      "loss": 4.6657,
      "step": 19397
    },
    {
      "epoch": 2.656532456861134,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004441830721629641,
      "loss": 4.7022,
      "step": 19398
    },
    {
      "epoch": 2.65666940564229,
      "grad_norm": 0.359375,
      "learning_rate": 0.00044414211018837,
      "loss": 4.6815,
      "step": 19399
    },
    {
      "epoch": 2.6568063544234457,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004441011482623216,
      "loss": 4.7531,
      "step": 19400
    },
    {
      "epoch": 2.6569433032046015,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004440601863851585,
      "loss": 4.6583,
      "step": 19401
    },
    {
      "epoch": 2.6570802519857573,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004440192245572201,
      "loss": 4.7057,
      "step": 19402
    },
    {
      "epoch": 2.657217200766913,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00044397826277884565,
      "loss": 4.6428,
      "step": 19403
    },
    {
      "epoch": 2.657354149548069,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004439373010503748,
      "loss": 4.7283,
      "step": 19404
    },
    {
      "epoch": 2.657491098329225,
      "grad_norm": 0.375,
      "learning_rate": 0.0004438963393721471,
      "loss": 4.643,
      "step": 19405
    },
    {
      "epoch": 2.657628047110381,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044385537774450175,
      "loss": 4.6219,
      "step": 19406
    },
    {
      "epoch": 2.6577649958915366,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004438144161677783,
      "loss": 4.6718,
      "step": 19407
    },
    {
      "epoch": 2.6579019446726924,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004437734546423165,
      "loss": 4.7292,
      "step": 19408
    },
    {
      "epoch": 2.658038893453848,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004437324931684554,
      "loss": 4.7434,
      "step": 19409
    },
    {
      "epoch": 2.6581758422350044,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004436915317465346,
      "loss": 4.6984,
      "step": 19410
    },
    {
      "epoch": 2.6583127910161597,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044365057037689363,
      "loss": 4.6426,
      "step": 19411
    },
    {
      "epoch": 2.658449739797316,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004436096090598719,
      "loss": 4.6611,
      "step": 19412
    },
    {
      "epoch": 2.6585866885784717,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00044356864779580894,
      "loss": 4.704,
      "step": 19413
    },
    {
      "epoch": 2.6587236373596275,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004435276865850439,
      "loss": 4.7509,
      "step": 19414
    },
    {
      "epoch": 2.6588605861407832,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004434867254279167,
      "loss": 4.5924,
      "step": 19415
    },
    {
      "epoch": 2.658997534921939,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004434457643247665,
      "loss": 4.6018,
      "step": 19416
    },
    {
      "epoch": 2.6591344837030952,
      "grad_norm": 0.359375,
      "learning_rate": 0.00044340480327593275,
      "loss": 4.6829,
      "step": 19417
    },
    {
      "epoch": 2.659271432484251,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000443363842281755,
      "loss": 4.75,
      "step": 19418
    },
    {
      "epoch": 2.659408381265407,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00044332288134257266,
      "loss": 4.7313,
      "step": 19419
    },
    {
      "epoch": 2.6595453300465626,
      "grad_norm": 0.40625,
      "learning_rate": 0.00044328192045872526,
      "loss": 4.6756,
      "step": 19420
    },
    {
      "epoch": 2.6596822788277183,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000443240959630552,
      "loss": 4.6321,
      "step": 19421
    },
    {
      "epoch": 2.659819227608874,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00044319999885839267,
      "loss": 4.6823,
      "step": 19422
    },
    {
      "epoch": 2.65995617639003,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0004431590381425866,
      "loss": 4.6384,
      "step": 19423
    },
    {
      "epoch": 2.660093125171186,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00044311807748347314,
      "loss": 4.689,
      "step": 19424
    },
    {
      "epoch": 2.660230073952342,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004430771168813918,
      "loss": 4.6292,
      "step": 19425
    },
    {
      "epoch": 2.6603670227334977,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004430361563366821,
      "loss": 4.7184,
      "step": 19426
    },
    {
      "epoch": 2.6605039715146535,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004429951958496835,
      "loss": 4.6988,
      "step": 19427
    },
    {
      "epoch": 2.6606409202958092,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004429542354207351,
      "loss": 4.6109,
      "step": 19428
    },
    {
      "epoch": 2.6607778690769655,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004429132750501769,
      "loss": 4.7058,
      "step": 19429
    },
    {
      "epoch": 2.6609148178581212,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00044287231473834804,
      "loss": 4.6708,
      "step": 19430
    },
    {
      "epoch": 2.661051766639277,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000442831354485588,
      "loss": 4.7576,
      "step": 19431
    },
    {
      "epoch": 2.661188715420433,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004427903942922361,
      "loss": 4.7097,
      "step": 19432
    },
    {
      "epoch": 2.6613256642015886,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00044274943415863205,
      "loss": 4.6939,
      "step": 19433
    },
    {
      "epoch": 2.6614626129827443,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004427084740851151,
      "loss": 4.6781,
      "step": 19434
    },
    {
      "epoch": 2.6615995617639,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0004426675140720247,
      "loss": 4.6892,
      "step": 19435
    },
    {
      "epoch": 2.6617365105450563,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00044262655411970054,
      "loss": 4.7099,
      "step": 19436
    },
    {
      "epoch": 2.661873459326212,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00044258559422848186,
      "loss": 4.6982,
      "step": 19437
    },
    {
      "epoch": 2.662010408107368,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00044254463439870806,
      "loss": 4.6296,
      "step": 19438
    },
    {
      "epoch": 2.6621473568885237,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00044250367463071855,
      "loss": 4.7154,
      "step": 19439
    },
    {
      "epoch": 2.6622843056696794,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000442462714924853,
      "loss": 4.6395,
      "step": 19440
    },
    {
      "epoch": 2.6624212544508357,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0004424217552814508,
      "loss": 4.6839,
      "step": 19441
    },
    {
      "epoch": 2.662558203231991,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004423807957008512,
      "loss": 4.753,
      "step": 19442
    },
    {
      "epoch": 2.662695152013147,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00044233983618339367,
      "loss": 4.6499,
      "step": 19443
    },
    {
      "epoch": 2.662832100794303,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044229887672941795,
      "loss": 4.6718,
      "step": 19444
    },
    {
      "epoch": 2.6629690495754588,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00044225791733926324,
      "loss": 4.6311,
      "step": 19445
    },
    {
      "epoch": 2.6631059983566145,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004422169580132688,
      "loss": 4.7247,
      "step": 19446
    },
    {
      "epoch": 2.6632429471377703,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044217599875177454,
      "loss": 4.7011,
      "step": 19447
    },
    {
      "epoch": 2.6633798959189265,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00044213503955511956,
      "loss": 4.7813,
      "step": 19448
    },
    {
      "epoch": 2.6635168447000823,
      "grad_norm": 0.390625,
      "learning_rate": 0.0004420940804236434,
      "loss": 4.6678,
      "step": 19449
    },
    {
      "epoch": 2.663653793481238,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00044205312135768527,
      "loss": 4.7357,
      "step": 19450
    },
    {
      "epoch": 2.663790742262394,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004420121623575851,
      "loss": 4.7255,
      "step": 19451
    },
    {
      "epoch": 2.6639276910435497,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000441971203423682,
      "loss": 4.6771,
      "step": 19452
    },
    {
      "epoch": 2.6640646398247054,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044193024455631524,
      "loss": 4.6438,
      "step": 19453
    },
    {
      "epoch": 2.664201588605861,
      "grad_norm": 0.375,
      "learning_rate": 0.00044188928575582465,
      "loss": 4.72,
      "step": 19454
    },
    {
      "epoch": 2.6643385373870174,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004418483270225495,
      "loss": 4.6343,
      "step": 19455
    },
    {
      "epoch": 2.664475486168173,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004418073683568292,
      "loss": 4.6872,
      "step": 19456
    },
    {
      "epoch": 2.664612434949329,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000441766409759003,
      "loss": 4.77,
      "step": 19457
    },
    {
      "epoch": 2.6647493837304848,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00044172545122941073,
      "loss": 4.6757,
      "step": 19458
    },
    {
      "epoch": 2.6648863325116405,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004416844927683916,
      "loss": 4.7522,
      "step": 19459
    },
    {
      "epoch": 2.6650232812927968,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00044164353437628506,
      "loss": 4.7114,
      "step": 19460
    },
    {
      "epoch": 2.665160230073952,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00044160257605343036,
      "loss": 4.663,
      "step": 19461
    },
    {
      "epoch": 2.6652971788551083,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004415616178001673,
      "loss": 4.6506,
      "step": 19462
    },
    {
      "epoch": 2.665434127636264,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004415206596168351,
      "loss": 4.6111,
      "step": 19463
    },
    {
      "epoch": 2.66557107641742,
      "grad_norm": 0.34375,
      "learning_rate": 0.00044147970150377303,
      "loss": 4.6818,
      "step": 19464
    },
    {
      "epoch": 2.6657080251985756,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00044143874346132093,
      "loss": 4.7243,
      "step": 19465
    },
    {
      "epoch": 2.6658449739797314,
      "grad_norm": 0.337890625,
      "learning_rate": 0.000441397785489818,
      "loss": 4.6607,
      "step": 19466
    },
    {
      "epoch": 2.6659819227608876,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004413568275896037,
      "loss": 4.6635,
      "step": 19467
    },
    {
      "epoch": 2.6661188715420434,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004413158697610171,
      "loss": 4.6518,
      "step": 19468
    },
    {
      "epoch": 2.666255820323199,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004412749120043983,
      "loss": 4.6413,
      "step": 19469
    },
    {
      "epoch": 2.666392769104355,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004412339543200863,
      "loss": 4.678,
      "step": 19470
    },
    {
      "epoch": 2.6665297178855107,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004411929967084204,
      "loss": 4.6659,
      "step": 19471
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004411520391697405,
      "loss": 4.6198,
      "step": 19472
    },
    {
      "epoch": 2.6668036154478223,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004411110817043857,
      "loss": 4.6829,
      "step": 19473
    },
    {
      "epoch": 2.6669405642289785,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00044107012431269556,
      "loss": 4.6519,
      "step": 19474
    },
    {
      "epoch": 2.6670775130101343,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004410291669950092,
      "loss": 4.6928,
      "step": 19475
    },
    {
      "epoch": 2.66721446179129,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00044098820975166646,
      "loss": 4.6922,
      "step": 19476
    },
    {
      "epoch": 2.667351410572446,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00044094725258300656,
      "loss": 4.6705,
      "step": 19477
    },
    {
      "epoch": 2.6674883593536016,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00044090629548936895,
      "loss": 4.6918,
      "step": 19478
    },
    {
      "epoch": 2.667625308134758,
      "grad_norm": 0.375,
      "learning_rate": 0.0004408653384710928,
      "loss": 4.6466,
      "step": 19479
    },
    {
      "epoch": 2.667762256915913,
      "grad_norm": 0.34375,
      "learning_rate": 0.000440824381528518,
      "loss": 4.6496,
      "step": 19480
    },
    {
      "epoch": 2.6678992056970694,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00044078342466198373,
      "loss": 4.6997,
      "step": 19481
    },
    {
      "epoch": 2.668036154478225,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004407424678718292,
      "loss": 4.6892,
      "step": 19482
    },
    {
      "epoch": 2.668173103259381,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004407015111583943,
      "loss": 4.7595,
      "step": 19483
    },
    {
      "epoch": 2.6683100520405367,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00044066055452201815,
      "loss": 4.6361,
      "step": 19484
    },
    {
      "epoch": 2.6684470008216925,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004406195979630402,
      "loss": 4.7005,
      "step": 19485
    },
    {
      "epoch": 2.6685839496028487,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004405786414817997,
      "loss": 4.6943,
      "step": 19486
    },
    {
      "epoch": 2.6687208983840045,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004405376850786364,
      "loss": 4.6483,
      "step": 19487
    },
    {
      "epoch": 2.6688578471651603,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004404967287538896,
      "loss": 4.7576,
      "step": 19488
    },
    {
      "epoch": 2.668994795946316,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00044045577250789866,
      "loss": 4.6634,
      "step": 19489
    },
    {
      "epoch": 2.669131744727472,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00044041481634100285,
      "loss": 4.6581,
      "step": 19490
    },
    {
      "epoch": 2.6692686935086276,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004403738602535419,
      "loss": 4.6941,
      "step": 19491
    },
    {
      "epoch": 2.6694056422897834,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044033290424585504,
      "loss": 4.6295,
      "step": 19492
    },
    {
      "epoch": 2.6695425910709396,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00044029194831828153,
      "loss": 4.6525,
      "step": 19493
    },
    {
      "epoch": 2.6696795398520954,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004402509924711612,
      "loss": 4.6392,
      "step": 19494
    },
    {
      "epoch": 2.669816488633251,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004402100367048332,
      "loss": 4.7416,
      "step": 19495
    },
    {
      "epoch": 2.669953437414407,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00044016908101963693,
      "loss": 4.6601,
      "step": 19496
    },
    {
      "epoch": 2.6700903861955627,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004401281254159116,
      "loss": 4.657,
      "step": 19497
    },
    {
      "epoch": 2.670227334976719,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004400871698939971,
      "loss": 4.6442,
      "step": 19498
    },
    {
      "epoch": 2.6703642837578747,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0004400462144542326,
      "loss": 4.7112,
      "step": 19499
    },
    {
      "epoch": 2.6705012325390305,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00044000525909695725,
      "loss": 4.6672,
      "step": 19500
    },
    {
      "epoch": 2.6706381813201863,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043996430382251087,
      "loss": 4.6819,
      "step": 19501
    },
    {
      "epoch": 2.670775130101342,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043992334863123277,
      "loss": 4.7216,
      "step": 19502
    },
    {
      "epoch": 2.670912078882498,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00043988239352346224,
      "loss": 4.7397,
      "step": 19503
    },
    {
      "epoch": 2.6710490276636536,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00043984143849953855,
      "loss": 4.7046,
      "step": 19504
    },
    {
      "epoch": 2.67118597644481,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004398004835598015,
      "loss": 4.7011,
      "step": 19505
    },
    {
      "epoch": 2.6713229252259656,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004397595287045902,
      "loss": 4.6598,
      "step": 19506
    },
    {
      "epoch": 2.6714598740071214,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00043971857393424417,
      "loss": 4.684,
      "step": 19507
    },
    {
      "epoch": 2.671596822788277,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00043967761924910254,
      "loss": 4.726,
      "step": 19508
    },
    {
      "epoch": 2.671733771569433,
      "grad_norm": 0.390625,
      "learning_rate": 0.00043963666464950517,
      "loss": 4.6626,
      "step": 19509
    },
    {
      "epoch": 2.671870720350589,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00043959571013579124,
      "loss": 4.744,
      "step": 19510
    },
    {
      "epoch": 2.6720076691317445,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004395547557082999,
      "loss": 4.7004,
      "step": 19511
    },
    {
      "epoch": 2.6721446179129007,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000439513801367371,
      "loss": 4.6989,
      "step": 19512
    },
    {
      "epoch": 2.6722815666940565,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004394728471133437,
      "loss": 4.6197,
      "step": 19513
    },
    {
      "epoch": 2.6724185154752123,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00043943189294655746,
      "loss": 4.7754,
      "step": 19514
    },
    {
      "epoch": 2.672555464256368,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004393909388673515,
      "loss": 4.6331,
      "step": 19515
    },
    {
      "epoch": 2.672692413037524,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004393499848760655,
      "loss": 4.6788,
      "step": 19516
    },
    {
      "epoch": 2.67282936181868,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004393090309730387,
      "loss": 4.7272,
      "step": 19517
    },
    {
      "epoch": 2.672966310599836,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004392680771586103,
      "loss": 4.822,
      "step": 19518
    },
    {
      "epoch": 2.6731032593809916,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004392271234331201,
      "loss": 4.6998,
      "step": 19519
    },
    {
      "epoch": 2.6732402081621474,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004391861697969073,
      "loss": 4.6561,
      "step": 19520
    },
    {
      "epoch": 2.673377156943303,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00043914521625031133,
      "loss": 4.6789,
      "step": 19521
    },
    {
      "epoch": 2.673514105724459,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004391042627936713,
      "loss": 4.6857,
      "step": 19522
    },
    {
      "epoch": 2.6736510545056147,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004390633094273271,
      "loss": 4.6761,
      "step": 19523
    },
    {
      "epoch": 2.673788003286771,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004390223561516178,
      "loss": 4.6881,
      "step": 19524
    },
    {
      "epoch": 2.6739249520679267,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00043898140296688286,
      "loss": 4.7177,
      "step": 19525
    },
    {
      "epoch": 2.6740619008490825,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004389404498734615,
      "loss": 4.6908,
      "step": 19526
    },
    {
      "epoch": 2.6741988496302382,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004388994968716934,
      "loss": 4.7292,
      "step": 19527
    },
    {
      "epoch": 2.674335798411394,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00043885854396191786,
      "loss": 4.715,
      "step": 19528
    },
    {
      "epoch": 2.6744727471925502,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000438817591144474,
      "loss": 4.6612,
      "step": 19529
    },
    {
      "epoch": 2.6746096959737056,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0004387766384197016,
      "loss": 4.5603,
      "step": 19530
    },
    {
      "epoch": 2.674746644754862,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00043873568578793994,
      "loss": 4.6759,
      "step": 19531
    },
    {
      "epoch": 2.6748835935360176,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004386947332495283,
      "loss": 4.6704,
      "step": 19532
    },
    {
      "epoch": 2.6750205423171733,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004386537808048059,
      "loss": 4.6549,
      "step": 19533
    },
    {
      "epoch": 2.675157491098329,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043861282845411256,
      "loss": 4.7312,
      "step": 19534
    },
    {
      "epoch": 2.675294439879485,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00043857187619778735,
      "loss": 4.6831,
      "step": 19535
    },
    {
      "epoch": 2.675431388660641,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004385309240361696,
      "loss": 4.6871,
      "step": 19536
    },
    {
      "epoch": 2.675568337441797,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004384899719695989,
      "loss": 4.8333,
      "step": 19537
    },
    {
      "epoch": 2.6757052862229527,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004384490199984147,
      "loss": 4.7388,
      "step": 19538
    },
    {
      "epoch": 2.6758422350041084,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004384080681229561,
      "loss": 4.7106,
      "step": 19539
    },
    {
      "epoch": 2.6759791837852642,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00043836711634356244,
      "loss": 4.651,
      "step": 19540
    },
    {
      "epoch": 2.67611613256642,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004383261646605735,
      "loss": 4.676,
      "step": 19541
    },
    {
      "epoch": 2.676253081347576,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043828521307432837,
      "loss": 4.7214,
      "step": 19542
    },
    {
      "epoch": 2.676390030128732,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00043824426158516647,
      "loss": 4.656,
      "step": 19543
    },
    {
      "epoch": 2.6765269789098878,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000438203310193427,
      "loss": 4.6826,
      "step": 19544
    },
    {
      "epoch": 2.6766639276910436,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004381623588994497,
      "loss": 4.6535,
      "step": 19545
    },
    {
      "epoch": 2.6768008764721993,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00043812140770357373,
      "loss": 4.6827,
      "step": 19546
    },
    {
      "epoch": 2.676937825253355,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00043808045660613835,
      "loss": 4.6551,
      "step": 19547
    },
    {
      "epoch": 2.6770747740345113,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004380395056074832,
      "loss": 4.7214,
      "step": 19548
    },
    {
      "epoch": 2.677211722815667,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00043799855470794753,
      "loss": 4.6617,
      "step": 19549
    },
    {
      "epoch": 2.677348671596823,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004379576039078707,
      "loss": 4.6876,
      "step": 19550
    },
    {
      "epoch": 2.6774856203779787,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00043791665320759186,
      "loss": 4.7446,
      "step": 19551
    },
    {
      "epoch": 2.6776225691591344,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00043787570260745085,
      "loss": 4.6723,
      "step": 19552
    },
    {
      "epoch": 2.67775951794029,
      "grad_norm": 0.375,
      "learning_rate": 0.0004378347521077867,
      "loss": 4.6917,
      "step": 19553
    },
    {
      "epoch": 2.677896466721446,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00043779380170893894,
      "loss": 4.6534,
      "step": 19554
    },
    {
      "epoch": 2.678033415502602,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00043775285141124665,
      "loss": 4.6991,
      "step": 19555
    },
    {
      "epoch": 2.678170364283758,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043771190121504955,
      "loss": 4.6593,
      "step": 19556
    },
    {
      "epoch": 2.6783073130649138,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00043767095112068685,
      "loss": 4.5902,
      "step": 19557
    },
    {
      "epoch": 2.6784442618460695,
      "grad_norm": 0.328125,
      "learning_rate": 0.00043763000112849774,
      "loss": 4.6607,
      "step": 19558
    },
    {
      "epoch": 2.6785812106272253,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00043758905123882194,
      "loss": 4.6951,
      "step": 19559
    },
    {
      "epoch": 2.6787181594083815,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00043754810145199866,
      "loss": 4.6565,
      "step": 19560
    },
    {
      "epoch": 2.678855108189537,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004375071517683672,
      "loss": 4.7253,
      "step": 19561
    },
    {
      "epoch": 2.678992056970693,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004374662021882668,
      "loss": 4.6806,
      "step": 19562
    },
    {
      "epoch": 2.679129005751849,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00043742525271203714,
      "loss": 4.625,
      "step": 19563
    },
    {
      "epoch": 2.6792659545330046,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00043738430334001736,
      "loss": 4.7286,
      "step": 19564
    },
    {
      "epoch": 2.6794029033141604,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00043734335407254676,
      "loss": 4.7357,
      "step": 19565
    },
    {
      "epoch": 2.679539852095316,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004373024049099649,
      "loss": 4.7468,
      "step": 19566
    },
    {
      "epoch": 2.6796768008764724,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043726145585261107,
      "loss": 4.7104,
      "step": 19567
    },
    {
      "epoch": 2.679813749657628,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00043722050690082464,
      "loss": 4.6913,
      "step": 19568
    },
    {
      "epoch": 2.679950698438784,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004371795580549447,
      "loss": 4.6961,
      "step": 19569
    },
    {
      "epoch": 2.6800876472199398,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000437138609315311,
      "loss": 4.7063,
      "step": 19570
    },
    {
      "epoch": 2.6802245960010955,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00043709766068226265,
      "loss": 4.7091,
      "step": 19571
    },
    {
      "epoch": 2.6803615447822513,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00043705671215613903,
      "loss": 4.7237,
      "step": 19572
    },
    {
      "epoch": 2.680498493563407,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043701576373727954,
      "loss": 4.6496,
      "step": 19573
    },
    {
      "epoch": 2.6806354423445633,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00043697481542602355,
      "loss": 4.7044,
      "step": 19574
    },
    {
      "epoch": 2.680772391125719,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00043693386722271034,
      "loss": 4.7082,
      "step": 19575
    },
    {
      "epoch": 2.680909339906875,
      "grad_norm": 0.328125,
      "learning_rate": 0.00043689291912767916,
      "loss": 4.7548,
      "step": 19576
    },
    {
      "epoch": 2.6810462886880306,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004368519711412697,
      "loss": 4.8243,
      "step": 19577
    },
    {
      "epoch": 2.6811832374691864,
      "grad_norm": 0.359375,
      "learning_rate": 0.00043681102326382105,
      "loss": 4.6715,
      "step": 19578
    },
    {
      "epoch": 2.6813201862503426,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00043677007549567244,
      "loss": 4.6799,
      "step": 19579
    },
    {
      "epoch": 2.681457135031498,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00043672912783716345,
      "loss": 4.614,
      "step": 19580
    },
    {
      "epoch": 2.681594083812654,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043668818028863343,
      "loss": 4.7393,
      "step": 19581
    },
    {
      "epoch": 2.68173103259381,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004366472328504216,
      "loss": 4.731,
      "step": 19582
    },
    {
      "epoch": 2.6818679813749657,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004366062855228671,
      "loss": 4.6581,
      "step": 19583
    },
    {
      "epoch": 2.6820049301561215,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00043656533830630976,
      "loss": 4.6242,
      "step": 19584
    },
    {
      "epoch": 2.6821418789372773,
      "grad_norm": 0.375,
      "learning_rate": 0.00043652439120108865,
      "loss": 4.6828,
      "step": 19585
    },
    {
      "epoch": 2.6822788277184335,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004364834442075431,
      "loss": 4.6885,
      "step": 19586
    },
    {
      "epoch": 2.6824157764995893,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00043644249732601234,
      "loss": 4.6568,
      "step": 19587
    },
    {
      "epoch": 2.682552725280745,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000436401550556836,
      "loss": 4.6414,
      "step": 19588
    },
    {
      "epoch": 2.682689674061901,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043636060390035323,
      "loss": 4.6814,
      "step": 19589
    },
    {
      "epoch": 2.6828266228430566,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0004363196573569033,
      "loss": 4.6505,
      "step": 19590
    },
    {
      "epoch": 2.6829635716242124,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004362787109268256,
      "loss": 4.631,
      "step": 19591
    },
    {
      "epoch": 2.683100520405368,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00043623776461045965,
      "loss": 4.6736,
      "step": 19592
    },
    {
      "epoch": 2.6832374691865244,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004361968184081445,
      "loss": 4.697,
      "step": 19593
    },
    {
      "epoch": 2.68337441796768,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00043615587232021956,
      "loss": 4.7188,
      "step": 19594
    },
    {
      "epoch": 2.683511366748836,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00043611492634702435,
      "loss": 4.6667,
      "step": 19595
    },
    {
      "epoch": 2.6836483155299917,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00043607398048889806,
      "loss": 4.7359,
      "step": 19596
    },
    {
      "epoch": 2.6837852643111475,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00043603303474617984,
      "loss": 4.6763,
      "step": 19597
    },
    {
      "epoch": 2.6839222130923037,
      "grad_norm": 0.375,
      "learning_rate": 0.00043599208911920925,
      "loss": 4.686,
      "step": 19598
    },
    {
      "epoch": 2.684059161873459,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00043595114360832566,
      "loss": 4.602,
      "step": 19599
    },
    {
      "epoch": 2.6841961106546153,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004359101982138682,
      "loss": 4.6674,
      "step": 19600
    },
    {
      "epoch": 2.684333059435771,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00043586925293617637,
      "loss": 4.6709,
      "step": 19601
    },
    {
      "epoch": 2.684470008216927,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0004358283077755893,
      "loss": 4.6977,
      "step": 19602
    },
    {
      "epoch": 2.6846069569980826,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004357873627324465,
      "loss": 4.6296,
      "step": 19603
    },
    {
      "epoch": 2.6847439057792384,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043574641780708714,
      "loss": 4.6102,
      "step": 19604
    },
    {
      "epoch": 2.6848808545603946,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0004357054729998506,
      "loss": 4.6988,
      "step": 19605
    },
    {
      "epoch": 2.6850178033415504,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004356645283110763,
      "loss": 4.6159,
      "step": 19606
    },
    {
      "epoch": 2.685154752122706,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004356235837411034,
      "loss": 4.6895,
      "step": 19607
    },
    {
      "epoch": 2.685291700903862,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004355826392902714,
      "loss": 4.6022,
      "step": 19608
    },
    {
      "epoch": 2.6854286496850177,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043554169495891936,
      "loss": 4.5819,
      "step": 19609
    },
    {
      "epoch": 2.6855655984661735,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00043550075074738685,
      "loss": 4.7235,
      "step": 19610
    },
    {
      "epoch": 2.6857025472473293,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000435459806656013,
      "loss": 4.6557,
      "step": 19611
    },
    {
      "epoch": 2.6858394960284855,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00043541886268513716,
      "loss": 4.6429,
      "step": 19612
    },
    {
      "epoch": 2.6859764448096413,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004353779188350988,
      "loss": 4.6962,
      "step": 19613
    },
    {
      "epoch": 2.686113393590797,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000435336975106237,
      "loss": 4.6869,
      "step": 19614
    },
    {
      "epoch": 2.686250342371953,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043529603149889133,
      "loss": 4.6655,
      "step": 19615
    },
    {
      "epoch": 2.6863872911531086,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00043525508801340077,
      "loss": 4.6129,
      "step": 19616
    },
    {
      "epoch": 2.686524239934265,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000435214144650105,
      "loss": 4.7041,
      "step": 19617
    },
    {
      "epoch": 2.6866611887154206,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000435173201409343,
      "loss": 4.7818,
      "step": 19618
    },
    {
      "epoch": 2.6867981374965764,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004351322582914542,
      "loss": 4.6463,
      "step": 19619
    },
    {
      "epoch": 2.686935086277732,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00043509131529677797,
      "loss": 4.6436,
      "step": 19620
    },
    {
      "epoch": 2.687072035058888,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004350503724256535,
      "loss": 4.6694,
      "step": 19621
    },
    {
      "epoch": 2.6872089838400437,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004350094296784203,
      "loss": 4.6093,
      "step": 19622
    },
    {
      "epoch": 2.6873459326211995,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00043496848705541736,
      "loss": 4.7284,
      "step": 19623
    },
    {
      "epoch": 2.6874828814023557,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004349275445569843,
      "loss": 4.7225,
      "step": 19624
    },
    {
      "epoch": 2.6876198301835115,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00043488660218346016,
      "loss": 4.6601,
      "step": 19625
    },
    {
      "epoch": 2.6877567789646672,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004348456599351844,
      "loss": 4.6968,
      "step": 19626
    },
    {
      "epoch": 2.687893727745823,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043480471781249624,
      "loss": 4.73,
      "step": 19627
    },
    {
      "epoch": 2.688030676526979,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000434763775815735,
      "loss": 4.7274,
      "step": 19628
    },
    {
      "epoch": 2.688167625308135,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00043472283394524003,
      "loss": 4.6711,
      "step": 19629
    },
    {
      "epoch": 2.6883045740892904,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004346818922013505,
      "loss": 4.7604,
      "step": 19630
    },
    {
      "epoch": 2.6884415228704466,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00043464095058440587,
      "loss": 4.628,
      "step": 19631
    },
    {
      "epoch": 2.6885784716516024,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004346000090947453,
      "loss": 4.7,
      "step": 19632
    },
    {
      "epoch": 2.688715420432758,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00043455906773270817,
      "loss": 4.6916,
      "step": 19633
    },
    {
      "epoch": 2.688852369213914,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004345181264986336,
      "loss": 4.7526,
      "step": 19634
    },
    {
      "epoch": 2.6889893179950697,
      "grad_norm": 0.32421875,
      "learning_rate": 0.000434477185392861,
      "loss": 4.7242,
      "step": 19635
    },
    {
      "epoch": 2.689126266776226,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00043443624441572986,
      "loss": 4.714,
      "step": 19636
    },
    {
      "epoch": 2.6892632155573817,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004343953035675792,
      "loss": 4.6809,
      "step": 19637
    },
    {
      "epoch": 2.6894001643385375,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043435436284874814,
      "loss": 4.6843,
      "step": 19638
    },
    {
      "epoch": 2.6895371131196932,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004343134222595765,
      "loss": 4.7202,
      "step": 19639
    },
    {
      "epoch": 2.689674061900849,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004342724818004032,
      "loss": 4.6889,
      "step": 19640
    },
    {
      "epoch": 2.689811010682005,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00043423154147156754,
      "loss": 4.6931,
      "step": 19641
    },
    {
      "epoch": 2.6899479594631606,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004341906012734088,
      "loss": 4.6431,
      "step": 19642
    },
    {
      "epoch": 2.690084908244317,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00043414966120626644,
      "loss": 4.6666,
      "step": 19643
    },
    {
      "epoch": 2.6902218570254726,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00043410872127047965,
      "loss": 4.628,
      "step": 19644
    },
    {
      "epoch": 2.6903588058066283,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004340677814663874,
      "loss": 4.7823,
      "step": 19645
    },
    {
      "epoch": 2.690495754587784,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004340268417943295,
      "loss": 4.6485,
      "step": 19646
    },
    {
      "epoch": 2.69063270336894,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004339859022546449,
      "loss": 4.6268,
      "step": 19647
    },
    {
      "epoch": 2.690769652150096,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004339449628476729,
      "loss": 4.6973,
      "step": 19648
    },
    {
      "epoch": 2.6909066009312514,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004339040235737528,
      "loss": 4.6171,
      "step": 19649
    },
    {
      "epoch": 2.6910435497124077,
      "grad_norm": 0.34375,
      "learning_rate": 0.000433863084433224,
      "loss": 4.6208,
      "step": 19650
    },
    {
      "epoch": 2.6911804984935634,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00043382214542642567,
      "loss": 4.64,
      "step": 19651
    },
    {
      "epoch": 2.691317447274719,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004337812065536969,
      "loss": 4.7157,
      "step": 19652
    },
    {
      "epoch": 2.691454396055875,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004337402678153773,
      "loss": 4.6559,
      "step": 19653
    },
    {
      "epoch": 2.6915913448370308,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000433699329211806,
      "loss": 4.6551,
      "step": 19654
    },
    {
      "epoch": 2.691728293618187,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004336583907433222,
      "loss": 4.6832,
      "step": 19655
    },
    {
      "epoch": 2.6918652423993428,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00043361745241026503,
      "loss": 4.6736,
      "step": 19656
    },
    {
      "epoch": 2.6920021911804985,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00043357651421297415,
      "loss": 4.6653,
      "step": 19657
    },
    {
      "epoch": 2.6921391399616543,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00043353557615178863,
      "loss": 4.7027,
      "step": 19658
    },
    {
      "epoch": 2.69227608874281,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00043349463822704747,
      "loss": 4.6619,
      "step": 19659
    },
    {
      "epoch": 2.692413037523966,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004334537004390904,
      "loss": 4.7029,
      "step": 19660
    },
    {
      "epoch": 2.6925499863051217,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00043341276278825643,
      "loss": 4.6729,
      "step": 19661
    },
    {
      "epoch": 2.692686935086278,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00043337182527488483,
      "loss": 4.6241,
      "step": 19662
    },
    {
      "epoch": 2.6928238838674337,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004333308878993148,
      "loss": 4.6247,
      "step": 19663
    },
    {
      "epoch": 2.6929608326485894,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00043328995066188575,
      "loss": 4.733,
      "step": 19664
    },
    {
      "epoch": 2.693097781429745,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004332490135629369,
      "loss": 4.6831,
      "step": 19665
    },
    {
      "epoch": 2.693234730210901,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043320807660280746,
      "loss": 4.6752,
      "step": 19666
    },
    {
      "epoch": 2.693371678992057,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004331671397818365,
      "loss": 4.6704,
      "step": 19667
    },
    {
      "epoch": 2.693508627773213,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004331262031003636,
      "loss": 4.6785,
      "step": 19668
    },
    {
      "epoch": 2.6936455765543688,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000433085266558728,
      "loss": 4.6655,
      "step": 19669
    },
    {
      "epoch": 2.6937825253355245,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043304433015726855,
      "loss": 4.6476,
      "step": 19670
    },
    {
      "epoch": 2.6939194741166803,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043300339389632496,
      "loss": 4.6774,
      "step": 19671
    },
    {
      "epoch": 2.694056422897836,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004329624577762363,
      "loss": 4.6806,
      "step": 19672
    },
    {
      "epoch": 2.694193371678992,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043292152179734184,
      "loss": 4.6413,
      "step": 19673
    },
    {
      "epoch": 2.694330320460148,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004328805859599806,
      "loss": 4.7237,
      "step": 19674
    },
    {
      "epoch": 2.694467269241304,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00043283965026449223,
      "loss": 4.6675,
      "step": 19675
    },
    {
      "epoch": 2.6946042180224596,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004327987147112157,
      "loss": 4.7488,
      "step": 19676
    },
    {
      "epoch": 2.6947411668036154,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0004327577793004902,
      "loss": 4.6939,
      "step": 19677
    },
    {
      "epoch": 2.694878115584771,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043271684403265527,
      "loss": 4.819,
      "step": 19678
    },
    {
      "epoch": 2.6950150643659274,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043267590890805,
      "loss": 4.681,
      "step": 19679
    },
    {
      "epoch": 2.6951520131470827,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004326349739270135,
      "loss": 4.7015,
      "step": 19680
    },
    {
      "epoch": 2.695288961928239,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000432594039089885,
      "loss": 4.7251,
      "step": 19681
    },
    {
      "epoch": 2.6954259107093947,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000432553104397004,
      "loss": 4.5733,
      "step": 19682
    },
    {
      "epoch": 2.6955628594905505,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0004325121698487096,
      "loss": 4.7187,
      "step": 19683
    },
    {
      "epoch": 2.6956998082717063,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043247123544534104,
      "loss": 4.6773,
      "step": 19684
    },
    {
      "epoch": 2.695836757052862,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004324303011872373,
      "loss": 4.6849,
      "step": 19685
    },
    {
      "epoch": 2.6959737058340183,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00043238936707473805,
      "loss": 4.6842,
      "step": 19686
    },
    {
      "epoch": 2.696110654615174,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043234843310818234,
      "loss": 4.6837,
      "step": 19687
    },
    {
      "epoch": 2.69624760339633,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004323074992879091,
      "loss": 4.7546,
      "step": 19688
    },
    {
      "epoch": 2.6963845521774856,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004322665656142581,
      "loss": 4.6624,
      "step": 19689
    },
    {
      "epoch": 2.6965215009586414,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004322256320875682,
      "loss": 4.6369,
      "step": 19690
    },
    {
      "epoch": 2.696658449739797,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043218469870817883,
      "loss": 4.6237,
      "step": 19691
    },
    {
      "epoch": 2.696795398520953,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00043214376547642884,
      "loss": 4.7,
      "step": 19692
    },
    {
      "epoch": 2.696932347302109,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00043210283239265797,
      "loss": 4.6633,
      "step": 19693
    },
    {
      "epoch": 2.697069296083265,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004320618994572052,
      "loss": 4.6573,
      "step": 19694
    },
    {
      "epoch": 2.6972062448644207,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004320209666704095,
      "loss": 4.6642,
      "step": 19695
    },
    {
      "epoch": 2.6973431936455765,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00043198003403261055,
      "loss": 4.7618,
      "step": 19696
    },
    {
      "epoch": 2.6974801424267323,
      "grad_norm": 0.359375,
      "learning_rate": 0.00043193910154414733,
      "loss": 4.6204,
      "step": 19697
    },
    {
      "epoch": 2.6976170912078885,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0004318981692053591,
      "loss": 4.7261,
      "step": 19698
    },
    {
      "epoch": 2.697754039989044,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043185723701658486,
      "loss": 4.684,
      "step": 19699
    },
    {
      "epoch": 2.6978909887702,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004318163049781642,
      "loss": 4.6534,
      "step": 19700
    },
    {
      "epoch": 2.698027937551356,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043177537309043617,
      "loss": 4.7322,
      "step": 19701
    },
    {
      "epoch": 2.6981648863325116,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00043173444135373993,
      "loss": 4.5811,
      "step": 19702
    },
    {
      "epoch": 2.6983018351136674,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0004316935097684145,
      "loss": 4.7164,
      "step": 19703
    },
    {
      "epoch": 2.698438783894823,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004316525783347996,
      "loss": 4.719,
      "step": 19704
    },
    {
      "epoch": 2.6985757326759794,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00043161164705323417,
      "loss": 4.5791,
      "step": 19705
    },
    {
      "epoch": 2.698712681457135,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004315707159240571,
      "loss": 4.7258,
      "step": 19706
    },
    {
      "epoch": 2.698849630238291,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004315297849476082,
      "loss": 4.59,
      "step": 19707
    },
    {
      "epoch": 2.6989865790194467,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00043148885412422637,
      "loss": 4.6643,
      "step": 19708
    },
    {
      "epoch": 2.6991235278006025,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004314479234542508,
      "loss": 4.7688,
      "step": 19709
    },
    {
      "epoch": 2.6992604765817583,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00043140699293802044,
      "loss": 4.6377,
      "step": 19710
    },
    {
      "epoch": 2.699397425362914,
      "grad_norm": 0.353515625,
      "learning_rate": 0.000431366062575875,
      "loss": 4.6848,
      "step": 19711
    },
    {
      "epoch": 2.6995343741440703,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004313251323681535,
      "loss": 4.7675,
      "step": 19712
    },
    {
      "epoch": 2.699671322925226,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00043128420231519496,
      "loss": 4.7141,
      "step": 19713
    },
    {
      "epoch": 2.699808271706382,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043124327241733857,
      "loss": 4.5672,
      "step": 19714
    },
    {
      "epoch": 2.6999452204875376,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00043120234267492386,
      "loss": 4.6986,
      "step": 19715
    },
    {
      "epoch": 2.7000821692686934,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0004311614130882898,
      "loss": 4.7182,
      "step": 19716
    },
    {
      "epoch": 2.7002191180498496,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004311204836577754,
      "loss": 4.7362,
      "step": 19717
    },
    {
      "epoch": 2.700356066831005,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00043107955438372026,
      "loss": 4.6553,
      "step": 19718
    },
    {
      "epoch": 2.700493015612161,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004310386252664633,
      "loss": 4.6769,
      "step": 19719
    },
    {
      "epoch": 2.700629964393317,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004309976963063439,
      "loss": 4.6702,
      "step": 19720
    },
    {
      "epoch": 2.7007669131744727,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004309567675037008,
      "loss": 4.6982,
      "step": 19721
    },
    {
      "epoch": 2.7009038619556285,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004309158388588737,
      "loss": 4.7406,
      "step": 19722
    },
    {
      "epoch": 2.7010408107367843,
      "grad_norm": 0.359375,
      "learning_rate": 0.00043087491037220165,
      "loss": 4.7284,
      "step": 19723
    },
    {
      "epoch": 2.7011777595179405,
      "grad_norm": 0.3125,
      "learning_rate": 0.00043083398204402356,
      "loss": 4.7513,
      "step": 19724
    },
    {
      "epoch": 2.7013147082990963,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004307930538746791,
      "loss": 4.748,
      "step": 19725
    },
    {
      "epoch": 2.701451657080252,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00043075212586450703,
      "loss": 4.7399,
      "step": 19726
    },
    {
      "epoch": 2.701588605861408,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00043071119801384676,
      "loss": 4.6663,
      "step": 19727
    },
    {
      "epoch": 2.7017255546425636,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004306702703230372,
      "loss": 4.696,
      "step": 19728
    },
    {
      "epoch": 2.7018625034237194,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00043062934279241796,
      "loss": 4.6895,
      "step": 19729
    },
    {
      "epoch": 2.701999452204875,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043058841542232786,
      "loss": 4.6041,
      "step": 19730
    },
    {
      "epoch": 2.7021364009860314,
      "grad_norm": 0.359375,
      "learning_rate": 0.00043054748821310626,
      "loss": 4.641,
      "step": 19731
    },
    {
      "epoch": 2.702273349767187,
      "grad_norm": 0.34375,
      "learning_rate": 0.00043050656116509203,
      "loss": 4.6649,
      "step": 19732
    },
    {
      "epoch": 2.702410298548343,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004304656342786248,
      "loss": 4.6764,
      "step": 19733
    },
    {
      "epoch": 2.7025472473294987,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004304247075540435,
      "loss": 4.699,
      "step": 19734
    },
    {
      "epoch": 2.7026841961106545,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00043038378099168707,
      "loss": 4.6701,
      "step": 19735
    },
    {
      "epoch": 2.7028211448918107,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00043034285459189514,
      "loss": 4.6991,
      "step": 19736
    },
    {
      "epoch": 2.7029580936729665,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004303019283550067,
      "loss": 4.6458,
      "step": 19737
    },
    {
      "epoch": 2.7030950424541222,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004302610022813609,
      "loss": 4.7417,
      "step": 19738
    },
    {
      "epoch": 2.703231991235278,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004302200763712966,
      "loss": 4.6615,
      "step": 19739
    },
    {
      "epoch": 2.703368940016434,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004301791506251535,
      "loss": 4.5976,
      "step": 19740
    },
    {
      "epoch": 2.7035058887975896,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004301382250432704,
      "loss": 4.734,
      "step": 19741
    },
    {
      "epoch": 2.7036428375787454,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00043009729962598646,
      "loss": 4.7387,
      "step": 19742
    },
    {
      "epoch": 2.7037797863599016,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00043005637437364114,
      "loss": 4.667,
      "step": 19743
    },
    {
      "epoch": 2.7039167351410573,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00043001544928657334,
      "loss": 4.6976,
      "step": 19744
    },
    {
      "epoch": 2.704053683922213,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004299745243651223,
      "loss": 4.6558,
      "step": 19745
    },
    {
      "epoch": 2.704190632703369,
      "grad_norm": 0.333984375,
      "learning_rate": 0.000429933599609627,
      "loss": 4.6518,
      "step": 19746
    },
    {
      "epoch": 2.7043275814845247,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00042989267502042684,
      "loss": 4.6095,
      "step": 19747
    },
    {
      "epoch": 2.704464530265681,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00042985175059786097,
      "loss": 4.6693,
      "step": 19748
    },
    {
      "epoch": 2.7046014790468362,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0004298108263422684,
      "loss": 4.6608,
      "step": 19749
    },
    {
      "epoch": 2.7047384278279925,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00042976990225398814,
      "loss": 4.7687,
      "step": 19750
    },
    {
      "epoch": 2.7048753766091482,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004297289783333597,
      "loss": 4.6582,
      "step": 19751
    },
    {
      "epoch": 2.705012325390304,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00042968805458072206,
      "loss": 4.7157,
      "step": 19752
    },
    {
      "epoch": 2.70514927417146,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004296471309964141,
      "loss": 4.6649,
      "step": 19753
    },
    {
      "epoch": 2.7052862229526156,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004296062075807755,
      "loss": 4.6512,
      "step": 19754
    },
    {
      "epoch": 2.705423171733772,
      "grad_norm": 0.375,
      "learning_rate": 0.00042956528433414507,
      "loss": 4.6872,
      "step": 19755
    },
    {
      "epoch": 2.7055601205149276,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000429524361256862,
      "loss": 4.6374,
      "step": 19756
    },
    {
      "epoch": 2.7056970692960833,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00042948343834926523,
      "loss": 4.6376,
      "step": 19757
    },
    {
      "epoch": 2.705834018077239,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004294425156116943,
      "loss": 4.7364,
      "step": 19758
    },
    {
      "epoch": 2.705970966858395,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00042940159304448813,
      "loss": 4.6439,
      "step": 19759
    },
    {
      "epoch": 2.7061079156395507,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004293606706479857,
      "loss": 4.7177,
      "step": 19760
    },
    {
      "epoch": 2.7062448644207064,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004293197484225265,
      "loss": 4.5976,
      "step": 19761
    },
    {
      "epoch": 2.7063818132018627,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004292788263684494,
      "loss": 4.738,
      "step": 19762
    },
    {
      "epoch": 2.7065187619830184,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0004292379044860937,
      "loss": 4.6498,
      "step": 19763
    },
    {
      "epoch": 2.706655710764174,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004291969827757982,
      "loss": 4.6778,
      "step": 19764
    },
    {
      "epoch": 2.70679265954533,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004291560612379025,
      "loss": 4.7386,
      "step": 19765
    },
    {
      "epoch": 2.7069296083264858,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042911513987274537,
      "loss": 4.6907,
      "step": 19766
    },
    {
      "epoch": 2.707066557107642,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004290742186806662,
      "loss": 4.7041,
      "step": 19767
    },
    {
      "epoch": 2.7072035058887973,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00042903329766200366,
      "loss": 4.6901,
      "step": 19768
    },
    {
      "epoch": 2.7073404546699535,
      "grad_norm": 0.328125,
      "learning_rate": 0.00042899237681709744,
      "loss": 4.6607,
      "step": 19769
    },
    {
      "epoch": 2.7074774034511093,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00042895145614628633,
      "loss": 4.6458,
      "step": 19770
    },
    {
      "epoch": 2.707614352232265,
      "grad_norm": 0.40625,
      "learning_rate": 0.00042891053564990937,
      "loss": 4.7373,
      "step": 19771
    },
    {
      "epoch": 2.707751301013421,
      "grad_norm": 0.3515625,
      "learning_rate": 0.000428869615328306,
      "loss": 4.6759,
      "step": 19772
    },
    {
      "epoch": 2.7078882497945767,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004288286951818152,
      "loss": 4.6962,
      "step": 19773
    },
    {
      "epoch": 2.708025198575733,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000428787775210776,
      "loss": 4.7598,
      "step": 19774
    },
    {
      "epoch": 2.7081621473568886,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004287468554155274,
      "loss": 4.6699,
      "step": 19775
    },
    {
      "epoch": 2.7082990961380444,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004287059357964089,
      "loss": 4.6589,
      "step": 19776
    },
    {
      "epoch": 2.7084360449192,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004286650163537593,
      "loss": 4.7496,
      "step": 19777
    },
    {
      "epoch": 2.708572993700356,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00042862409708791785,
      "loss": 4.6906,
      "step": 19778
    },
    {
      "epoch": 2.7087099424815118,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004285831779992234,
      "loss": 4.6846,
      "step": 19779
    },
    {
      "epoch": 2.7088468912626675,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00042854225908801544,
      "loss": 4.6656,
      "step": 19780
    },
    {
      "epoch": 2.7089838400438238,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004285013403546329,
      "loss": 4.6097,
      "step": 19781
    },
    {
      "epoch": 2.7091207888249795,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00042846042179941465,
      "loss": 4.6563,
      "step": 19782
    },
    {
      "epoch": 2.7092577376061353,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00042841950342270026,
      "loss": 4.8101,
      "step": 19783
    },
    {
      "epoch": 2.709394686387291,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004283785852248286,
      "loss": 4.7185,
      "step": 19784
    },
    {
      "epoch": 2.709531635168447,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004283376672061387,
      "loss": 4.6972,
      "step": 19785
    },
    {
      "epoch": 2.709668583949603,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00042829674936696957,
      "loss": 4.7008,
      "step": 19786
    },
    {
      "epoch": 2.709805532730759,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004282558317076606,
      "loss": 4.6367,
      "step": 19787
    },
    {
      "epoch": 2.7099424815119146,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004282149142285508,
      "loss": 4.6766,
      "step": 19788
    },
    {
      "epoch": 2.7100794302930704,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004281739969299789,
      "loss": 4.6532,
      "step": 19789
    },
    {
      "epoch": 2.710216379074226,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004281330798122845,
      "loss": 4.658,
      "step": 19790
    },
    {
      "epoch": 2.710353327855382,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004280921628758065,
      "loss": 4.7733,
      "step": 19791
    },
    {
      "epoch": 2.7104902766365377,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004280512461208841,
      "loss": 4.742,
      "step": 19792
    },
    {
      "epoch": 2.710627225417694,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042801032954785585,
      "loss": 4.6436,
      "step": 19793
    },
    {
      "epoch": 2.7107641741988497,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042796941315706156,
      "loss": 4.7835,
      "step": 19794
    },
    {
      "epoch": 2.7109011229800055,
      "grad_norm": 0.359375,
      "learning_rate": 0.00042792849694883993,
      "loss": 4.662,
      "step": 19795
    },
    {
      "epoch": 2.7110380717611613,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00042788758092353015,
      "loss": 4.6846,
      "step": 19796
    },
    {
      "epoch": 2.711175020542317,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000427846665081471,
      "loss": 4.6755,
      "step": 19797
    },
    {
      "epoch": 2.7113119693234733,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004278057494230021,
      "loss": 4.6688,
      "step": 19798
    },
    {
      "epoch": 2.7114489181046286,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00042776483394846217,
      "loss": 4.766,
      "step": 19799
    },
    {
      "epoch": 2.711585866885785,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004277239186581902,
      "loss": 4.7473,
      "step": 19800
    },
    {
      "epoch": 2.7117228156669406,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004276830035525256,
      "loss": 4.6971,
      "step": 19801
    },
    {
      "epoch": 2.7118597644480964,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004276420886318073,
      "loss": 4.6198,
      "step": 19802
    },
    {
      "epoch": 2.711996713229252,
      "grad_norm": 0.359375,
      "learning_rate": 0.00042760117389637435,
      "loss": 4.7236,
      "step": 19803
    },
    {
      "epoch": 2.712133662010408,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00042756025934656555,
      "loss": 4.6286,
      "step": 19804
    },
    {
      "epoch": 2.712270610791564,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004275193449827204,
      "loss": 4.6889,
      "step": 19805
    },
    {
      "epoch": 2.71240755957272,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004274784308051779,
      "loss": 4.7051,
      "step": 19806
    },
    {
      "epoch": 2.7125445083538757,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004274375168142768,
      "loss": 4.6374,
      "step": 19807
    },
    {
      "epoch": 2.7126814571350315,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00042739660301035646,
      "loss": 4.7411,
      "step": 19808
    },
    {
      "epoch": 2.7128184059161873,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004273556893937559,
      "loss": 4.7287,
      "step": 19809
    },
    {
      "epoch": 2.712955354697343,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00042731477596481425,
      "loss": 4.6905,
      "step": 19810
    },
    {
      "epoch": 2.713092303478499,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004272738627238702,
      "loss": 4.7501,
      "step": 19811
    },
    {
      "epoch": 2.713229252259655,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042723294967126317,
      "loss": 4.7255,
      "step": 19812
    },
    {
      "epoch": 2.713366201040811,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004271920368073322,
      "loss": 4.6715,
      "step": 19813
    },
    {
      "epoch": 2.7135031498219666,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00042715112413241617,
      "loss": 4.6222,
      "step": 19814
    },
    {
      "epoch": 2.7136400986031224,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00042711021164685414,
      "loss": 4.6569,
      "step": 19815
    },
    {
      "epoch": 2.713777047384278,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00042706929935098543,
      "loss": 4.6936,
      "step": 19816
    },
    {
      "epoch": 2.7139139961654344,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00042702838724514885,
      "loss": 4.7207,
      "step": 19817
    },
    {
      "epoch": 2.7140509449465897,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004269874753296833,
      "loss": 4.679,
      "step": 19818
    },
    {
      "epoch": 2.714187893727746,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004269465636049283,
      "loss": 4.6941,
      "step": 19819
    },
    {
      "epoch": 2.7143248425089017,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0004269056520712225,
      "loss": 4.6206,
      "step": 19820
    },
    {
      "epoch": 2.7144617912900575,
      "grad_norm": 0.359375,
      "learning_rate": 0.000426864740728905,
      "loss": 4.6836,
      "step": 19821
    },
    {
      "epoch": 2.7145987400712133,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004268238295783149,
      "loss": 4.7158,
      "step": 19822
    },
    {
      "epoch": 2.714735688852369,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004267829186197914,
      "loss": 4.7235,
      "step": 19823
    },
    {
      "epoch": 2.7148726376335253,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00042674200785367336,
      "loss": 4.657,
      "step": 19824
    },
    {
      "epoch": 2.715009586414681,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00042670109728029973,
      "loss": 4.7209,
      "step": 19825
    },
    {
      "epoch": 2.715146535195837,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00042666018690000967,
      "loss": 4.7328,
      "step": 19826
    },
    {
      "epoch": 2.7152834839769926,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00042661927671314235,
      "loss": 4.7125,
      "step": 19827
    },
    {
      "epoch": 2.7154204327581484,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042657836672003644,
      "loss": 4.6853,
      "step": 19828
    },
    {
      "epoch": 2.715557381539304,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004265374569210313,
      "loss": 4.718,
      "step": 19829
    },
    {
      "epoch": 2.71569433032046,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00042649654731646586,
      "loss": 4.7463,
      "step": 19830
    },
    {
      "epoch": 2.715831279101616,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004264556379066792,
      "loss": 4.6615,
      "step": 19831
    },
    {
      "epoch": 2.715968227882772,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004264147286920101,
      "loss": 4.6719,
      "step": 19832
    },
    {
      "epoch": 2.7161051766639277,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042637381967279786,
      "loss": 4.7399,
      "step": 19833
    },
    {
      "epoch": 2.7162421254450835,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00042633291084938146,
      "loss": 4.7073,
      "step": 19834
    },
    {
      "epoch": 2.7163790742262393,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0004262920022220998,
      "loss": 4.6249,
      "step": 19835
    },
    {
      "epoch": 2.7165160230073955,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004262510937912919,
      "loss": 4.7634,
      "step": 19836
    },
    {
      "epoch": 2.716652971788551,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000426210185557297,
      "loss": 4.6781,
      "step": 19837
    },
    {
      "epoch": 2.716789920569707,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042616927752045383,
      "loss": 4.6544,
      "step": 19838
    },
    {
      "epoch": 2.716926869350863,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00042612836968110166,
      "loss": 4.6901,
      "step": 19839
    },
    {
      "epoch": 2.7170638181320186,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042608746203957925,
      "loss": 4.6344,
      "step": 19840
    },
    {
      "epoch": 2.7172007669131744,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042604655459622584,
      "loss": 4.6451,
      "step": 19841
    },
    {
      "epoch": 2.71733771569433,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00042600564735138026,
      "loss": 4.655,
      "step": 19842
    },
    {
      "epoch": 2.7174746644754864,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004259647403053817,
      "loss": 4.6213,
      "step": 19843
    },
    {
      "epoch": 2.717611613256642,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00042592383345856896,
      "loss": 4.7034,
      "step": 19844
    },
    {
      "epoch": 2.717748562037798,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00042588292681128114,
      "loss": 4.709,
      "step": 19845
    },
    {
      "epoch": 2.7178855108189537,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00042584202036385736,
      "loss": 4.707,
      "step": 19846
    },
    {
      "epoch": 2.7180224596001095,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004258011141166364,
      "loss": 4.6605,
      "step": 19847
    },
    {
      "epoch": 2.7181594083812652,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004257602080699575,
      "loss": 4.7218,
      "step": 19848
    },
    {
      "epoch": 2.718296357162421,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0004257193022241594,
      "loss": 4.6401,
      "step": 19849
    },
    {
      "epoch": 2.7184333059435772,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004256783965795813,
      "loss": 4.606,
      "step": 19850
    },
    {
      "epoch": 2.718570254724733,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004256374911365621,
      "loss": 4.7392,
      "step": 19851
    },
    {
      "epoch": 2.718707203505889,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004255965858954408,
      "loss": 4.76,
      "step": 19852
    },
    {
      "epoch": 2.7188441522870446,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004255556808565565,
      "loss": 4.7497,
      "step": 19853
    },
    {
      "epoch": 2.7189811010682003,
      "grad_norm": 0.328125,
      "learning_rate": 0.000425514776020248,
      "loss": 4.7239,
      "step": 19854
    },
    {
      "epoch": 2.7191180498493566,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00042547387138685443,
      "loss": 4.6752,
      "step": 19855
    },
    {
      "epoch": 2.7192549986305123,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00042543296695671473,
      "loss": 4.645,
      "step": 19856
    },
    {
      "epoch": 2.719391947411668,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0004253920627301679,
      "loss": 4.6129,
      "step": 19857
    },
    {
      "epoch": 2.719528896192824,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004253511587075529,
      "loss": 4.7316,
      "step": 19858
    },
    {
      "epoch": 2.7196658449739797,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004253102548892087,
      "loss": 4.6919,
      "step": 19859
    },
    {
      "epoch": 2.7198027937551355,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004252693512754743,
      "loss": 4.6761,
      "step": 19860
    },
    {
      "epoch": 2.7199397425362912,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00042522844786668876,
      "loss": 4.6032,
      "step": 19861
    },
    {
      "epoch": 2.7200766913174474,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042518754466319084,
      "loss": 4.6965,
      "step": 19862
    },
    {
      "epoch": 2.7202136400986032,
      "grad_norm": 0.328125,
      "learning_rate": 0.00042514664166531967,
      "loss": 4.7154,
      "step": 19863
    },
    {
      "epoch": 2.720350588879759,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00042510573887341433,
      "loss": 4.6707,
      "step": 19864
    },
    {
      "epoch": 2.720487537660915,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004250648362878135,
      "loss": 4.7741,
      "step": 19865
    },
    {
      "epoch": 2.7206244864420706,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004250239339088564,
      "loss": 4.642,
      "step": 19866
    },
    {
      "epoch": 2.7207614352232268,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00042498303173688193,
      "loss": 4.7138,
      "step": 19867
    },
    {
      "epoch": 2.720898384004382,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004249421297722291,
      "loss": 4.6711,
      "step": 19868
    },
    {
      "epoch": 2.7210353327855383,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004249012280152366,
      "loss": 4.6057,
      "step": 19869
    },
    {
      "epoch": 2.721172281566694,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00042486032646624374,
      "loss": 4.7856,
      "step": 19870
    },
    {
      "epoch": 2.72130923034785,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00042481942512558935,
      "loss": 4.6612,
      "step": 19871
    },
    {
      "epoch": 2.7214461791290057,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004247785239936124,
      "loss": 4.6825,
      "step": 19872
    },
    {
      "epoch": 2.7215831279101614,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004247376230706517,
      "loss": 4.6795,
      "step": 19873
    },
    {
      "epoch": 2.7217200766913177,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00042469672235704653,
      "loss": 4.6678,
      "step": 19874
    },
    {
      "epoch": 2.7218570254724734,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004246558218531356,
      "loss": 4.6127,
      "step": 19875
    },
    {
      "epoch": 2.721993974253629,
      "grad_norm": 0.359375,
      "learning_rate": 0.0004246149215592578,
      "loss": 4.6398,
      "step": 19876
    },
    {
      "epoch": 2.722130923034785,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004245740214757522,
      "loss": 4.6441,
      "step": 19877
    },
    {
      "epoch": 2.7222678718159408,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004245331216029579,
      "loss": 4.7153,
      "step": 19878
    },
    {
      "epoch": 2.7224048205970965,
      "grad_norm": 0.328125,
      "learning_rate": 0.00042449222194121367,
      "loss": 4.6238,
      "step": 19879
    },
    {
      "epoch": 2.7225417693782523,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00042445132249085824,
      "loss": 4.6736,
      "step": 19880
    },
    {
      "epoch": 2.7226787181594085,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000424410423252231,
      "loss": 4.6969,
      "step": 19881
    },
    {
      "epoch": 2.7228156669405643,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004243695242256707,
      "loss": 4.6945,
      "step": 19882
    },
    {
      "epoch": 2.72295261572172,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042432862541151616,
      "loss": 4.6904,
      "step": 19883
    },
    {
      "epoch": 2.723089564502876,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00042428772681010643,
      "loss": 4.745,
      "step": 19884
    },
    {
      "epoch": 2.7232265132840316,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004242468284217805,
      "loss": 4.7266,
      "step": 19885
    },
    {
      "epoch": 2.723363462065188,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00042420593024687723,
      "loss": 4.6126,
      "step": 19886
    },
    {
      "epoch": 2.723500410846343,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004241650322857354,
      "loss": 4.659,
      "step": 19887
    },
    {
      "epoch": 2.7236373596274994,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00042412413453869427,
      "loss": 4.6914,
      "step": 19888
    },
    {
      "epoch": 2.723774308408655,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004240832370060926,
      "loss": 4.7357,
      "step": 19889
    },
    {
      "epoch": 2.723911257189811,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00042404233968826934,
      "loss": 4.7094,
      "step": 19890
    },
    {
      "epoch": 2.7240482059709668,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004240014425855632,
      "loss": 4.6555,
      "step": 19891
    },
    {
      "epoch": 2.7241851547521225,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004239605456983135,
      "loss": 4.6322,
      "step": 19892
    },
    {
      "epoch": 2.7243221035332787,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004239196490268589,
      "loss": 4.7038,
      "step": 19893
    },
    {
      "epoch": 2.7244590523144345,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042387875257153833,
      "loss": 4.6213,
      "step": 19894
    },
    {
      "epoch": 2.7245960010955903,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00042383785633269083,
      "loss": 4.7201,
      "step": 19895
    },
    {
      "epoch": 2.724732949876746,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004237969603106553,
      "loss": 4.6496,
      "step": 19896
    },
    {
      "epoch": 2.724869898657902,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004237560645057706,
      "loss": 4.6269,
      "step": 19897
    },
    {
      "epoch": 2.7250068474390576,
      "grad_norm": 0.34375,
      "learning_rate": 0.00042371516891837545,
      "loss": 4.6918,
      "step": 19898
    },
    {
      "epoch": 2.7251437962202134,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004236742735488091,
      "loss": 4.6581,
      "step": 19899
    },
    {
      "epoch": 2.7252807450013696,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042363337839741044,
      "loss": 4.7269,
      "step": 19900
    },
    {
      "epoch": 2.7254176937825254,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00042359248346451796,
      "loss": 4.7509,
      "step": 19901
    },
    {
      "epoch": 2.725554642563681,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0004235515887504711,
      "loss": 4.7107,
      "step": 19902
    },
    {
      "epoch": 2.725691591344837,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004235106942556085,
      "loss": 4.6775,
      "step": 19903
    },
    {
      "epoch": 2.7258285401259927,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00042346979998026914,
      "loss": 4.6859,
      "step": 19904
    },
    {
      "epoch": 2.725965488907149,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004234289059247916,
      "loss": 4.6695,
      "step": 19905
    },
    {
      "epoch": 2.7261024376883047,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0004233880120895153,
      "loss": 4.7207,
      "step": 19906
    },
    {
      "epoch": 2.7262393864694605,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004233471184747789,
      "loss": 4.6857,
      "step": 19907
    },
    {
      "epoch": 2.7263763352506163,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004233062250809212,
      "loss": 4.6132,
      "step": 19908
    },
    {
      "epoch": 2.726513284031772,
      "grad_norm": 0.365234375,
      "learning_rate": 0.000423265331908281,
      "loss": 4.6437,
      "step": 19909
    },
    {
      "epoch": 2.726650232812928,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0004232244389571976,
      "loss": 4.6418,
      "step": 19910
    },
    {
      "epoch": 2.7267871815940836,
      "grad_norm": 0.359375,
      "learning_rate": 0.00042318354622800965,
      "loss": 4.6285,
      "step": 19911
    },
    {
      "epoch": 2.72692413037524,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00042314265372105585,
      "loss": 4.7087,
      "step": 19912
    },
    {
      "epoch": 2.7270610791563956,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00042310176143667543,
      "loss": 4.6936,
      "step": 19913
    },
    {
      "epoch": 2.7271980279375514,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004230608693752072,
      "loss": 4.6748,
      "step": 19914
    },
    {
      "epoch": 2.727334976718707,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004230199775369899,
      "loss": 4.6517,
      "step": 19915
    },
    {
      "epoch": 2.727471925499863,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004229790859223623,
      "loss": 4.7608,
      "step": 19916
    },
    {
      "epoch": 2.727608874281019,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00042293819453166365,
      "loss": 4.6883,
      "step": 19917
    },
    {
      "epoch": 2.7277458230621745,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004228973033652326,
      "loss": 4.7418,
      "step": 19918
    },
    {
      "epoch": 2.7278827718433307,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004228564124234079,
      "loss": 4.7203,
      "step": 19919
    },
    {
      "epoch": 2.7280197206244865,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004228155217065287,
      "loss": 4.6876,
      "step": 19920
    },
    {
      "epoch": 2.7281566694056423,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00042277463121493384,
      "loss": 4.6234,
      "step": 19921
    },
    {
      "epoch": 2.728293618186798,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004227337409489621,
      "loss": 4.7285,
      "step": 19922
    },
    {
      "epoch": 2.728430566967954,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004226928509089521,
      "loss": 4.754,
      "step": 19923
    },
    {
      "epoch": 2.72856751574911,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0004226519610952432,
      "loss": 4.6884,
      "step": 19924
    },
    {
      "epoch": 2.728704464530266,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00042261107150817393,
      "loss": 4.7674,
      "step": 19925
    },
    {
      "epoch": 2.7288414133114216,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00042257018214808333,
      "loss": 4.7877,
      "step": 19926
    },
    {
      "epoch": 2.7289783620925774,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00042252929301530987,
      "loss": 4.5771,
      "step": 19927
    },
    {
      "epoch": 2.729115310873733,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00042248840411019295,
      "loss": 4.6955,
      "step": 19928
    },
    {
      "epoch": 2.729252259654889,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004224475154330712,
      "loss": 4.6676,
      "step": 19929
    },
    {
      "epoch": 2.7293892084360447,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00042240662698428324,
      "loss": 4.6468,
      "step": 19930
    },
    {
      "epoch": 2.729526157217201,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004223657387641684,
      "loss": 4.6509,
      "step": 19931
    },
    {
      "epoch": 2.7296631059983567,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004223248507730652,
      "loss": 4.6768,
      "step": 19932
    },
    {
      "epoch": 2.7298000547795125,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00042228396301131253,
      "loss": 4.7084,
      "step": 19933
    },
    {
      "epoch": 2.7299370035606683,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004222430754792491,
      "loss": 4.7131,
      "step": 19934
    },
    {
      "epoch": 2.730073952341824,
      "grad_norm": 0.328125,
      "learning_rate": 0.00042220218817721415,
      "loss": 4.7132,
      "step": 19935
    },
    {
      "epoch": 2.7302109011229803,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00042216130110554626,
      "loss": 4.6677,
      "step": 19936
    },
    {
      "epoch": 2.7303478499041356,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004221204142645843,
      "loss": 4.5914,
      "step": 19937
    },
    {
      "epoch": 2.730484798685292,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004220795276546669,
      "loss": 4.7193,
      "step": 19938
    },
    {
      "epoch": 2.7306217474664476,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0004220386412761333,
      "loss": 4.6853,
      "step": 19939
    },
    {
      "epoch": 2.7307586962476034,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00042199775512932215,
      "loss": 4.7497,
      "step": 19940
    },
    {
      "epoch": 2.730895645028759,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004219568692145721,
      "loss": 4.7005,
      "step": 19941
    },
    {
      "epoch": 2.731032593809915,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004219159835322224,
      "loss": 4.6879,
      "step": 19942
    },
    {
      "epoch": 2.731169542591071,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004218750980826115,
      "loss": 4.6496,
      "step": 19943
    },
    {
      "epoch": 2.731306491372227,
      "grad_norm": 0.328125,
      "learning_rate": 0.0004218342128660785,
      "loss": 4.7642,
      "step": 19944
    },
    {
      "epoch": 2.7314434401533827,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004217933278829618,
      "loss": 4.5885,
      "step": 19945
    },
    {
      "epoch": 2.7315803889345385,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00042175244313360073,
      "loss": 4.702,
      "step": 19946
    },
    {
      "epoch": 2.7317173377156942,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004217115586183339,
      "loss": 4.7246,
      "step": 19947
    },
    {
      "epoch": 2.73185428649685,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004216706743375,
      "loss": 4.6531,
      "step": 19948
    },
    {
      "epoch": 2.731991235278006,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00042162979029143806,
      "loss": 4.7307,
      "step": 19949
    },
    {
      "epoch": 2.732128184059162,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0004215889064804869,
      "loss": 4.7572,
      "step": 19950
    },
    {
      "epoch": 2.732265132840318,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00042154802290498514,
      "loss": 4.7099,
      "step": 19951
    },
    {
      "epoch": 2.7324020816214736,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00042150713956527156,
      "loss": 4.6721,
      "step": 19952
    },
    {
      "epoch": 2.7325390304026294,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0004214662564616853,
      "loss": 4.6534,
      "step": 19953
    },
    {
      "epoch": 2.732675979183785,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00042142537359456496,
      "loss": 4.7198,
      "step": 19954
    },
    {
      "epoch": 2.7328129279649414,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00042138449096424936,
      "loss": 4.7245,
      "step": 19955
    },
    {
      "epoch": 2.7329498767460967,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004213436085710771,
      "loss": 4.7455,
      "step": 19956
    },
    {
      "epoch": 2.733086825527253,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00042130272641538745,
      "loss": 4.6369,
      "step": 19957
    },
    {
      "epoch": 2.7332237743084087,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004212618444975188,
      "loss": 4.6635,
      "step": 19958
    },
    {
      "epoch": 2.7333607230895645,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042122096281781,
      "loss": 4.6599,
      "step": 19959
    },
    {
      "epoch": 2.7334976718707202,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0004211800813766001,
      "loss": 4.691,
      "step": 19960
    },
    {
      "epoch": 2.733634620651876,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0004211392001742278,
      "loss": 4.6848,
      "step": 19961
    },
    {
      "epoch": 2.7337715694330322,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042109831921103175,
      "loss": 4.6465,
      "step": 19962
    },
    {
      "epoch": 2.733908518214188,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00042105743848735067,
      "loss": 4.7122,
      "step": 19963
    },
    {
      "epoch": 2.734045466995344,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004210165580035236,
      "loss": 4.6936,
      "step": 19964
    },
    {
      "epoch": 2.7341824157764996,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00042097567775988936,
      "loss": 4.676,
      "step": 19965
    },
    {
      "epoch": 2.7343193645576553,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0004209347977567863,
      "loss": 4.6679,
      "step": 19966
    },
    {
      "epoch": 2.7344563133388116,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00042089391799455373,
      "loss": 4.7067,
      "step": 19967
    },
    {
      "epoch": 2.734593262119967,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00042085303847353014,
      "loss": 4.7292,
      "step": 19968
    },
    {
      "epoch": 2.734730210901123,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00042081215919405443,
      "loss": 4.7278,
      "step": 19969
    },
    {
      "epoch": 2.734867159682279,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004207712801564651,
      "loss": 4.6622,
      "step": 19970
    },
    {
      "epoch": 2.7350041084634347,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00042073040136110133,
      "loss": 4.6912,
      "step": 19971
    },
    {
      "epoch": 2.7351410572445904,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00042068952280830167,
      "loss": 4.6842,
      "step": 19972
    },
    {
      "epoch": 2.7352780060257462,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004206486444984049,
      "loss": 4.6282,
      "step": 19973
    },
    {
      "epoch": 2.7354149548069024,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004206077664317497,
      "loss": 4.6963,
      "step": 19974
    },
    {
      "epoch": 2.735551903588058,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0004205668886086751,
      "loss": 4.7083,
      "step": 19975
    },
    {
      "epoch": 2.735688852369214,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004205260110295197,
      "loss": 4.6856,
      "step": 19976
    },
    {
      "epoch": 2.7358258011503698,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00042048513369462206,
      "loss": 4.7354,
      "step": 19977
    },
    {
      "epoch": 2.7359627499315256,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0004204442566043214,
      "loss": 4.6094,
      "step": 19978
    },
    {
      "epoch": 2.7360996987126813,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004204033797589562,
      "loss": 4.6956,
      "step": 19979
    },
    {
      "epoch": 2.736236647493837,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0004203625031588652,
      "loss": 4.7638,
      "step": 19980
    },
    {
      "epoch": 2.7363735962749933,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00042032162680438707,
      "loss": 4.6888,
      "step": 19981
    },
    {
      "epoch": 2.736510545056149,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0004202807506958608,
      "loss": 4.7132,
      "step": 19982
    },
    {
      "epoch": 2.736647493837305,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0004202398748336251,
      "loss": 4.5585,
      "step": 19983
    },
    {
      "epoch": 2.7367844426184607,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00042019899921801843,
      "loss": 4.6317,
      "step": 19984
    },
    {
      "epoch": 2.7369213913996164,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00042015812384938,
      "loss": 4.7439,
      "step": 19985
    },
    {
      "epoch": 2.7370583401807727,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0004201172487280482,
      "loss": 4.7067,
      "step": 19986
    },
    {
      "epoch": 2.737195288961928,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000420076373854362,
      "loss": 4.6387,
      "step": 19987
    },
    {
      "epoch": 2.737332237743084,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00042003549922865974,
      "loss": 4.7165,
      "step": 19988
    },
    {
      "epoch": 2.73746918652424,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00041999462485128067,
      "loss": 4.6586,
      "step": 19989
    },
    {
      "epoch": 2.7376061353053958,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0004199537507225633,
      "loss": 4.7094,
      "step": 19990
    },
    {
      "epoch": 2.7377430840865515,
      "grad_norm": 0.328125,
      "learning_rate": 0.00041991287684284637,
      "loss": 4.6417,
      "step": 19991
    },
    {
      "epoch": 2.7378800328677073,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00041987200321246833,
      "loss": 4.6705,
      "step": 19992
    },
    {
      "epoch": 2.7380169816488635,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00041983112983176845,
      "loss": 4.6779,
      "step": 19993
    },
    {
      "epoch": 2.7381539304300193,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00041979025670108516,
      "loss": 4.6821,
      "step": 19994
    },
    {
      "epoch": 2.738290879211175,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000419749383820757,
      "loss": 4.7889,
      "step": 19995
    },
    {
      "epoch": 2.738427827992331,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0004197085111911231,
      "loss": 4.7355,
      "step": 19996
    },
    {
      "epoch": 2.7385647767734866,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000419667638812522,
      "loss": 4.7079,
      "step": 19997
    },
    {
      "epoch": 2.7387017255546424,
      "grad_norm": 0.34375,
      "learning_rate": 0.0004196267666852924,
      "loss": 4.7134,
      "step": 19998
    },
    {
      "epoch": 2.738838674335798,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00041958589480977287,
      "loss": 4.6473,
      "step": 19999
    },
    {
      "epoch": 2.7389756231169544,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0004195450231863024,
      "loss": 4.6819,
      "step": 20000
    }
  ],
  "logging_steps": 1,
  "max_steps": 36510,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.772381735332348e+18,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}