{
  "best_metric": 0.4816867411136627,
  "best_model_checkpoint": "paligema_cafe/checkpoint-700",
  "epoch": 49.840255591054316,
  "eval_steps": 100,
  "global_step": 3900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012779552715654952,
      "grad_norm": 0.8142021894454956,
      "learning_rate": 1.99948717948718e-05,
      "loss": 2.9034,
      "step": 1
    },
    {
      "epoch": 0.025559105431309903,
      "grad_norm": 0.8559026718139648,
      "learning_rate": 1.998974358974359e-05,
      "loss": 2.9373,
      "step": 2
    },
    {
      "epoch": 0.038338658146964855,
      "grad_norm": 0.9021930694580078,
      "learning_rate": 1.9984615384615387e-05,
      "loss": 2.925,
      "step": 3
    },
    {
      "epoch": 0.051118210862619806,
      "grad_norm": 0.8621866106987,
      "learning_rate": 1.997948717948718e-05,
      "loss": 3.0668,
      "step": 4
    },
    {
      "epoch": 0.06389776357827476,
      "grad_norm": 1.0313794612884521,
      "learning_rate": 1.9974358974358975e-05,
      "loss": 2.9666,
      "step": 5
    },
    {
      "epoch": 0.07667731629392971,
      "grad_norm": 0.7908412218093872,
      "learning_rate": 1.9969230769230773e-05,
      "loss": 3.0073,
      "step": 6
    },
    {
      "epoch": 0.08945686900958466,
      "grad_norm": 0.8272828459739685,
      "learning_rate": 1.9964102564102567e-05,
      "loss": 2.784,
      "step": 7
    },
    {
      "epoch": 0.10223642172523961,
      "grad_norm": 0.9401512742042542,
      "learning_rate": 1.995897435897436e-05,
      "loss": 2.9083,
      "step": 8
    },
    {
      "epoch": 0.11501597444089456,
      "grad_norm": 0.9094955325126648,
      "learning_rate": 1.9953846153846155e-05,
      "loss": 2.806,
      "step": 9
    },
    {
      "epoch": 0.12779552715654952,
      "grad_norm": 0.9415909647941589,
      "learning_rate": 1.994871794871795e-05,
      "loss": 2.9022,
      "step": 10
    },
    {
      "epoch": 0.14057507987220447,
      "grad_norm": 0.9576209783554077,
      "learning_rate": 1.9943589743589746e-05,
      "loss": 2.8489,
      "step": 11
    },
    {
      "epoch": 0.15335463258785942,
      "grad_norm": 0.9657884240150452,
      "learning_rate": 1.993846153846154e-05,
      "loss": 2.9371,
      "step": 12
    },
    {
      "epoch": 0.16613418530351437,
      "grad_norm": 0.9966011047363281,
      "learning_rate": 1.9933333333333334e-05,
      "loss": 2.8831,
      "step": 13
    },
    {
      "epoch": 0.17891373801916932,
      "grad_norm": 1.2204116582870483,
      "learning_rate": 1.992820512820513e-05,
      "loss": 3.0116,
      "step": 14
    },
    {
      "epoch": 0.19169329073482427,
      "grad_norm": 1.1095342636108398,
      "learning_rate": 1.9923076923076926e-05,
      "loss": 2.6358,
      "step": 15
    },
    {
      "epoch": 0.20447284345047922,
      "grad_norm": 1.0488131046295166,
      "learning_rate": 1.991794871794872e-05,
      "loss": 2.7842,
      "step": 16
    },
    {
      "epoch": 0.21725239616613418,
      "grad_norm": 1.1298291683197021,
      "learning_rate": 1.9912820512820514e-05,
      "loss": 2.6071,
      "step": 17
    },
    {
      "epoch": 0.23003194888178913,
      "grad_norm": 1.20250403881073,
      "learning_rate": 1.990769230769231e-05,
      "loss": 2.2891,
      "step": 18
    },
    {
      "epoch": 0.24281150159744408,
      "grad_norm": 1.0653283596038818,
      "learning_rate": 1.9902564102564102e-05,
      "loss": 2.694,
      "step": 19
    },
    {
      "epoch": 0.25559105431309903,
      "grad_norm": 1.0217392444610596,
      "learning_rate": 1.98974358974359e-05,
      "loss": 2.7254,
      "step": 20
    },
    {
      "epoch": 0.268370607028754,
      "grad_norm": 1.1248751878738403,
      "learning_rate": 1.9892307692307694e-05,
      "loss": 2.8926,
      "step": 21
    },
    {
      "epoch": 0.28115015974440893,
      "grad_norm": 1.0898598432540894,
      "learning_rate": 1.9887179487179488e-05,
      "loss": 2.6239,
      "step": 22
    },
    {
      "epoch": 0.2939297124600639,
      "grad_norm": 1.0956969261169434,
      "learning_rate": 1.9882051282051285e-05,
      "loss": 2.5163,
      "step": 23
    },
    {
      "epoch": 0.30670926517571884,
      "grad_norm": 1.1412856578826904,
      "learning_rate": 1.987692307692308e-05,
      "loss": 2.6186,
      "step": 24
    },
    {
      "epoch": 0.3194888178913738,
      "grad_norm": 1.1162058115005493,
      "learning_rate": 1.9871794871794873e-05,
      "loss": 2.505,
      "step": 25
    },
    {
      "epoch": 0.33226837060702874,
      "grad_norm": 1.301196575164795,
      "learning_rate": 1.9866666666666667e-05,
      "loss": 2.1748,
      "step": 26
    },
    {
      "epoch": 0.3450479233226837,
      "grad_norm": 1.2784515619277954,
      "learning_rate": 1.9861538461538465e-05,
      "loss": 2.3362,
      "step": 27
    },
    {
      "epoch": 0.35782747603833864,
      "grad_norm": 1.2827022075653076,
      "learning_rate": 1.985641025641026e-05,
      "loss": 2.403,
      "step": 28
    },
    {
      "epoch": 0.3706070287539936,
      "grad_norm": 1.1135437488555908,
      "learning_rate": 1.9851282051282053e-05,
      "loss": 2.2757,
      "step": 29
    },
    {
      "epoch": 0.38338658146964855,
      "grad_norm": 1.2342116832733154,
      "learning_rate": 1.9846153846153847e-05,
      "loss": 2.6038,
      "step": 30
    },
    {
      "epoch": 0.3961661341853035,
      "grad_norm": 1.5493252277374268,
      "learning_rate": 1.984102564102564e-05,
      "loss": 2.4121,
      "step": 31
    },
    {
      "epoch": 0.40894568690095845,
      "grad_norm": 1.2742743492126465,
      "learning_rate": 1.983589743589744e-05,
      "loss": 2.2992,
      "step": 32
    },
    {
      "epoch": 0.4217252396166134,
      "grad_norm": 1.1998932361602783,
      "learning_rate": 1.9830769230769232e-05,
      "loss": 2.2267,
      "step": 33
    },
    {
      "epoch": 0.43450479233226835,
      "grad_norm": 1.302520751953125,
      "learning_rate": 1.9825641025641027e-05,
      "loss": 2.308,
      "step": 34
    },
    {
      "epoch": 0.4472843450479233,
      "grad_norm": 1.4223449230194092,
      "learning_rate": 1.9820512820512824e-05,
      "loss": 2.045,
      "step": 35
    },
    {
      "epoch": 0.46006389776357826,
      "grad_norm": 1.3312417268753052,
      "learning_rate": 1.9815384615384618e-05,
      "loss": 2.2282,
      "step": 36
    },
    {
      "epoch": 0.4728434504792332,
      "grad_norm": 1.2069036960601807,
      "learning_rate": 1.9810256410256412e-05,
      "loss": 2.102,
      "step": 37
    },
    {
      "epoch": 0.48562300319488816,
      "grad_norm": 1.3504729270935059,
      "learning_rate": 1.9805128205128206e-05,
      "loss": 2.1236,
      "step": 38
    },
    {
      "epoch": 0.4984025559105431,
      "grad_norm": 1.258859395980835,
      "learning_rate": 1.98e-05,
      "loss": 2.3477,
      "step": 39
    },
    {
      "epoch": 0.5111821086261981,
      "grad_norm": 1.3494940996170044,
      "learning_rate": 1.9794871794871798e-05,
      "loss": 2.2073,
      "step": 40
    },
    {
      "epoch": 0.5239616613418531,
      "grad_norm": 1.217779278755188,
      "learning_rate": 1.978974358974359e-05,
      "loss": 2.2316,
      "step": 41
    },
    {
      "epoch": 0.536741214057508,
      "grad_norm": 1.3344950675964355,
      "learning_rate": 1.9784615384615386e-05,
      "loss": 2.0516,
      "step": 42
    },
    {
      "epoch": 0.549520766773163,
      "grad_norm": 1.2877260446548462,
      "learning_rate": 1.977948717948718e-05,
      "loss": 1.9832,
      "step": 43
    },
    {
      "epoch": 0.5623003194888179,
      "grad_norm": 1.225043773651123,
      "learning_rate": 1.9774358974358977e-05,
      "loss": 2.1805,
      "step": 44
    },
    {
      "epoch": 0.5750798722044729,
      "grad_norm": 1.3049123287200928,
      "learning_rate": 1.976923076923077e-05,
      "loss": 1.7603,
      "step": 45
    },
    {
      "epoch": 0.5878594249201278,
      "grad_norm": 1.2719806432724,
      "learning_rate": 1.9764102564102565e-05,
      "loss": 1.928,
      "step": 46
    },
    {
      "epoch": 0.6006389776357828,
      "grad_norm": 1.4088648557662964,
      "learning_rate": 1.975897435897436e-05,
      "loss": 1.8306,
      "step": 47
    },
    {
      "epoch": 0.6134185303514377,
      "grad_norm": 1.3367751836776733,
      "learning_rate": 1.9753846153846153e-05,
      "loss": 2.0066,
      "step": 48
    },
    {
      "epoch": 0.6261980830670927,
      "grad_norm": 1.2356927394866943,
      "learning_rate": 1.974871794871795e-05,
      "loss": 1.9573,
      "step": 49
    },
    {
      "epoch": 0.6389776357827476,
      "grad_norm": 1.4789074659347534,
      "learning_rate": 1.9743589743589745e-05,
      "loss": 1.9171,
      "step": 50
    },
    {
      "epoch": 0.6517571884984026,
      "grad_norm": 1.335368037223816,
      "learning_rate": 1.973846153846154e-05,
      "loss": 1.9178,
      "step": 51
    },
    {
      "epoch": 0.6645367412140575,
      "grad_norm": 1.4464950561523438,
      "learning_rate": 1.9733333333333336e-05,
      "loss": 1.9905,
      "step": 52
    },
    {
      "epoch": 0.6773162939297125,
      "grad_norm": 1.6089545488357544,
      "learning_rate": 1.972820512820513e-05,
      "loss": 1.8002,
      "step": 53
    },
    {
      "epoch": 0.6900958466453674,
      "grad_norm": 1.397189736366272,
      "learning_rate": 1.9723076923076924e-05,
      "loss": 1.8588,
      "step": 54
    },
    {
      "epoch": 0.7028753993610224,
      "grad_norm": 1.346503734588623,
      "learning_rate": 1.9717948717948722e-05,
      "loss": 1.6663,
      "step": 55
    },
    {
      "epoch": 0.7156549520766773,
      "grad_norm": 1.2886115312576294,
      "learning_rate": 1.9712820512820513e-05,
      "loss": 1.7527,
      "step": 56
    },
    {
      "epoch": 0.7284345047923323,
      "grad_norm": 1.4284838438034058,
      "learning_rate": 1.970769230769231e-05,
      "loss": 1.7273,
      "step": 57
    },
    {
      "epoch": 0.7412140575079872,
      "grad_norm": 1.3520196676254272,
      "learning_rate": 1.9702564102564104e-05,
      "loss": 1.6347,
      "step": 58
    },
    {
      "epoch": 0.7539936102236422,
      "grad_norm": 1.7988076210021973,
      "learning_rate": 1.9697435897435898e-05,
      "loss": 1.5542,
      "step": 59
    },
    {
      "epoch": 0.7667731629392971,
      "grad_norm": 1.6694620847702026,
      "learning_rate": 1.9692307692307696e-05,
      "loss": 1.6283,
      "step": 60
    },
    {
      "epoch": 0.7795527156549521,
      "grad_norm": 1.3663523197174072,
      "learning_rate": 1.968717948717949e-05,
      "loss": 1.73,
      "step": 61
    },
    {
      "epoch": 0.792332268370607,
      "grad_norm": 1.4919695854187012,
      "learning_rate": 1.9682051282051284e-05,
      "loss": 1.6079,
      "step": 62
    },
    {
      "epoch": 0.805111821086262,
      "grad_norm": 1.550233244895935,
      "learning_rate": 1.9676923076923078e-05,
      "loss": 1.4532,
      "step": 63
    },
    {
      "epoch": 0.8178913738019169,
      "grad_norm": 1.4926998615264893,
      "learning_rate": 1.9671794871794875e-05,
      "loss": 1.3253,
      "step": 64
    },
    {
      "epoch": 0.8306709265175719,
      "grad_norm": 1.656699538230896,
      "learning_rate": 1.9666666666666666e-05,
      "loss": 1.4525,
      "step": 65
    },
    {
      "epoch": 0.8434504792332268,
      "grad_norm": 1.774846076965332,
      "learning_rate": 1.9661538461538463e-05,
      "loss": 1.5266,
      "step": 66
    },
    {
      "epoch": 0.8562300319488818,
      "grad_norm": 1.41655695438385,
      "learning_rate": 1.9656410256410257e-05,
      "loss": 1.724,
      "step": 67
    },
    {
      "epoch": 0.8690095846645367,
      "grad_norm": 1.3376843929290771,
      "learning_rate": 1.965128205128205e-05,
      "loss": 1.62,
      "step": 68
    },
    {
      "epoch": 0.8817891373801917,
      "grad_norm": 1.439816951751709,
      "learning_rate": 1.964615384615385e-05,
      "loss": 1.4697,
      "step": 69
    },
    {
      "epoch": 0.8945686900958466,
      "grad_norm": 1.5517547130584717,
      "learning_rate": 1.9641025641025643e-05,
      "loss": 1.4436,
      "step": 70
    },
    {
      "epoch": 0.9073482428115016,
      "grad_norm": 1.4276851415634155,
      "learning_rate": 1.9635897435897437e-05,
      "loss": 1.4793,
      "step": 71
    },
    {
      "epoch": 0.9201277955271565,
      "grad_norm": 1.402268409729004,
      "learning_rate": 1.9630769230769234e-05,
      "loss": 1.3503,
      "step": 72
    },
    {
      "epoch": 0.9329073482428115,
      "grad_norm": 1.4884088039398193,
      "learning_rate": 1.962564102564103e-05,
      "loss": 1.2738,
      "step": 73
    },
    {
      "epoch": 0.9456869009584664,
      "grad_norm": 2.289491653442383,
      "learning_rate": 1.9620512820512822e-05,
      "loss": 1.4683,
      "step": 74
    },
    {
      "epoch": 0.9584664536741214,
      "grad_norm": 1.571138620376587,
      "learning_rate": 1.9615384615384617e-05,
      "loss": 1.3477,
      "step": 75
    },
    {
      "epoch": 0.9712460063897763,
      "grad_norm": 1.9306889772415161,
      "learning_rate": 1.961025641025641e-05,
      "loss": 1.3338,
      "step": 76
    },
    {
      "epoch": 0.9840255591054313,
      "grad_norm": 1.4522299766540527,
      "learning_rate": 1.9605128205128208e-05,
      "loss": 1.2846,
      "step": 77
    },
    {
      "epoch": 0.9968051118210862,
      "grad_norm": 1.491416335105896,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 1.1654,
      "step": 78
    },
    {
      "epoch": 1.0095846645367412,
      "grad_norm": 1.2546143531799316,
      "learning_rate": 1.9594871794871796e-05,
      "loss": 1.0005,
      "step": 79
    },
    {
      "epoch": 1.0223642172523961,
      "grad_norm": 1.8387315273284912,
      "learning_rate": 1.958974358974359e-05,
      "loss": 1.309,
      "step": 80
    },
    {
      "epoch": 1.035143769968051,
      "grad_norm": 2.288738250732422,
      "learning_rate": 1.9584615384615388e-05,
      "loss": 1.25,
      "step": 81
    },
    {
      "epoch": 1.0479233226837061,
      "grad_norm": 1.4508757591247559,
      "learning_rate": 1.957948717948718e-05,
      "loss": 1.3196,
      "step": 82
    },
    {
      "epoch": 1.060702875399361,
      "grad_norm": 1.538571834564209,
      "learning_rate": 1.9574358974358976e-05,
      "loss": 1.0223,
      "step": 83
    },
    {
      "epoch": 1.073482428115016,
      "grad_norm": 1.5000869035720825,
      "learning_rate": 1.9569230769230773e-05,
      "loss": 1.1089,
      "step": 84
    },
    {
      "epoch": 1.0862619808306708,
      "grad_norm": 1.6408222913742065,
      "learning_rate": 1.9564102564102564e-05,
      "loss": 1.0924,
      "step": 85
    },
    {
      "epoch": 1.099041533546326,
      "grad_norm": 1.7912194728851318,
      "learning_rate": 1.955897435897436e-05,
      "loss": 1.0234,
      "step": 86
    },
    {
      "epoch": 1.1118210862619808,
      "grad_norm": 1.4508275985717773,
      "learning_rate": 1.9553846153846155e-05,
      "loss": 1.1314,
      "step": 87
    },
    {
      "epoch": 1.1246006389776357,
      "grad_norm": 1.8075443506240845,
      "learning_rate": 1.954871794871795e-05,
      "loss": 1.007,
      "step": 88
    },
    {
      "epoch": 1.1373801916932909,
      "grad_norm": 1.595278024673462,
      "learning_rate": 1.9543589743589747e-05,
      "loss": 1.1304,
      "step": 89
    },
    {
      "epoch": 1.1501597444089458,
      "grad_norm": 1.43287193775177,
      "learning_rate": 1.953846153846154e-05,
      "loss": 0.9737,
      "step": 90
    },
    {
      "epoch": 1.1629392971246006,
      "grad_norm": 1.7888563871383667,
      "learning_rate": 1.9533333333333335e-05,
      "loss": 1.0763,
      "step": 91
    },
    {
      "epoch": 1.1757188498402555,
      "grad_norm": 1.7963002920150757,
      "learning_rate": 1.952820512820513e-05,
      "loss": 1.0133,
      "step": 92
    },
    {
      "epoch": 1.1884984025559104,
      "grad_norm": 1.6066761016845703,
      "learning_rate": 1.9523076923076923e-05,
      "loss": 0.9496,
      "step": 93
    },
    {
      "epoch": 1.2012779552715656,
      "grad_norm": 1.7649800777435303,
      "learning_rate": 1.951794871794872e-05,
      "loss": 0.9979,
      "step": 94
    },
    {
      "epoch": 1.2140575079872205,
      "grad_norm": 1.8012830018997192,
      "learning_rate": 1.9512820512820515e-05,
      "loss": 1.0609,
      "step": 95
    },
    {
      "epoch": 1.2268370607028753,
      "grad_norm": 1.6527281999588013,
      "learning_rate": 1.950769230769231e-05,
      "loss": 0.8774,
      "step": 96
    },
    {
      "epoch": 1.2396166134185305,
      "grad_norm": 1.5712525844573975,
      "learning_rate": 1.9502564102564103e-05,
      "loss": 0.9831,
      "step": 97
    },
    {
      "epoch": 1.2523961661341854,
      "grad_norm": 2.5428738594055176,
      "learning_rate": 1.94974358974359e-05,
      "loss": 1.3413,
      "step": 98
    },
    {
      "epoch": 1.2651757188498403,
      "grad_norm": 1.9915602207183838,
      "learning_rate": 1.9492307692307694e-05,
      "loss": 0.8742,
      "step": 99
    },
    {
      "epoch": 1.2779552715654952,
      "grad_norm": 2.5296716690063477,
      "learning_rate": 1.9487179487179488e-05,
      "loss": 1.0443,
      "step": 100
    },
    {
      "epoch": 1.2779552715654952,
      "eval_loss": 0.992189347743988,
      "eval_runtime": 181.8595,
      "eval_samples_per_second": 0.863,
      "eval_steps_per_second": 0.11,
      "step": 100
    },
    {
      "epoch": 1.29073482428115,
      "grad_norm": 1.5496196746826172,
      "learning_rate": 1.9482051282051286e-05,
      "loss": 0.9173,
      "step": 101
    },
    {
      "epoch": 1.3035143769968052,
      "grad_norm": 1.8158091306686401,
      "learning_rate": 1.9476923076923076e-05,
      "loss": 1.2035,
      "step": 102
    },
    {
      "epoch": 1.31629392971246,
      "grad_norm": 2.238765239715576,
      "learning_rate": 1.9471794871794874e-05,
      "loss": 1.0669,
      "step": 103
    },
    {
      "epoch": 1.329073482428115,
      "grad_norm": 1.706765055656433,
      "learning_rate": 1.9466666666666668e-05,
      "loss": 0.7441,
      "step": 104
    },
    {
      "epoch": 1.34185303514377,
      "grad_norm": 1.669560432434082,
      "learning_rate": 1.9461538461538462e-05,
      "loss": 0.8697,
      "step": 105
    },
    {
      "epoch": 1.354632587859425,
      "grad_norm": 1.5571810007095337,
      "learning_rate": 1.945641025641026e-05,
      "loss": 0.836,
      "step": 106
    },
    {
      "epoch": 1.3674121405750799,
      "grad_norm": 2.0145392417907715,
      "learning_rate": 1.9451282051282053e-05,
      "loss": 1.0022,
      "step": 107
    },
    {
      "epoch": 1.3801916932907348,
      "grad_norm": 2.5267364978790283,
      "learning_rate": 1.9446153846153847e-05,
      "loss": 1.188,
      "step": 108
    },
    {
      "epoch": 1.3929712460063897,
      "grad_norm": 2.094905138015747,
      "learning_rate": 1.9441025641025645e-05,
      "loss": 0.9611,
      "step": 109
    },
    {
      "epoch": 1.4057507987220448,
      "grad_norm": 1.7281832695007324,
      "learning_rate": 1.943589743589744e-05,
      "loss": 0.7503,
      "step": 110
    },
    {
      "epoch": 1.4185303514376997,
      "grad_norm": 1.7954721450805664,
      "learning_rate": 1.9430769230769233e-05,
      "loss": 0.7309,
      "step": 111
    },
    {
      "epoch": 1.4313099041533546,
      "grad_norm": 1.8233072757720947,
      "learning_rate": 1.9425641025641027e-05,
      "loss": 0.9671,
      "step": 112
    },
    {
      "epoch": 1.4440894568690097,
      "grad_norm": 2.052917003631592,
      "learning_rate": 1.942051282051282e-05,
      "loss": 0.9148,
      "step": 113
    },
    {
      "epoch": 1.4568690095846646,
      "grad_norm": 1.862432599067688,
      "learning_rate": 1.9415384615384615e-05,
      "loss": 0.9417,
      "step": 114
    },
    {
      "epoch": 1.4696485623003195,
      "grad_norm": 1.6614469289779663,
      "learning_rate": 1.9410256410256413e-05,
      "loss": 0.7467,
      "step": 115
    },
    {
      "epoch": 1.4824281150159744,
      "grad_norm": 1.9247630834579468,
      "learning_rate": 1.9405128205128207e-05,
      "loss": 0.92,
      "step": 116
    },
    {
      "epoch": 1.4952076677316293,
      "grad_norm": 1.9821908473968506,
      "learning_rate": 1.94e-05,
      "loss": 0.7205,
      "step": 117
    },
    {
      "epoch": 1.5079872204472844,
      "grad_norm": 2.299802303314209,
      "learning_rate": 1.9394871794871798e-05,
      "loss": 0.7411,
      "step": 118
    },
    {
      "epoch": 1.5207667731629393,
      "grad_norm": 1.7966883182525635,
      "learning_rate": 1.938974358974359e-05,
      "loss": 0.8728,
      "step": 119
    },
    {
      "epoch": 1.5335463258785942,
      "grad_norm": 2.048807382583618,
      "learning_rate": 1.9384615384615386e-05,
      "loss": 0.8071,
      "step": 120
    },
    {
      "epoch": 1.5463258785942493,
      "grad_norm": 1.8083820343017578,
      "learning_rate": 1.9379487179487184e-05,
      "loss": 1.044,
      "step": 121
    },
    {
      "epoch": 1.5591054313099042,
      "grad_norm": 1.692712664604187,
      "learning_rate": 1.9374358974358974e-05,
      "loss": 0.7637,
      "step": 122
    },
    {
      "epoch": 1.571884984025559,
      "grad_norm": 1.9947776794433594,
      "learning_rate": 1.936923076923077e-05,
      "loss": 0.7374,
      "step": 123
    },
    {
      "epoch": 1.5846645367412142,
      "grad_norm": 1.7471729516983032,
      "learning_rate": 1.9364102564102566e-05,
      "loss": 0.72,
      "step": 124
    },
    {
      "epoch": 1.5974440894568689,
      "grad_norm": 1.6320607662200928,
      "learning_rate": 1.935897435897436e-05,
      "loss": 0.8545,
      "step": 125
    },
    {
      "epoch": 1.610223642172524,
      "grad_norm": 1.4828823804855347,
      "learning_rate": 1.9353846153846157e-05,
      "loss": 0.7754,
      "step": 126
    },
    {
      "epoch": 1.623003194888179,
      "grad_norm": 1.6667225360870361,
      "learning_rate": 1.934871794871795e-05,
      "loss": 0.7953,
      "step": 127
    },
    {
      "epoch": 1.6357827476038338,
      "grad_norm": 1.9879682064056396,
      "learning_rate": 1.9343589743589745e-05,
      "loss": 0.8459,
      "step": 128
    },
    {
      "epoch": 1.648562300319489,
      "grad_norm": 1.8809915781021118,
      "learning_rate": 1.933846153846154e-05,
      "loss": 0.7419,
      "step": 129
    },
    {
      "epoch": 1.6613418530351438,
      "grad_norm": 2.041389226913452,
      "learning_rate": 1.9333333333333333e-05,
      "loss": 0.8847,
      "step": 130
    },
    {
      "epoch": 1.6741214057507987,
      "grad_norm": 2.201057195663452,
      "learning_rate": 1.932820512820513e-05,
      "loss": 0.5984,
      "step": 131
    },
    {
      "epoch": 1.6869009584664538,
      "grad_norm": 2.1080873012542725,
      "learning_rate": 1.9323076923076925e-05,
      "loss": 0.9419,
      "step": 132
    },
    {
      "epoch": 1.6996805111821085,
      "grad_norm": 2.061424493789673,
      "learning_rate": 1.931794871794872e-05,
      "loss": 0.7766,
      "step": 133
    },
    {
      "epoch": 1.7124600638977636,
      "grad_norm": 1.7074888944625854,
      "learning_rate": 1.9312820512820513e-05,
      "loss": 0.6979,
      "step": 134
    },
    {
      "epoch": 1.7252396166134185,
      "grad_norm": 1.6944959163665771,
      "learning_rate": 1.930769230769231e-05,
      "loss": 0.8957,
      "step": 135
    },
    {
      "epoch": 1.7380191693290734,
      "grad_norm": 1.9344595670700073,
      "learning_rate": 1.9302564102564105e-05,
      "loss": 0.7251,
      "step": 136
    },
    {
      "epoch": 1.7507987220447285,
      "grad_norm": 2.4287431240081787,
      "learning_rate": 1.92974358974359e-05,
      "loss": 0.641,
      "step": 137
    },
    {
      "epoch": 1.7635782747603834,
      "grad_norm": 2.3612871170043945,
      "learning_rate": 1.9292307692307696e-05,
      "loss": 0.7049,
      "step": 138
    },
    {
      "epoch": 1.7763578274760383,
      "grad_norm": 2.3364453315734863,
      "learning_rate": 1.9287179487179487e-05,
      "loss": 0.7515,
      "step": 139
    },
    {
      "epoch": 1.7891373801916934,
      "grad_norm": 1.9627810716629028,
      "learning_rate": 1.9282051282051284e-05,
      "loss": 0.6797,
      "step": 140
    },
    {
      "epoch": 1.8019169329073481,
      "grad_norm": 2.451805830001831,
      "learning_rate": 1.9276923076923078e-05,
      "loss": 0.7781,
      "step": 141
    },
    {
      "epoch": 1.8146964856230032,
      "grad_norm": 1.8913015127182007,
      "learning_rate": 1.9271794871794872e-05,
      "loss": 0.8206,
      "step": 142
    },
    {
      "epoch": 1.8274760383386581,
      "grad_norm": 1.6889513731002808,
      "learning_rate": 1.926666666666667e-05,
      "loss": 0.5143,
      "step": 143
    },
    {
      "epoch": 1.840255591054313,
      "grad_norm": 2.1671924591064453,
      "learning_rate": 1.9261538461538464e-05,
      "loss": 1.0332,
      "step": 144
    },
    {
      "epoch": 1.8530351437699681,
      "grad_norm": 1.8647410869598389,
      "learning_rate": 1.9256410256410258e-05,
      "loss": 0.687,
      "step": 145
    },
    {
      "epoch": 1.865814696485623,
      "grad_norm": 1.7664998769760132,
      "learning_rate": 1.9251282051282052e-05,
      "loss": 0.8675,
      "step": 146
    },
    {
      "epoch": 1.878594249201278,
      "grad_norm": 2.581796407699585,
      "learning_rate": 1.924615384615385e-05,
      "loss": 1.0433,
      "step": 147
    },
    {
      "epoch": 1.891373801916933,
      "grad_norm": 2.2406113147735596,
      "learning_rate": 1.9241025641025643e-05,
      "loss": 1.2425,
      "step": 148
    },
    {
      "epoch": 1.9041533546325877,
      "grad_norm": 2.4278011322021484,
      "learning_rate": 1.9235897435897437e-05,
      "loss": 0.5769,
      "step": 149
    },
    {
      "epoch": 1.9169329073482428,
      "grad_norm": 2.2902305126190186,
      "learning_rate": 1.923076923076923e-05,
      "loss": 0.6431,
      "step": 150
    },
    {
      "epoch": 1.9297124600638977,
      "grad_norm": 1.8957128524780273,
      "learning_rate": 1.9225641025641025e-05,
      "loss": 0.8498,
      "step": 151
    },
    {
      "epoch": 1.9424920127795526,
      "grad_norm": 1.6397674083709717,
      "learning_rate": 1.9220512820512823e-05,
      "loss": 0.8161,
      "step": 152
    },
    {
      "epoch": 1.9552715654952078,
      "grad_norm": 1.9125725030899048,
      "learning_rate": 1.9215384615384617e-05,
      "loss": 0.5361,
      "step": 153
    },
    {
      "epoch": 1.9680511182108626,
      "grad_norm": 1.9646354913711548,
      "learning_rate": 1.921025641025641e-05,
      "loss": 0.8507,
      "step": 154
    },
    {
      "epoch": 1.9808306709265175,
      "grad_norm": 1.851611852645874,
      "learning_rate": 1.920512820512821e-05,
      "loss": 0.6281,
      "step": 155
    },
    {
      "epoch": 1.9936102236421727,
      "grad_norm": 1.6641134023666382,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.6391,
      "step": 156
    },
    {
      "epoch": 2.0063897763578273,
      "grad_norm": 2.319157361984253,
      "learning_rate": 1.9194871794871797e-05,
      "loss": 1.1003,
      "step": 157
    },
    {
      "epoch": 2.0191693290734825,
      "grad_norm": 2.1189239025115967,
      "learning_rate": 1.9189743589743594e-05,
      "loss": 0.7064,
      "step": 158
    },
    {
      "epoch": 2.0319488817891376,
      "grad_norm": 2.2989776134490967,
      "learning_rate": 1.9184615384615385e-05,
      "loss": 0.7314,
      "step": 159
    },
    {
      "epoch": 2.0447284345047922,
      "grad_norm": 1.766953468322754,
      "learning_rate": 1.9179487179487182e-05,
      "loss": 0.5125,
      "step": 160
    },
    {
      "epoch": 2.0575079872204474,
      "grad_norm": 2.2920029163360596,
      "learning_rate": 1.9174358974358976e-05,
      "loss": 0.6074,
      "step": 161
    },
    {
      "epoch": 2.070287539936102,
      "grad_norm": 1.7585970163345337,
      "learning_rate": 1.916923076923077e-05,
      "loss": 0.8434,
      "step": 162
    },
    {
      "epoch": 2.083067092651757,
      "grad_norm": 1.5543984174728394,
      "learning_rate": 1.9164102564102564e-05,
      "loss": 0.5749,
      "step": 163
    },
    {
      "epoch": 2.0958466453674123,
      "grad_norm": 2.3347678184509277,
      "learning_rate": 1.9158974358974362e-05,
      "loss": 0.7675,
      "step": 164
    },
    {
      "epoch": 2.108626198083067,
      "grad_norm": 1.7411854267120361,
      "learning_rate": 1.9153846153846156e-05,
      "loss": 0.5865,
      "step": 165
    },
    {
      "epoch": 2.121405750798722,
      "grad_norm": 3.4430465698242188,
      "learning_rate": 1.914871794871795e-05,
      "loss": 0.7566,
      "step": 166
    },
    {
      "epoch": 2.134185303514377,
      "grad_norm": 1.8736684322357178,
      "learning_rate": 1.9143589743589744e-05,
      "loss": 0.6378,
      "step": 167
    },
    {
      "epoch": 2.146964856230032,
      "grad_norm": 1.6040446758270264,
      "learning_rate": 1.9138461538461538e-05,
      "loss": 0.5306,
      "step": 168
    },
    {
      "epoch": 2.159744408945687,
      "grad_norm": 2.1889142990112305,
      "learning_rate": 1.9133333333333335e-05,
      "loss": 0.5335,
      "step": 169
    },
    {
      "epoch": 2.1725239616613417,
      "grad_norm": 1.7461152076721191,
      "learning_rate": 1.912820512820513e-05,
      "loss": 0.5002,
      "step": 170
    },
    {
      "epoch": 2.1853035143769968,
      "grad_norm": 1.7458964586257935,
      "learning_rate": 1.9123076923076923e-05,
      "loss": 0.6409,
      "step": 171
    },
    {
      "epoch": 2.198083067092652,
      "grad_norm": 2.153862953186035,
      "learning_rate": 1.911794871794872e-05,
      "loss": 0.5139,
      "step": 172
    },
    {
      "epoch": 2.2108626198083066,
      "grad_norm": 1.9740614891052246,
      "learning_rate": 1.9112820512820515e-05,
      "loss": 0.7204,
      "step": 173
    },
    {
      "epoch": 2.2236421725239617,
      "grad_norm": 2.859555721282959,
      "learning_rate": 1.910769230769231e-05,
      "loss": 0.8671,
      "step": 174
    },
    {
      "epoch": 2.236421725239617,
      "grad_norm": 2.5436668395996094,
      "learning_rate": 1.9102564102564106e-05,
      "loss": 0.5918,
      "step": 175
    },
    {
      "epoch": 2.2492012779552715,
      "grad_norm": 2.3637661933898926,
      "learning_rate": 1.9097435897435897e-05,
      "loss": 0.6223,
      "step": 176
    },
    {
      "epoch": 2.2619808306709266,
      "grad_norm": 2.048809766769409,
      "learning_rate": 1.9092307692307695e-05,
      "loss": 0.5968,
      "step": 177
    },
    {
      "epoch": 2.2747603833865817,
      "grad_norm": 2.409816265106201,
      "learning_rate": 1.908717948717949e-05,
      "loss": 0.6192,
      "step": 178
    },
    {
      "epoch": 2.2875399361022364,
      "grad_norm": 2.0535144805908203,
      "learning_rate": 1.9082051282051283e-05,
      "loss": 0.5019,
      "step": 179
    },
    {
      "epoch": 2.3003194888178915,
      "grad_norm": 2.1206672191619873,
      "learning_rate": 1.907692307692308e-05,
      "loss": 0.8196,
      "step": 180
    },
    {
      "epoch": 2.313099041533546,
      "grad_norm": 1.8516058921813965,
      "learning_rate": 1.9071794871794874e-05,
      "loss": 0.6362,
      "step": 181
    },
    {
      "epoch": 2.3258785942492013,
      "grad_norm": 2.1362321376800537,
      "learning_rate": 1.9066666666666668e-05,
      "loss": 0.5798,
      "step": 182
    },
    {
      "epoch": 2.3386581469648564,
      "grad_norm": 1.775614619255066,
      "learning_rate": 1.9061538461538462e-05,
      "loss": 0.6961,
      "step": 183
    },
    {
      "epoch": 2.351437699680511,
      "grad_norm": 2.2190535068511963,
      "learning_rate": 1.905641025641026e-05,
      "loss": 0.7895,
      "step": 184
    },
    {
      "epoch": 2.364217252396166,
      "grad_norm": 1.8019541501998901,
      "learning_rate": 1.905128205128205e-05,
      "loss": 0.4996,
      "step": 185
    },
    {
      "epoch": 2.376996805111821,
      "grad_norm": 3.500237226486206,
      "learning_rate": 1.9046153846153848e-05,
      "loss": 0.8526,
      "step": 186
    },
    {
      "epoch": 2.389776357827476,
      "grad_norm": 2.187237024307251,
      "learning_rate": 1.9041025641025642e-05,
      "loss": 0.6901,
      "step": 187
    },
    {
      "epoch": 2.402555910543131,
      "grad_norm": 1.8445521593093872,
      "learning_rate": 1.9035897435897436e-05,
      "loss": 0.6152,
      "step": 188
    },
    {
      "epoch": 2.415335463258786,
      "grad_norm": 2.152160882949829,
      "learning_rate": 1.9030769230769233e-05,
      "loss": 0.6574,
      "step": 189
    },
    {
      "epoch": 2.428115015974441,
      "grad_norm": 2.056286573410034,
      "learning_rate": 1.9025641025641027e-05,
      "loss": 0.5641,
      "step": 190
    },
    {
      "epoch": 2.440894568690096,
      "grad_norm": 1.9102013111114502,
      "learning_rate": 1.902051282051282e-05,
      "loss": 0.666,
      "step": 191
    },
    {
      "epoch": 2.4536741214057507,
      "grad_norm": 2.1371514797210693,
      "learning_rate": 1.901538461538462e-05,
      "loss": 0.768,
      "step": 192
    },
    {
      "epoch": 2.466453674121406,
      "grad_norm": 2.719141721725464,
      "learning_rate": 1.9010256410256413e-05,
      "loss": 0.7495,
      "step": 193
    },
    {
      "epoch": 2.479233226837061,
      "grad_norm": 2.1122324466705322,
      "learning_rate": 1.9005128205128207e-05,
      "loss": 0.5684,
      "step": 194
    },
    {
      "epoch": 2.4920127795527156,
      "grad_norm": 1.485161542892456,
      "learning_rate": 1.9e-05,
      "loss": 0.5679,
      "step": 195
    },
    {
      "epoch": 2.5047923322683707,
      "grad_norm": 1.7819613218307495,
      "learning_rate": 1.8994871794871795e-05,
      "loss": 0.4931,
      "step": 196
    },
    {
      "epoch": 2.5175718849840254,
      "grad_norm": 1.859328269958496,
      "learning_rate": 1.8989743589743593e-05,
      "loss": 0.6118,
      "step": 197
    },
    {
      "epoch": 2.5303514376996805,
      "grad_norm": 1.929071068763733,
      "learning_rate": 1.8984615384615387e-05,
      "loss": 0.6645,
      "step": 198
    },
    {
      "epoch": 2.543130990415335,
      "grad_norm": 2.3670995235443115,
      "learning_rate": 1.897948717948718e-05,
      "loss": 0.696,
      "step": 199
    },
    {
      "epoch": 2.5559105431309903,
      "grad_norm": 2.6638221740722656,
      "learning_rate": 1.8974358974358975e-05,
      "loss": 0.8578,
      "step": 200
    },
    {
      "epoch": 2.5559105431309903,
      "eval_loss": 0.6453202366828918,
      "eval_runtime": 181.0992,
      "eval_samples_per_second": 0.867,
      "eval_steps_per_second": 0.11,
      "step": 200
    },
    {
      "epoch": 2.5686900958466454,
      "grad_norm": 2.1391708850860596,
      "learning_rate": 1.8969230769230772e-05,
      "loss": 0.6726,
      "step": 201
    },
    {
      "epoch": 2.5814696485623,
      "grad_norm": 2.205821990966797,
      "learning_rate": 1.8964102564102566e-05,
      "loss": 0.6483,
      "step": 202
    },
    {
      "epoch": 2.594249201277955,
      "grad_norm": 2.030141592025757,
      "learning_rate": 1.895897435897436e-05,
      "loss": 0.7137,
      "step": 203
    },
    {
      "epoch": 2.6070287539936103,
      "grad_norm": 1.7788156270980835,
      "learning_rate": 1.8953846153846158e-05,
      "loss": 0.7249,
      "step": 204
    },
    {
      "epoch": 2.619808306709265,
      "grad_norm": 1.799312710762024,
      "learning_rate": 1.894871794871795e-05,
      "loss": 0.6609,
      "step": 205
    },
    {
      "epoch": 2.63258785942492,
      "grad_norm": 1.8184287548065186,
      "learning_rate": 1.8943589743589746e-05,
      "loss": 0.5924,
      "step": 206
    },
    {
      "epoch": 2.6453674121405752,
      "grad_norm": 1.925370216369629,
      "learning_rate": 1.893846153846154e-05,
      "loss": 0.6466,
      "step": 207
    },
    {
      "epoch": 2.65814696485623,
      "grad_norm": 1.9518979787826538,
      "learning_rate": 1.8933333333333334e-05,
      "loss": 0.5184,
      "step": 208
    },
    {
      "epoch": 2.670926517571885,
      "grad_norm": 1.5550258159637451,
      "learning_rate": 1.892820512820513e-05,
      "loss": 0.5325,
      "step": 209
    },
    {
      "epoch": 2.68370607028754,
      "grad_norm": 1.674920678138733,
      "learning_rate": 1.8923076923076925e-05,
      "loss": 0.4642,
      "step": 210
    },
    {
      "epoch": 2.696485623003195,
      "grad_norm": 2.142885684967041,
      "learning_rate": 1.891794871794872e-05,
      "loss": 0.7174,
      "step": 211
    },
    {
      "epoch": 2.70926517571885,
      "grad_norm": 2.322446346282959,
      "learning_rate": 1.8912820512820513e-05,
      "loss": 0.7566,
      "step": 212
    },
    {
      "epoch": 2.722044728434505,
      "grad_norm": 1.7620545625686646,
      "learning_rate": 1.8907692307692308e-05,
      "loss": 0.5529,
      "step": 213
    },
    {
      "epoch": 2.7348242811501597,
      "grad_norm": 1.606350064277649,
      "learning_rate": 1.8902564102564105e-05,
      "loss": 0.469,
      "step": 214
    },
    {
      "epoch": 2.747603833865815,
      "grad_norm": 2.2989253997802734,
      "learning_rate": 1.88974358974359e-05,
      "loss": 0.7478,
      "step": 215
    },
    {
      "epoch": 2.7603833865814695,
      "grad_norm": 2.260450601577759,
      "learning_rate": 1.8892307692307693e-05,
      "loss": 0.5506,
      "step": 216
    },
    {
      "epoch": 2.7731629392971247,
      "grad_norm": 2.017993211746216,
      "learning_rate": 1.8887179487179487e-05,
      "loss": 0.4649,
      "step": 217
    },
    {
      "epoch": 2.7859424920127793,
      "grad_norm": 1.8710030317306519,
      "learning_rate": 1.8882051282051285e-05,
      "loss": 0.6181,
      "step": 218
    },
    {
      "epoch": 2.7987220447284344,
      "grad_norm": 2.171832799911499,
      "learning_rate": 1.887692307692308e-05,
      "loss": 0.566,
      "step": 219
    },
    {
      "epoch": 2.8115015974440896,
      "grad_norm": 2.444894552230835,
      "learning_rate": 1.8871794871794873e-05,
      "loss": 0.6019,
      "step": 220
    },
    {
      "epoch": 2.8242811501597442,
      "grad_norm": 2.516892194747925,
      "learning_rate": 1.886666666666667e-05,
      "loss": 0.609,
      "step": 221
    },
    {
      "epoch": 2.8370607028753994,
      "grad_norm": 1.8376013040542603,
      "learning_rate": 1.886153846153846e-05,
      "loss": 0.4318,
      "step": 222
    },
    {
      "epoch": 2.8498402555910545,
      "grad_norm": 2.615744113922119,
      "learning_rate": 1.8856410256410258e-05,
      "loss": 0.6561,
      "step": 223
    },
    {
      "epoch": 2.862619808306709,
      "grad_norm": 2.35602068901062,
      "learning_rate": 1.8851282051282052e-05,
      "loss": 0.7015,
      "step": 224
    },
    {
      "epoch": 2.8753993610223643,
      "grad_norm": 1.974861979484558,
      "learning_rate": 1.8846153846153846e-05,
      "loss": 0.6246,
      "step": 225
    },
    {
      "epoch": 2.8881789137380194,
      "grad_norm": 1.9995675086975098,
      "learning_rate": 1.8841025641025644e-05,
      "loss": 0.5532,
      "step": 226
    },
    {
      "epoch": 2.900958466453674,
      "grad_norm": 2.0074105262756348,
      "learning_rate": 1.8835897435897438e-05,
      "loss": 0.5717,
      "step": 227
    },
    {
      "epoch": 2.913738019169329,
      "grad_norm": 2.0054516792297363,
      "learning_rate": 1.8830769230769232e-05,
      "loss": 0.6862,
      "step": 228
    },
    {
      "epoch": 2.9265175718849843,
      "grad_norm": 2.4561843872070312,
      "learning_rate": 1.882564102564103e-05,
      "loss": 0.6273,
      "step": 229
    },
    {
      "epoch": 2.939297124600639,
      "grad_norm": 2.190880537033081,
      "learning_rate": 1.8820512820512823e-05,
      "loss": 0.618,
      "step": 230
    },
    {
      "epoch": 2.952076677316294,
      "grad_norm": 2.0371289253234863,
      "learning_rate": 1.8815384615384617e-05,
      "loss": 0.4063,
      "step": 231
    },
    {
      "epoch": 2.9648562300319488,
      "grad_norm": 2.6498465538024902,
      "learning_rate": 1.881025641025641e-05,
      "loss": 0.6254,
      "step": 232
    },
    {
      "epoch": 2.977635782747604,
      "grad_norm": 1.9762572050094604,
      "learning_rate": 1.8805128205128206e-05,
      "loss": 0.6503,
      "step": 233
    },
    {
      "epoch": 2.9904153354632586,
      "grad_norm": 1.9959864616394043,
      "learning_rate": 1.88e-05,
      "loss": 0.5692,
      "step": 234
    },
    {
      "epoch": 3.0031948881789137,
      "grad_norm": 2.349118232727051,
      "learning_rate": 1.8794871794871797e-05,
      "loss": 0.7357,
      "step": 235
    },
    {
      "epoch": 3.015974440894569,
      "grad_norm": 2.024961471557617,
      "learning_rate": 1.878974358974359e-05,
      "loss": 0.5339,
      "step": 236
    },
    {
      "epoch": 3.0287539936102235,
      "grad_norm": 2.3128573894500732,
      "learning_rate": 1.8784615384615385e-05,
      "loss": 0.6399,
      "step": 237
    },
    {
      "epoch": 3.0415335463258786,
      "grad_norm": 2.0917768478393555,
      "learning_rate": 1.8779487179487183e-05,
      "loss": 0.6905,
      "step": 238
    },
    {
      "epoch": 3.0543130990415337,
      "grad_norm": 2.443096876144409,
      "learning_rate": 1.8774358974358977e-05,
      "loss": 0.58,
      "step": 239
    },
    {
      "epoch": 3.0670926517571884,
      "grad_norm": 2.049499988555908,
      "learning_rate": 1.876923076923077e-05,
      "loss": 0.5741,
      "step": 240
    },
    {
      "epoch": 3.0798722044728435,
      "grad_norm": 2.377434730529785,
      "learning_rate": 1.8764102564102568e-05,
      "loss": 0.6907,
      "step": 241
    },
    {
      "epoch": 3.0926517571884986,
      "grad_norm": 2.2461509704589844,
      "learning_rate": 1.875897435897436e-05,
      "loss": 0.6526,
      "step": 242
    },
    {
      "epoch": 3.1054313099041533,
      "grad_norm": 2.7188053131103516,
      "learning_rate": 1.8753846153846156e-05,
      "loss": 0.603,
      "step": 243
    },
    {
      "epoch": 3.1182108626198084,
      "grad_norm": 2.164670705795288,
      "learning_rate": 1.874871794871795e-05,
      "loss": 0.5971,
      "step": 244
    },
    {
      "epoch": 3.130990415335463,
      "grad_norm": 2.7351036071777344,
      "learning_rate": 1.8743589743589744e-05,
      "loss": 0.5675,
      "step": 245
    },
    {
      "epoch": 3.143769968051118,
      "grad_norm": 1.7953331470489502,
      "learning_rate": 1.8738461538461542e-05,
      "loss": 0.4819,
      "step": 246
    },
    {
      "epoch": 3.1565495207667733,
      "grad_norm": 2.31479811668396,
      "learning_rate": 1.8733333333333336e-05,
      "loss": 0.7826,
      "step": 247
    },
    {
      "epoch": 3.169329073482428,
      "grad_norm": 1.7311149835586548,
      "learning_rate": 1.872820512820513e-05,
      "loss": 0.6296,
      "step": 248
    },
    {
      "epoch": 3.182108626198083,
      "grad_norm": 1.964176893234253,
      "learning_rate": 1.8723076923076924e-05,
      "loss": 0.5487,
      "step": 249
    },
    {
      "epoch": 3.194888178913738,
      "grad_norm": 2.2353405952453613,
      "learning_rate": 1.8717948717948718e-05,
      "loss": 0.7026,
      "step": 250
    },
    {
      "epoch": 3.207667731629393,
      "grad_norm": 2.1692943572998047,
      "learning_rate": 1.8712820512820515e-05,
      "loss": 0.4247,
      "step": 251
    },
    {
      "epoch": 3.220447284345048,
      "grad_norm": 1.894005298614502,
      "learning_rate": 1.870769230769231e-05,
      "loss": 0.577,
      "step": 252
    },
    {
      "epoch": 3.2332268370607027,
      "grad_norm": 2.4834699630737305,
      "learning_rate": 1.8702564102564104e-05,
      "loss": 0.7699,
      "step": 253
    },
    {
      "epoch": 3.246006389776358,
      "grad_norm": 2.7217936515808105,
      "learning_rate": 1.8697435897435898e-05,
      "loss": 0.5841,
      "step": 254
    },
    {
      "epoch": 3.258785942492013,
      "grad_norm": 2.007357120513916,
      "learning_rate": 1.8692307692307695e-05,
      "loss": 0.4561,
      "step": 255
    },
    {
      "epoch": 3.2715654952076676,
      "grad_norm": 2.752572536468506,
      "learning_rate": 1.868717948717949e-05,
      "loss": 0.7493,
      "step": 256
    },
    {
      "epoch": 3.2843450479233227,
      "grad_norm": 1.9526993036270142,
      "learning_rate": 1.8682051282051283e-05,
      "loss": 0.62,
      "step": 257
    },
    {
      "epoch": 3.297124600638978,
      "grad_norm": 2.3786776065826416,
      "learning_rate": 1.867692307692308e-05,
      "loss": 0.6224,
      "step": 258
    },
    {
      "epoch": 3.3099041533546325,
      "grad_norm": 2.043469190597534,
      "learning_rate": 1.867179487179487e-05,
      "loss": 0.5053,
      "step": 259
    },
    {
      "epoch": 3.3226837060702876,
      "grad_norm": 1.7549484968185425,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.4134,
      "step": 260
    },
    {
      "epoch": 3.3354632587859427,
      "grad_norm": 2.2081825733184814,
      "learning_rate": 1.8661538461538463e-05,
      "loss": 0.8591,
      "step": 261
    },
    {
      "epoch": 3.3482428115015974,
      "grad_norm": 1.8951700925827026,
      "learning_rate": 1.8656410256410257e-05,
      "loss": 0.6255,
      "step": 262
    },
    {
      "epoch": 3.3610223642172525,
      "grad_norm": 2.757758140563965,
      "learning_rate": 1.8651282051282054e-05,
      "loss": 0.4654,
      "step": 263
    },
    {
      "epoch": 3.373801916932907,
      "grad_norm": 2.7067980766296387,
      "learning_rate": 1.8646153846153848e-05,
      "loss": 0.6433,
      "step": 264
    },
    {
      "epoch": 3.3865814696485623,
      "grad_norm": 2.0272269248962402,
      "learning_rate": 1.8641025641025642e-05,
      "loss": 0.4873,
      "step": 265
    },
    {
      "epoch": 3.3993610223642174,
      "grad_norm": 2.0535693168640137,
      "learning_rate": 1.8635897435897436e-05,
      "loss": 0.6818,
      "step": 266
    },
    {
      "epoch": 3.412140575079872,
      "grad_norm": 2.127767562866211,
      "learning_rate": 1.8630769230769234e-05,
      "loss": 0.4724,
      "step": 267
    },
    {
      "epoch": 3.4249201277955272,
      "grad_norm": 1.7211772203445435,
      "learning_rate": 1.8625641025641028e-05,
      "loss": 0.4135,
      "step": 268
    },
    {
      "epoch": 3.437699680511182,
      "grad_norm": 1.9719951152801514,
      "learning_rate": 1.8620512820512822e-05,
      "loss": 0.4409,
      "step": 269
    },
    {
      "epoch": 3.450479233226837,
      "grad_norm": 1.8113923072814941,
      "learning_rate": 1.8615384615384616e-05,
      "loss": 0.6051,
      "step": 270
    },
    {
      "epoch": 3.463258785942492,
      "grad_norm": 2.094407796859741,
      "learning_rate": 1.861025641025641e-05,
      "loss": 0.4492,
      "step": 271
    },
    {
      "epoch": 3.476038338658147,
      "grad_norm": 2.2038345336914062,
      "learning_rate": 1.8605128205128207e-05,
      "loss": 0.6206,
      "step": 272
    },
    {
      "epoch": 3.488817891373802,
      "grad_norm": 2.0385520458221436,
      "learning_rate": 1.86e-05,
      "loss": 0.4314,
      "step": 273
    },
    {
      "epoch": 3.501597444089457,
      "grad_norm": 2.0348079204559326,
      "learning_rate": 1.8594871794871796e-05,
      "loss": 0.5783,
      "step": 274
    },
    {
      "epoch": 3.5143769968051117,
      "grad_norm": 1.742039680480957,
      "learning_rate": 1.8589743589743593e-05,
      "loss": 0.4207,
      "step": 275
    },
    {
      "epoch": 3.527156549520767,
      "grad_norm": 2.7231643199920654,
      "learning_rate": 1.8584615384615387e-05,
      "loss": 0.703,
      "step": 276
    },
    {
      "epoch": 3.539936102236422,
      "grad_norm": 2.291921615600586,
      "learning_rate": 1.857948717948718e-05,
      "loss": 0.5661,
      "step": 277
    },
    {
      "epoch": 3.5527156549520766,
      "grad_norm": 1.8833556175231934,
      "learning_rate": 1.857435897435898e-05,
      "loss": 0.4881,
      "step": 278
    },
    {
      "epoch": 3.5654952076677318,
      "grad_norm": 2.2225875854492188,
      "learning_rate": 1.856923076923077e-05,
      "loss": 0.4945,
      "step": 279
    },
    {
      "epoch": 3.5782747603833864,
      "grad_norm": 2.337555408477783,
      "learning_rate": 1.8564102564102567e-05,
      "loss": 0.6395,
      "step": 280
    },
    {
      "epoch": 3.5910543130990416,
      "grad_norm": 2.4982399940490723,
      "learning_rate": 1.855897435897436e-05,
      "loss": 0.6697,
      "step": 281
    },
    {
      "epoch": 3.6038338658146962,
      "grad_norm": 3.495062828063965,
      "learning_rate": 1.8553846153846155e-05,
      "loss": 0.5231,
      "step": 282
    },
    {
      "epoch": 3.6166134185303513,
      "grad_norm": 1.9106608629226685,
      "learning_rate": 1.854871794871795e-05,
      "loss": 0.4807,
      "step": 283
    },
    {
      "epoch": 3.6293929712460065,
      "grad_norm": 2.2010486125946045,
      "learning_rate": 1.8543589743589746e-05,
      "loss": 0.5052,
      "step": 284
    },
    {
      "epoch": 3.642172523961661,
      "grad_norm": 2.2390029430389404,
      "learning_rate": 1.853846153846154e-05,
      "loss": 0.4321,
      "step": 285
    },
    {
      "epoch": 3.6549520766773163,
      "grad_norm": 2.0129148960113525,
      "learning_rate": 1.8533333333333334e-05,
      "loss": 0.4684,
      "step": 286
    },
    {
      "epoch": 3.6677316293929714,
      "grad_norm": 2.1632564067840576,
      "learning_rate": 1.852820512820513e-05,
      "loss": 0.5153,
      "step": 287
    },
    {
      "epoch": 3.680511182108626,
      "grad_norm": 2.2095839977264404,
      "learning_rate": 1.8523076923076922e-05,
      "loss": 0.6905,
      "step": 288
    },
    {
      "epoch": 3.693290734824281,
      "grad_norm": 2.3573625087738037,
      "learning_rate": 1.851794871794872e-05,
      "loss": 0.7279,
      "step": 289
    },
    {
      "epoch": 3.7060702875399363,
      "grad_norm": 1.7171701192855835,
      "learning_rate": 1.8512820512820514e-05,
      "loss": 0.5147,
      "step": 290
    },
    {
      "epoch": 3.718849840255591,
      "grad_norm": 1.9186244010925293,
      "learning_rate": 1.8507692307692308e-05,
      "loss": 0.475,
      "step": 291
    },
    {
      "epoch": 3.731629392971246,
      "grad_norm": 2.364483118057251,
      "learning_rate": 1.8502564102564105e-05,
      "loss": 0.5201,
      "step": 292
    },
    {
      "epoch": 3.744408945686901,
      "grad_norm": 2.2425689697265625,
      "learning_rate": 1.84974358974359e-05,
      "loss": 0.5556,
      "step": 293
    },
    {
      "epoch": 3.757188498402556,
      "grad_norm": 2.286992311477661,
      "learning_rate": 1.8492307692307694e-05,
      "loss": 0.5985,
      "step": 294
    },
    {
      "epoch": 3.769968051118211,
      "grad_norm": 1.652174949645996,
      "learning_rate": 1.848717948717949e-05,
      "loss": 0.3692,
      "step": 295
    },
    {
      "epoch": 3.7827476038338657,
      "grad_norm": 2.004711627960205,
      "learning_rate": 1.848205128205128e-05,
      "loss": 0.6071,
      "step": 296
    },
    {
      "epoch": 3.7955271565495208,
      "grad_norm": 1.9623836278915405,
      "learning_rate": 1.847692307692308e-05,
      "loss": 0.5376,
      "step": 297
    },
    {
      "epoch": 3.8083067092651754,
      "grad_norm": 1.977774977684021,
      "learning_rate": 1.8471794871794873e-05,
      "loss": 0.4829,
      "step": 298
    },
    {
      "epoch": 3.8210862619808306,
      "grad_norm": 2.2446415424346924,
      "learning_rate": 1.8466666666666667e-05,
      "loss": 0.4534,
      "step": 299
    },
    {
      "epoch": 3.8338658146964857,
      "grad_norm": 2.3083629608154297,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 0.6234,
      "step": 300
    },
    {
      "epoch": 3.8338658146964857,
      "eval_loss": 0.5684083700180054,
      "eval_runtime": 180.8233,
      "eval_samples_per_second": 0.868,
      "eval_steps_per_second": 0.111,
      "step": 300
    },
    {
      "epoch": 3.8466453674121404,
      "grad_norm": 2.4002017974853516,
      "learning_rate": 1.845641025641026e-05,
      "loss": 0.6042,
      "step": 301
    },
    {
      "epoch": 3.8594249201277955,
      "grad_norm": 1.574140191078186,
      "learning_rate": 1.8451282051282053e-05,
      "loss": 0.3668,
      "step": 302
    },
    {
      "epoch": 3.8722044728434506,
      "grad_norm": 1.857406497001648,
      "learning_rate": 1.8446153846153847e-05,
      "loss": 0.4021,
      "step": 303
    },
    {
      "epoch": 3.8849840255591053,
      "grad_norm": 2.3369333744049072,
      "learning_rate": 1.8441025641025644e-05,
      "loss": 0.4543,
      "step": 304
    },
    {
      "epoch": 3.8977635782747604,
      "grad_norm": 2.348994255065918,
      "learning_rate": 1.8435897435897435e-05,
      "loss": 0.5099,
      "step": 305
    },
    {
      "epoch": 3.9105431309904155,
      "grad_norm": 2.381173610687256,
      "learning_rate": 1.8430769230769232e-05,
      "loss": 0.5886,
      "step": 306
    },
    {
      "epoch": 3.92332268370607,
      "grad_norm": 2.188190460205078,
      "learning_rate": 1.8425641025641026e-05,
      "loss": 0.5159,
      "step": 307
    },
    {
      "epoch": 3.9361022364217253,
      "grad_norm": 2.331792116165161,
      "learning_rate": 1.842051282051282e-05,
      "loss": 0.5538,
      "step": 308
    },
    {
      "epoch": 3.9488817891373804,
      "grad_norm": 2.4262869358062744,
      "learning_rate": 1.8415384615384618e-05,
      "loss": 0.4675,
      "step": 309
    },
    {
      "epoch": 3.961661341853035,
      "grad_norm": 2.115399122238159,
      "learning_rate": 1.8410256410256412e-05,
      "loss": 0.4631,
      "step": 310
    },
    {
      "epoch": 3.97444089456869,
      "grad_norm": 2.0459988117218018,
      "learning_rate": 1.8405128205128206e-05,
      "loss": 0.6768,
      "step": 311
    },
    {
      "epoch": 3.987220447284345,
      "grad_norm": 3.0487661361694336,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.6092,
      "step": 312
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.4635841846466064,
      "learning_rate": 1.8394871794871797e-05,
      "loss": 0.7452,
      "step": 313
    },
    {
      "epoch": 4.012779552715655,
      "grad_norm": 3.330390214920044,
      "learning_rate": 1.838974358974359e-05,
      "loss": 0.6009,
      "step": 314
    },
    {
      "epoch": 4.02555910543131,
      "grad_norm": 2.3095197677612305,
      "learning_rate": 1.8384615384615386e-05,
      "loss": 0.4163,
      "step": 315
    },
    {
      "epoch": 4.038338658146965,
      "grad_norm": 2.4546499252319336,
      "learning_rate": 1.837948717948718e-05,
      "loss": 0.5153,
      "step": 316
    },
    {
      "epoch": 4.05111821086262,
      "grad_norm": 2.220017671585083,
      "learning_rate": 1.8374358974358977e-05,
      "loss": 0.5124,
      "step": 317
    },
    {
      "epoch": 4.063897763578275,
      "grad_norm": 2.0015223026275635,
      "learning_rate": 1.836923076923077e-05,
      "loss": 0.4947,
      "step": 318
    },
    {
      "epoch": 4.07667731629393,
      "grad_norm": 2.588055372238159,
      "learning_rate": 1.8364102564102565e-05,
      "loss": 0.6403,
      "step": 319
    },
    {
      "epoch": 4.0894568690095845,
      "grad_norm": 2.0123562812805176,
      "learning_rate": 1.835897435897436e-05,
      "loss": 0.4782,
      "step": 320
    },
    {
      "epoch": 4.102236421725239,
      "grad_norm": 2.275867462158203,
      "learning_rate": 1.8353846153846157e-05,
      "loss": 0.5304,
      "step": 321
    },
    {
      "epoch": 4.115015974440895,
      "grad_norm": 2.5205776691436768,
      "learning_rate": 1.834871794871795e-05,
      "loss": 0.5028,
      "step": 322
    },
    {
      "epoch": 4.127795527156549,
      "grad_norm": 2.3919801712036133,
      "learning_rate": 1.8343589743589745e-05,
      "loss": 0.6356,
      "step": 323
    },
    {
      "epoch": 4.140575079872204,
      "grad_norm": 2.4463231563568115,
      "learning_rate": 1.8338461538461542e-05,
      "loss": 0.6879,
      "step": 324
    },
    {
      "epoch": 4.15335463258786,
      "grad_norm": 2.269652843475342,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.561,
      "step": 325
    },
    {
      "epoch": 4.166134185303514,
      "grad_norm": 2.2289443016052246,
      "learning_rate": 1.832820512820513e-05,
      "loss": 0.5963,
      "step": 326
    },
    {
      "epoch": 4.178913738019169,
      "grad_norm": 1.8987523317337036,
      "learning_rate": 1.8323076923076924e-05,
      "loss": 0.4023,
      "step": 327
    },
    {
      "epoch": 4.1916932907348246,
      "grad_norm": 1.9962126016616821,
      "learning_rate": 1.831794871794872e-05,
      "loss": 0.4075,
      "step": 328
    },
    {
      "epoch": 4.204472843450479,
      "grad_norm": 2.6134021282196045,
      "learning_rate": 1.8312820512820516e-05,
      "loss": 0.5599,
      "step": 329
    },
    {
      "epoch": 4.217252396166134,
      "grad_norm": 1.8641389608383179,
      "learning_rate": 1.830769230769231e-05,
      "loss": 0.4824,
      "step": 330
    },
    {
      "epoch": 4.2300319488817895,
      "grad_norm": 2.134913921356201,
      "learning_rate": 1.8302564102564104e-05,
      "loss": 0.4844,
      "step": 331
    },
    {
      "epoch": 4.242811501597444,
      "grad_norm": 2.1769747734069824,
      "learning_rate": 1.8297435897435898e-05,
      "loss": 0.429,
      "step": 332
    },
    {
      "epoch": 4.255591054313099,
      "grad_norm": 1.921721339225769,
      "learning_rate": 1.8292307692307692e-05,
      "loss": 0.3981,
      "step": 333
    },
    {
      "epoch": 4.268370607028754,
      "grad_norm": 2.113063335418701,
      "learning_rate": 1.828717948717949e-05,
      "loss": 0.499,
      "step": 334
    },
    {
      "epoch": 4.281150159744409,
      "grad_norm": 2.1482808589935303,
      "learning_rate": 1.8282051282051284e-05,
      "loss": 0.5937,
      "step": 335
    },
    {
      "epoch": 4.293929712460064,
      "grad_norm": 2.5962321758270264,
      "learning_rate": 1.8276923076923078e-05,
      "loss": 0.4059,
      "step": 336
    },
    {
      "epoch": 4.306709265175719,
      "grad_norm": 1.8513988256454468,
      "learning_rate": 1.827179487179487e-05,
      "loss": 0.4338,
      "step": 337
    },
    {
      "epoch": 4.319488817891374,
      "grad_norm": 2.2054789066314697,
      "learning_rate": 1.826666666666667e-05,
      "loss": 0.4734,
      "step": 338
    },
    {
      "epoch": 4.332268370607029,
      "grad_norm": 2.223257064819336,
      "learning_rate": 1.8261538461538463e-05,
      "loss": 0.5564,
      "step": 339
    },
    {
      "epoch": 4.345047923322683,
      "grad_norm": 2.102458953857422,
      "learning_rate": 1.8256410256410257e-05,
      "loss": 0.4475,
      "step": 340
    },
    {
      "epoch": 4.357827476038339,
      "grad_norm": 2.7025585174560547,
      "learning_rate": 1.8251282051282055e-05,
      "loss": 0.6849,
      "step": 341
    },
    {
      "epoch": 4.3706070287539935,
      "grad_norm": 2.189150094985962,
      "learning_rate": 1.8246153846153845e-05,
      "loss": 0.6558,
      "step": 342
    },
    {
      "epoch": 4.383386581469648,
      "grad_norm": 3.0373575687408447,
      "learning_rate": 1.8241025641025643e-05,
      "loss": 0.7083,
      "step": 343
    },
    {
      "epoch": 4.396166134185304,
      "grad_norm": 2.166837692260742,
      "learning_rate": 1.8235897435897437e-05,
      "loss": 0.3849,
      "step": 344
    },
    {
      "epoch": 4.4089456869009584,
      "grad_norm": 2.1708340644836426,
      "learning_rate": 1.823076923076923e-05,
      "loss": 0.4122,
      "step": 345
    },
    {
      "epoch": 4.421725239616613,
      "grad_norm": 2.3275482654571533,
      "learning_rate": 1.8225641025641028e-05,
      "loss": 0.3932,
      "step": 346
    },
    {
      "epoch": 4.434504792332269,
      "grad_norm": 3.253466844558716,
      "learning_rate": 1.8220512820512822e-05,
      "loss": 0.6906,
      "step": 347
    },
    {
      "epoch": 4.447284345047923,
      "grad_norm": 2.473674774169922,
      "learning_rate": 1.8215384615384616e-05,
      "loss": 0.6586,
      "step": 348
    },
    {
      "epoch": 4.460063897763578,
      "grad_norm": 2.164419651031494,
      "learning_rate": 1.8210256410256414e-05,
      "loss": 0.4456,
      "step": 349
    },
    {
      "epoch": 4.472843450479234,
      "grad_norm": 2.3611209392547607,
      "learning_rate": 1.8205128205128208e-05,
      "loss": 0.4352,
      "step": 350
    },
    {
      "epoch": 4.485623003194888,
      "grad_norm": 2.4800209999084473,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.4834,
      "step": 351
    },
    {
      "epoch": 4.498402555910543,
      "grad_norm": 2.4389426708221436,
      "learning_rate": 1.8194871794871796e-05,
      "loss": 0.57,
      "step": 352
    },
    {
      "epoch": 4.511182108626198,
      "grad_norm": 2.4953975677490234,
      "learning_rate": 1.818974358974359e-05,
      "loss": 0.3799,
      "step": 353
    },
    {
      "epoch": 4.523961661341853,
      "grad_norm": 1.9845151901245117,
      "learning_rate": 1.8184615384615384e-05,
      "loss": 0.4779,
      "step": 354
    },
    {
      "epoch": 4.536741214057508,
      "grad_norm": 2.2074553966522217,
      "learning_rate": 1.817948717948718e-05,
      "loss": 0.3435,
      "step": 355
    },
    {
      "epoch": 4.549520766773163,
      "grad_norm": 2.5801925659179688,
      "learning_rate": 1.8174358974358976e-05,
      "loss": 0.4775,
      "step": 356
    },
    {
      "epoch": 4.562300319488818,
      "grad_norm": 2.800104856491089,
      "learning_rate": 1.816923076923077e-05,
      "loss": 0.6591,
      "step": 357
    },
    {
      "epoch": 4.575079872204473,
      "grad_norm": 2.1821672916412354,
      "learning_rate": 1.8164102564102567e-05,
      "loss": 0.3604,
      "step": 358
    },
    {
      "epoch": 4.587859424920127,
      "grad_norm": 2.050424575805664,
      "learning_rate": 1.815897435897436e-05,
      "loss": 0.4017,
      "step": 359
    },
    {
      "epoch": 4.600638977635783,
      "grad_norm": 2.6958281993865967,
      "learning_rate": 1.8153846153846155e-05,
      "loss": 0.613,
      "step": 360
    },
    {
      "epoch": 4.613418530351438,
      "grad_norm": 2.337655544281006,
      "learning_rate": 1.8148717948717953e-05,
      "loss": 0.6911,
      "step": 361
    },
    {
      "epoch": 4.626198083067092,
      "grad_norm": 2.8686039447784424,
      "learning_rate": 1.8143589743589743e-05,
      "loss": 0.5681,
      "step": 362
    },
    {
      "epoch": 4.638977635782748,
      "grad_norm": 2.4105000495910645,
      "learning_rate": 1.813846153846154e-05,
      "loss": 0.6666,
      "step": 363
    },
    {
      "epoch": 4.651757188498403,
      "grad_norm": 2.0860559940338135,
      "learning_rate": 1.8133333333333335e-05,
      "loss": 0.411,
      "step": 364
    },
    {
      "epoch": 4.664536741214057,
      "grad_norm": 2.3210160732269287,
      "learning_rate": 1.812820512820513e-05,
      "loss": 0.4459,
      "step": 365
    },
    {
      "epoch": 4.677316293929713,
      "grad_norm": 1.9436842203140259,
      "learning_rate": 1.8123076923076926e-05,
      "loss": 0.493,
      "step": 366
    },
    {
      "epoch": 4.6900958466453675,
      "grad_norm": 2.178853750228882,
      "learning_rate": 1.811794871794872e-05,
      "loss": 0.6059,
      "step": 367
    },
    {
      "epoch": 4.702875399361022,
      "grad_norm": 2.5311412811279297,
      "learning_rate": 1.8112820512820514e-05,
      "loss": 0.504,
      "step": 368
    },
    {
      "epoch": 4.715654952076678,
      "grad_norm": 1.6586120128631592,
      "learning_rate": 1.810769230769231e-05,
      "loss": 0.4196,
      "step": 369
    },
    {
      "epoch": 4.728434504792332,
      "grad_norm": 2.5906753540039062,
      "learning_rate": 1.8102564102564102e-05,
      "loss": 0.462,
      "step": 370
    },
    {
      "epoch": 4.741214057507987,
      "grad_norm": 2.0996625423431396,
      "learning_rate": 1.80974358974359e-05,
      "loss": 0.5456,
      "step": 371
    },
    {
      "epoch": 4.753993610223642,
      "grad_norm": 2.0776517391204834,
      "learning_rate": 1.8092307692307694e-05,
      "loss": 0.5379,
      "step": 372
    },
    {
      "epoch": 4.766773162939297,
      "grad_norm": 1.8418030738830566,
      "learning_rate": 1.8087179487179488e-05,
      "loss": 0.4525,
      "step": 373
    },
    {
      "epoch": 4.779552715654952,
      "grad_norm": 2.2306466102600098,
      "learning_rate": 1.8082051282051282e-05,
      "loss": 0.4133,
      "step": 374
    },
    {
      "epoch": 4.792332268370607,
      "grad_norm": 2.0150136947631836,
      "learning_rate": 1.807692307692308e-05,
      "loss": 0.5694,
      "step": 375
    },
    {
      "epoch": 4.805111821086262,
      "grad_norm": 2.669576644897461,
      "learning_rate": 1.8071794871794874e-05,
      "loss": 0.7247,
      "step": 376
    },
    {
      "epoch": 4.817891373801917,
      "grad_norm": 2.7284021377563477,
      "learning_rate": 1.8066666666666668e-05,
      "loss": 0.4458,
      "step": 377
    },
    {
      "epoch": 4.830670926517572,
      "grad_norm": 1.9394841194152832,
      "learning_rate": 1.8061538461538465e-05,
      "loss": 0.4246,
      "step": 378
    },
    {
      "epoch": 4.843450479233227,
      "grad_norm": 2.0008041858673096,
      "learning_rate": 1.8056410256410256e-05,
      "loss": 0.3953,
      "step": 379
    },
    {
      "epoch": 4.856230031948882,
      "grad_norm": 2.4139251708984375,
      "learning_rate": 1.8051282051282053e-05,
      "loss": 0.5843,
      "step": 380
    },
    {
      "epoch": 4.8690095846645365,
      "grad_norm": 3.531747817993164,
      "learning_rate": 1.8046153846153847e-05,
      "loss": 0.4477,
      "step": 381
    },
    {
      "epoch": 4.881789137380192,
      "grad_norm": 2.701043128967285,
      "learning_rate": 1.804102564102564e-05,
      "loss": 0.5375,
      "step": 382
    },
    {
      "epoch": 4.894568690095847,
      "grad_norm": 2.51804256439209,
      "learning_rate": 1.803589743589744e-05,
      "loss": 0.442,
      "step": 383
    },
    {
      "epoch": 4.907348242811501,
      "grad_norm": 1.9672200679779053,
      "learning_rate": 1.8030769230769233e-05,
      "loss": 0.4485,
      "step": 384
    },
    {
      "epoch": 4.920127795527156,
      "grad_norm": 2.2077980041503906,
      "learning_rate": 1.8025641025641027e-05,
      "loss": 0.5409,
      "step": 385
    },
    {
      "epoch": 4.932907348242812,
      "grad_norm": 2.3517746925354004,
      "learning_rate": 1.802051282051282e-05,
      "loss": 0.5482,
      "step": 386
    },
    {
      "epoch": 4.945686900958466,
      "grad_norm": 2.220970869064331,
      "learning_rate": 1.8015384615384618e-05,
      "loss": 0.757,
      "step": 387
    },
    {
      "epoch": 4.958466453674122,
      "grad_norm": 2.4243104457855225,
      "learning_rate": 1.8010256410256412e-05,
      "loss": 0.5577,
      "step": 388
    },
    {
      "epoch": 4.9712460063897765,
      "grad_norm": 2.6521098613739014,
      "learning_rate": 1.8005128205128206e-05,
      "loss": 0.6587,
      "step": 389
    },
    {
      "epoch": 4.984025559105431,
      "grad_norm": 2.181779384613037,
      "learning_rate": 1.8e-05,
      "loss": 0.4766,
      "step": 390
    },
    {
      "epoch": 4.996805111821086,
      "grad_norm": 2.2270212173461914,
      "learning_rate": 1.7994871794871795e-05,
      "loss": 0.528,
      "step": 391
    },
    {
      "epoch": 5.0095846645367414,
      "grad_norm": 2.2594521045684814,
      "learning_rate": 1.7989743589743592e-05,
      "loss": 0.4741,
      "step": 392
    },
    {
      "epoch": 5.022364217252396,
      "grad_norm": 2.1782121658325195,
      "learning_rate": 1.7984615384615386e-05,
      "loss": 0.599,
      "step": 393
    },
    {
      "epoch": 5.035143769968051,
      "grad_norm": 2.0388574600219727,
      "learning_rate": 1.797948717948718e-05,
      "loss": 0.7343,
      "step": 394
    },
    {
      "epoch": 5.047923322683706,
      "grad_norm": 2.135293483734131,
      "learning_rate": 1.7974358974358977e-05,
      "loss": 0.6222,
      "step": 395
    },
    {
      "epoch": 5.060702875399361,
      "grad_norm": 2.4240000247955322,
      "learning_rate": 1.796923076923077e-05,
      "loss": 0.5916,
      "step": 396
    },
    {
      "epoch": 5.073482428115016,
      "grad_norm": 2.2460944652557373,
      "learning_rate": 1.7964102564102566e-05,
      "loss": 0.571,
      "step": 397
    },
    {
      "epoch": 5.086261980830671,
      "grad_norm": 2.330200433731079,
      "learning_rate": 1.7958974358974363e-05,
      "loss": 0.4237,
      "step": 398
    },
    {
      "epoch": 5.099041533546326,
      "grad_norm": 2.227525234222412,
      "learning_rate": 1.7953846153846154e-05,
      "loss": 0.64,
      "step": 399
    },
    {
      "epoch": 5.111821086261981,
      "grad_norm": 1.9940121173858643,
      "learning_rate": 1.794871794871795e-05,
      "loss": 0.4471,
      "step": 400
    },
    {
      "epoch": 5.111821086261981,
      "eval_loss": 0.5250003337860107,
      "eval_runtime": 181.0045,
      "eval_samples_per_second": 0.867,
      "eval_steps_per_second": 0.11,
      "step": 400
    },
    {
      "epoch": 5.124600638977636,
      "grad_norm": 1.8769639730453491,
      "learning_rate": 1.7943589743589745e-05,
      "loss": 0.3262,
      "step": 401
    },
    {
      "epoch": 5.137380191693291,
      "grad_norm": 2.1913700103759766,
      "learning_rate": 1.793846153846154e-05,
      "loss": 0.502,
      "step": 402
    },
    {
      "epoch": 5.1501597444089455,
      "grad_norm": 1.8566724061965942,
      "learning_rate": 1.7933333333333333e-05,
      "loss": 0.4005,
      "step": 403
    },
    {
      "epoch": 5.1629392971246,
      "grad_norm": 2.3526904582977295,
      "learning_rate": 1.792820512820513e-05,
      "loss": 0.6051,
      "step": 404
    },
    {
      "epoch": 5.175718849840256,
      "grad_norm": 2.3529820442199707,
      "learning_rate": 1.7923076923076925e-05,
      "loss": 0.5727,
      "step": 405
    },
    {
      "epoch": 5.18849840255591,
      "grad_norm": 2.2235405445098877,
      "learning_rate": 1.791794871794872e-05,
      "loss": 0.3353,
      "step": 406
    },
    {
      "epoch": 5.201277955271565,
      "grad_norm": 2.3372035026550293,
      "learning_rate": 1.7912820512820516e-05,
      "loss": 0.4259,
      "step": 407
    },
    {
      "epoch": 5.214057507987221,
      "grad_norm": 2.4938113689422607,
      "learning_rate": 1.7907692307692307e-05,
      "loss": 0.4206,
      "step": 408
    },
    {
      "epoch": 5.226837060702875,
      "grad_norm": 2.129610776901245,
      "learning_rate": 1.7902564102564104e-05,
      "loss": 0.5185,
      "step": 409
    },
    {
      "epoch": 5.23961661341853,
      "grad_norm": 2.3278634548187256,
      "learning_rate": 1.78974358974359e-05,
      "loss": 0.4314,
      "step": 410
    },
    {
      "epoch": 5.252396166134186,
      "grad_norm": 2.271420478820801,
      "learning_rate": 1.7892307692307692e-05,
      "loss": 0.3916,
      "step": 411
    },
    {
      "epoch": 5.26517571884984,
      "grad_norm": 1.915278434753418,
      "learning_rate": 1.788717948717949e-05,
      "loss": 0.4356,
      "step": 412
    },
    {
      "epoch": 5.277955271565495,
      "grad_norm": 2.2529425621032715,
      "learning_rate": 1.7882051282051284e-05,
      "loss": 0.4071,
      "step": 413
    },
    {
      "epoch": 5.2907348242811505,
      "grad_norm": 2.971188545227051,
      "learning_rate": 1.7876923076923078e-05,
      "loss": 0.3759,
      "step": 414
    },
    {
      "epoch": 5.303514376996805,
      "grad_norm": 1.9062541723251343,
      "learning_rate": 1.7871794871794875e-05,
      "loss": 0.3086,
      "step": 415
    },
    {
      "epoch": 5.31629392971246,
      "grad_norm": 2.860849618911743,
      "learning_rate": 1.7866666666666666e-05,
      "loss": 0.529,
      "step": 416
    },
    {
      "epoch": 5.329073482428115,
      "grad_norm": 2.6254971027374268,
      "learning_rate": 1.7861538461538464e-05,
      "loss": 0.4456,
      "step": 417
    },
    {
      "epoch": 5.34185303514377,
      "grad_norm": 2.4462695121765137,
      "learning_rate": 1.7856410256410258e-05,
      "loss": 0.5505,
      "step": 418
    },
    {
      "epoch": 5.354632587859425,
      "grad_norm": 3.001704216003418,
      "learning_rate": 1.785128205128205e-05,
      "loss": 0.4583,
      "step": 419
    },
    {
      "epoch": 5.36741214057508,
      "grad_norm": 2.2287330627441406,
      "learning_rate": 1.784615384615385e-05,
      "loss": 0.4334,
      "step": 420
    },
    {
      "epoch": 5.380191693290735,
      "grad_norm": 2.3208580017089844,
      "learning_rate": 1.7841025641025643e-05,
      "loss": 0.4962,
      "step": 421
    },
    {
      "epoch": 5.39297124600639,
      "grad_norm": 2.5191850662231445,
      "learning_rate": 1.7835897435897437e-05,
      "loss": 0.5019,
      "step": 422
    },
    {
      "epoch": 5.405750798722044,
      "grad_norm": 2.3573968410491943,
      "learning_rate": 1.783076923076923e-05,
      "loss": 0.5416,
      "step": 423
    },
    {
      "epoch": 5.4185303514377,
      "grad_norm": 2.279006242752075,
      "learning_rate": 1.782564102564103e-05,
      "loss": 0.3585,
      "step": 424
    },
    {
      "epoch": 5.431309904153355,
      "grad_norm": 1.831701397895813,
      "learning_rate": 1.7820512820512823e-05,
      "loss": 0.4025,
      "step": 425
    },
    {
      "epoch": 5.444089456869009,
      "grad_norm": 2.1646835803985596,
      "learning_rate": 1.7815384615384617e-05,
      "loss": 0.3548,
      "step": 426
    },
    {
      "epoch": 5.456869009584665,
      "grad_norm": 2.598620653152466,
      "learning_rate": 1.781025641025641e-05,
      "loss": 0.4204,
      "step": 427
    },
    {
      "epoch": 5.4696485623003195,
      "grad_norm": 3.2102348804473877,
      "learning_rate": 1.7805128205128205e-05,
      "loss": 0.6015,
      "step": 428
    },
    {
      "epoch": 5.482428115015974,
      "grad_norm": 2.23750376701355,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 0.3484,
      "step": 429
    },
    {
      "epoch": 5.49520766773163,
      "grad_norm": 2.155402421951294,
      "learning_rate": 1.7794871794871796e-05,
      "loss": 0.4067,
      "step": 430
    },
    {
      "epoch": 5.507987220447284,
      "grad_norm": 2.757122755050659,
      "learning_rate": 1.778974358974359e-05,
      "loss": 0.5165,
      "step": 431
    },
    {
      "epoch": 5.520766773162939,
      "grad_norm": 2.5486087799072266,
      "learning_rate": 1.7784615384615388e-05,
      "loss": 0.5124,
      "step": 432
    },
    {
      "epoch": 5.533546325878595,
      "grad_norm": 2.1844065189361572,
      "learning_rate": 1.7779487179487182e-05,
      "loss": 0.5247,
      "step": 433
    },
    {
      "epoch": 5.546325878594249,
      "grad_norm": 2.0668201446533203,
      "learning_rate": 1.7774358974358976e-05,
      "loss": 0.6111,
      "step": 434
    },
    {
      "epoch": 5.559105431309904,
      "grad_norm": 2.5725159645080566,
      "learning_rate": 1.776923076923077e-05,
      "loss": 0.4193,
      "step": 435
    },
    {
      "epoch": 5.571884984025559,
      "grad_norm": 2.1765971183776855,
      "learning_rate": 1.7764102564102564e-05,
      "loss": 0.3695,
      "step": 436
    },
    {
      "epoch": 5.584664536741214,
      "grad_norm": 2.1050305366516113,
      "learning_rate": 1.775897435897436e-05,
      "loss": 0.4419,
      "step": 437
    },
    {
      "epoch": 5.597444089456869,
      "grad_norm": 2.6420412063598633,
      "learning_rate": 1.7753846153846156e-05,
      "loss": 0.6949,
      "step": 438
    },
    {
      "epoch": 5.6102236421725244,
      "grad_norm": 3.008789300918579,
      "learning_rate": 1.774871794871795e-05,
      "loss": 0.4901,
      "step": 439
    },
    {
      "epoch": 5.623003194888179,
      "grad_norm": 2.481837511062622,
      "learning_rate": 1.7743589743589744e-05,
      "loss": 0.6561,
      "step": 440
    },
    {
      "epoch": 5.635782747603834,
      "grad_norm": 2.06085467338562,
      "learning_rate": 1.773846153846154e-05,
      "loss": 0.4835,
      "step": 441
    },
    {
      "epoch": 5.6485623003194885,
      "grad_norm": 1.9079381227493286,
      "learning_rate": 1.7733333333333335e-05,
      "loss": 0.3557,
      "step": 442
    },
    {
      "epoch": 5.661341853035144,
      "grad_norm": 2.1730477809906006,
      "learning_rate": 1.772820512820513e-05,
      "loss": 0.4584,
      "step": 443
    },
    {
      "epoch": 5.674121405750799,
      "grad_norm": 2.3779098987579346,
      "learning_rate": 1.7723076923076927e-05,
      "loss": 0.2891,
      "step": 444
    },
    {
      "epoch": 5.686900958466453,
      "grad_norm": 2.3261685371398926,
      "learning_rate": 1.7717948717948717e-05,
      "loss": 0.3954,
      "step": 445
    },
    {
      "epoch": 5.699680511182109,
      "grad_norm": 1.8987905979156494,
      "learning_rate": 1.7712820512820515e-05,
      "loss": 0.4915,
      "step": 446
    },
    {
      "epoch": 5.712460063897764,
      "grad_norm": 2.268406867980957,
      "learning_rate": 1.770769230769231e-05,
      "loss": 0.3933,
      "step": 447
    },
    {
      "epoch": 5.725239616613418,
      "grad_norm": 2.096902847290039,
      "learning_rate": 1.7702564102564103e-05,
      "loss": 0.3868,
      "step": 448
    },
    {
      "epoch": 5.738019169329074,
      "grad_norm": 2.9784042835235596,
      "learning_rate": 1.76974358974359e-05,
      "loss": 0.4829,
      "step": 449
    },
    {
      "epoch": 5.7507987220447285,
      "grad_norm": 2.718769073486328,
      "learning_rate": 1.7692307692307694e-05,
      "loss": 0.523,
      "step": 450
    },
    {
      "epoch": 5.763578274760383,
      "grad_norm": 1.9845243692398071,
      "learning_rate": 1.768717948717949e-05,
      "loss": 0.3861,
      "step": 451
    },
    {
      "epoch": 5.776357827476039,
      "grad_norm": 2.1359245777130127,
      "learning_rate": 1.7682051282051283e-05,
      "loss": 0.4859,
      "step": 452
    },
    {
      "epoch": 5.789137380191693,
      "grad_norm": 2.4216926097869873,
      "learning_rate": 1.7676923076923077e-05,
      "loss": 0.4118,
      "step": 453
    },
    {
      "epoch": 5.801916932907348,
      "grad_norm": 2.8170251846313477,
      "learning_rate": 1.7671794871794874e-05,
      "loss": 0.5945,
      "step": 454
    },
    {
      "epoch": 5.814696485623003,
      "grad_norm": 2.7723488807678223,
      "learning_rate": 1.7666666666666668e-05,
      "loss": 0.6911,
      "step": 455
    },
    {
      "epoch": 5.827476038338658,
      "grad_norm": 2.405487537384033,
      "learning_rate": 1.7661538461538462e-05,
      "loss": 0.5594,
      "step": 456
    },
    {
      "epoch": 5.840255591054313,
      "grad_norm": 2.5241634845733643,
      "learning_rate": 1.7656410256410256e-05,
      "loss": 0.446,
      "step": 457
    },
    {
      "epoch": 5.853035143769968,
      "grad_norm": 2.443591833114624,
      "learning_rate": 1.7651282051282054e-05,
      "loss": 0.4842,
      "step": 458
    },
    {
      "epoch": 5.865814696485623,
      "grad_norm": 2.4201200008392334,
      "learning_rate": 1.7646153846153848e-05,
      "loss": 0.5486,
      "step": 459
    },
    {
      "epoch": 5.878594249201278,
      "grad_norm": 2.012280225753784,
      "learning_rate": 1.7641025641025642e-05,
      "loss": 0.514,
      "step": 460
    },
    {
      "epoch": 5.891373801916933,
      "grad_norm": 2.587604284286499,
      "learning_rate": 1.763589743589744e-05,
      "loss": 0.522,
      "step": 461
    },
    {
      "epoch": 5.904153354632588,
      "grad_norm": 2.1839914321899414,
      "learning_rate": 1.763076923076923e-05,
      "loss": 0.4902,
      "step": 462
    },
    {
      "epoch": 5.916932907348243,
      "grad_norm": 2.1530771255493164,
      "learning_rate": 1.7625641025641027e-05,
      "loss": 0.3949,
      "step": 463
    },
    {
      "epoch": 5.9297124600638975,
      "grad_norm": 2.5098278522491455,
      "learning_rate": 1.762051282051282e-05,
      "loss": 0.5611,
      "step": 464
    },
    {
      "epoch": 5.942492012779553,
      "grad_norm": 2.27274227142334,
      "learning_rate": 1.7615384615384615e-05,
      "loss": 0.4455,
      "step": 465
    },
    {
      "epoch": 5.955271565495208,
      "grad_norm": 2.370197057723999,
      "learning_rate": 1.7610256410256413e-05,
      "loss": 0.4224,
      "step": 466
    },
    {
      "epoch": 5.968051118210862,
      "grad_norm": 1.684553623199463,
      "learning_rate": 1.7605128205128207e-05,
      "loss": 0.4083,
      "step": 467
    },
    {
      "epoch": 5.980830670926517,
      "grad_norm": 2.499558210372925,
      "learning_rate": 1.76e-05,
      "loss": 0.4626,
      "step": 468
    },
    {
      "epoch": 5.993610223642173,
      "grad_norm": 2.5789244174957275,
      "learning_rate": 1.75948717948718e-05,
      "loss": 0.5812,
      "step": 469
    },
    {
      "epoch": 6.006389776357827,
      "grad_norm": 2.08774471282959,
      "learning_rate": 1.7589743589743592e-05,
      "loss": 0.4607,
      "step": 470
    },
    {
      "epoch": 6.019169329073482,
      "grad_norm": 2.242676258087158,
      "learning_rate": 1.7584615384615386e-05,
      "loss": 0.3949,
      "step": 471
    },
    {
      "epoch": 6.031948881789138,
      "grad_norm": 2.3085381984710693,
      "learning_rate": 1.757948717948718e-05,
      "loss": 0.518,
      "step": 472
    },
    {
      "epoch": 6.044728434504792,
      "grad_norm": 2.412428855895996,
      "learning_rate": 1.7574358974358975e-05,
      "loss": 0.5979,
      "step": 473
    },
    {
      "epoch": 6.057507987220447,
      "grad_norm": 2.2813498973846436,
      "learning_rate": 1.7569230769230772e-05,
      "loss": 0.3837,
      "step": 474
    },
    {
      "epoch": 6.0702875399361025,
      "grad_norm": 2.6682474613189697,
      "learning_rate": 1.7564102564102566e-05,
      "loss": 0.4495,
      "step": 475
    },
    {
      "epoch": 6.083067092651757,
      "grad_norm": 3.1144282817840576,
      "learning_rate": 1.755897435897436e-05,
      "loss": 0.4583,
      "step": 476
    },
    {
      "epoch": 6.095846645367412,
      "grad_norm": 2.3847062587738037,
      "learning_rate": 1.7553846153846154e-05,
      "loss": 0.4137,
      "step": 477
    },
    {
      "epoch": 6.108626198083067,
      "grad_norm": 2.61763334274292,
      "learning_rate": 1.754871794871795e-05,
      "loss": 0.3974,
      "step": 478
    },
    {
      "epoch": 6.121405750798722,
      "grad_norm": 2.383402109146118,
      "learning_rate": 1.7543589743589746e-05,
      "loss": 0.4834,
      "step": 479
    },
    {
      "epoch": 6.134185303514377,
      "grad_norm": 3.521724224090576,
      "learning_rate": 1.753846153846154e-05,
      "loss": 0.3352,
      "step": 480
    },
    {
      "epoch": 6.146964856230032,
      "grad_norm": 2.1919541358947754,
      "learning_rate": 1.7533333333333337e-05,
      "loss": 0.3448,
      "step": 481
    },
    {
      "epoch": 6.159744408945687,
      "grad_norm": 2.2474188804626465,
      "learning_rate": 1.7528205128205128e-05,
      "loss": 0.4474,
      "step": 482
    },
    {
      "epoch": 6.172523961661342,
      "grad_norm": 2.5635788440704346,
      "learning_rate": 1.7523076923076925e-05,
      "loss": 0.492,
      "step": 483
    },
    {
      "epoch": 6.185303514376997,
      "grad_norm": 2.388453245162964,
      "learning_rate": 1.751794871794872e-05,
      "loss": 0.3088,
      "step": 484
    },
    {
      "epoch": 6.198083067092652,
      "grad_norm": 2.631474256515503,
      "learning_rate": 1.7512820512820513e-05,
      "loss": 0.4037,
      "step": 485
    },
    {
      "epoch": 6.210862619808307,
      "grad_norm": 2.8673110008239746,
      "learning_rate": 1.750769230769231e-05,
      "loss": 0.454,
      "step": 486
    },
    {
      "epoch": 6.223642172523961,
      "grad_norm": 3.3764970302581787,
      "learning_rate": 1.7502564102564105e-05,
      "loss": 0.5351,
      "step": 487
    },
    {
      "epoch": 6.236421725239617,
      "grad_norm": 2.687863349914551,
      "learning_rate": 1.74974358974359e-05,
      "loss": 0.4543,
      "step": 488
    },
    {
      "epoch": 6.2492012779552715,
      "grad_norm": 2.591404914855957,
      "learning_rate": 1.7492307692307693e-05,
      "loss": 0.4279,
      "step": 489
    },
    {
      "epoch": 6.261980830670926,
      "grad_norm": 2.737273693084717,
      "learning_rate": 1.7487179487179487e-05,
      "loss": 0.6408,
      "step": 490
    },
    {
      "epoch": 6.274760383386582,
      "grad_norm": 2.39886736869812,
      "learning_rate": 1.7482051282051284e-05,
      "loss": 0.3358,
      "step": 491
    },
    {
      "epoch": 6.287539936102236,
      "grad_norm": 2.9275259971618652,
      "learning_rate": 1.747692307692308e-05,
      "loss": 0.6465,
      "step": 492
    },
    {
      "epoch": 6.300319488817891,
      "grad_norm": 2.810349941253662,
      "learning_rate": 1.7471794871794873e-05,
      "loss": 0.3819,
      "step": 493
    },
    {
      "epoch": 6.313099041533547,
      "grad_norm": 2.2829222679138184,
      "learning_rate": 1.7466666666666667e-05,
      "loss": 0.5803,
      "step": 494
    },
    {
      "epoch": 6.325878594249201,
      "grad_norm": 2.397336721420288,
      "learning_rate": 1.7461538461538464e-05,
      "loss": 0.4829,
      "step": 495
    },
    {
      "epoch": 6.338658146964856,
      "grad_norm": 2.189927101135254,
      "learning_rate": 1.7456410256410258e-05,
      "loss": 0.5757,
      "step": 496
    },
    {
      "epoch": 6.3514376996805115,
      "grad_norm": 2.369382381439209,
      "learning_rate": 1.7451282051282052e-05,
      "loss": 0.2897,
      "step": 497
    },
    {
      "epoch": 6.364217252396166,
      "grad_norm": 2.5294623374938965,
      "learning_rate": 1.744615384615385e-05,
      "loss": 0.5625,
      "step": 498
    },
    {
      "epoch": 6.376996805111821,
      "grad_norm": 2.429304361343384,
      "learning_rate": 1.744102564102564e-05,
      "loss": 0.5052,
      "step": 499
    },
    {
      "epoch": 6.389776357827476,
      "grad_norm": 2.7437777519226074,
      "learning_rate": 1.7435897435897438e-05,
      "loss": 0.4608,
      "step": 500
    },
    {
      "epoch": 6.389776357827476,
      "eval_loss": 0.5022287964820862,
      "eval_runtime": 182.7071,
      "eval_samples_per_second": 0.859,
      "eval_steps_per_second": 0.109,
      "step": 500
    },
    {
      "epoch": 6.402555910543131,
      "grad_norm": 3.0428144931793213,
      "learning_rate": 1.7430769230769232e-05,
      "loss": 0.5481,
      "step": 501
    },
    {
      "epoch": 6.415335463258786,
      "grad_norm": 1.789218544960022,
      "learning_rate": 1.7425641025641026e-05,
      "loss": 0.379,
      "step": 502
    },
    {
      "epoch": 6.428115015974441,
      "grad_norm": 2.447986364364624,
      "learning_rate": 1.7420512820512823e-05,
      "loss": 0.36,
      "step": 503
    },
    {
      "epoch": 6.440894568690096,
      "grad_norm": 2.2227280139923096,
      "learning_rate": 1.7415384615384617e-05,
      "loss": 0.4233,
      "step": 504
    },
    {
      "epoch": 6.453674121405751,
      "grad_norm": 2.918168783187866,
      "learning_rate": 1.741025641025641e-05,
      "loss": 0.4857,
      "step": 505
    },
    {
      "epoch": 6.466453674121405,
      "grad_norm": 2.720658779144287,
      "learning_rate": 1.7405128205128205e-05,
      "loss": 0.3815,
      "step": 506
    },
    {
      "epoch": 6.479233226837061,
      "grad_norm": 2.6913065910339355,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 0.4462,
      "step": 507
    },
    {
      "epoch": 6.492012779552716,
      "grad_norm": 2.7055540084838867,
      "learning_rate": 1.7394871794871797e-05,
      "loss": 0.474,
      "step": 508
    },
    {
      "epoch": 6.50479233226837,
      "grad_norm": 2.182354688644409,
      "learning_rate": 1.738974358974359e-05,
      "loss": 0.4572,
      "step": 509
    },
    {
      "epoch": 6.517571884984026,
      "grad_norm": 2.5811660289764404,
      "learning_rate": 1.7384615384615385e-05,
      "loss": 0.3676,
      "step": 510
    },
    {
      "epoch": 6.5303514376996805,
      "grad_norm": 2.9754414558410645,
      "learning_rate": 1.737948717948718e-05,
      "loss": 0.5659,
      "step": 511
    },
    {
      "epoch": 6.543130990415335,
      "grad_norm": 3.129412889480591,
      "learning_rate": 1.7374358974358976e-05,
      "loss": 0.4372,
      "step": 512
    },
    {
      "epoch": 6.555910543130991,
      "grad_norm": 3.000218391418457,
      "learning_rate": 1.736923076923077e-05,
      "loss": 0.5066,
      "step": 513
    },
    {
      "epoch": 6.568690095846645,
      "grad_norm": 3.06145977973938,
      "learning_rate": 1.7364102564102565e-05,
      "loss": 0.456,
      "step": 514
    },
    {
      "epoch": 6.5814696485623,
      "grad_norm": 2.7481515407562256,
      "learning_rate": 1.7358974358974362e-05,
      "loss": 0.3037,
      "step": 515
    },
    {
      "epoch": 6.594249201277956,
      "grad_norm": 2.6574368476867676,
      "learning_rate": 1.7353846153846156e-05,
      "loss": 0.4023,
      "step": 516
    },
    {
      "epoch": 6.60702875399361,
      "grad_norm": 2.2366273403167725,
      "learning_rate": 1.734871794871795e-05,
      "loss": 0.4253,
      "step": 517
    },
    {
      "epoch": 6.619808306709265,
      "grad_norm": 2.5262792110443115,
      "learning_rate": 1.7343589743589748e-05,
      "loss": 0.4337,
      "step": 518
    },
    {
      "epoch": 6.63258785942492,
      "grad_norm": 2.2034177780151367,
      "learning_rate": 1.7338461538461538e-05,
      "loss": 0.4043,
      "step": 519
    },
    {
      "epoch": 6.645367412140575,
      "grad_norm": 3.698709487915039,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.6953,
      "step": 520
    },
    {
      "epoch": 6.65814696485623,
      "grad_norm": 2.3171446323394775,
      "learning_rate": 1.732820512820513e-05,
      "loss": 0.4593,
      "step": 521
    },
    {
      "epoch": 6.6709265175718855,
      "grad_norm": 2.540459156036377,
      "learning_rate": 1.7323076923076924e-05,
      "loss": 0.5598,
      "step": 522
    },
    {
      "epoch": 6.68370607028754,
      "grad_norm": 2.8060197830200195,
      "learning_rate": 1.731794871794872e-05,
      "loss": 0.3275,
      "step": 523
    },
    {
      "epoch": 6.696485623003195,
      "grad_norm": 2.904841899871826,
      "learning_rate": 1.7312820512820515e-05,
      "loss": 0.5502,
      "step": 524
    },
    {
      "epoch": 6.7092651757188495,
      "grad_norm": 2.756107807159424,
      "learning_rate": 1.730769230769231e-05,
      "loss": 0.4434,
      "step": 525
    },
    {
      "epoch": 6.722044728434505,
      "grad_norm": 2.744088888168335,
      "learning_rate": 1.7302564102564103e-05,
      "loss": 0.4104,
      "step": 526
    },
    {
      "epoch": 6.73482428115016,
      "grad_norm": 2.26206636428833,
      "learning_rate": 1.72974358974359e-05,
      "loss": 0.4231,
      "step": 527
    },
    {
      "epoch": 6.747603833865814,
      "grad_norm": 2.0125904083251953,
      "learning_rate": 1.729230769230769e-05,
      "loss": 0.4656,
      "step": 528
    },
    {
      "epoch": 6.76038338658147,
      "grad_norm": 2.1869449615478516,
      "learning_rate": 1.728717948717949e-05,
      "loss": 0.3672,
      "step": 529
    },
    {
      "epoch": 6.773162939297125,
      "grad_norm": 1.9915401935577393,
      "learning_rate": 1.7282051282051283e-05,
      "loss": 0.3877,
      "step": 530
    },
    {
      "epoch": 6.785942492012779,
      "grad_norm": 2.237344741821289,
      "learning_rate": 1.7276923076923077e-05,
      "loss": 0.3781,
      "step": 531
    },
    {
      "epoch": 6.798722044728435,
      "grad_norm": 2.5108916759490967,
      "learning_rate": 1.7271794871794874e-05,
      "loss": 0.5929,
      "step": 532
    },
    {
      "epoch": 6.81150159744409,
      "grad_norm": 2.3276259899139404,
      "learning_rate": 1.726666666666667e-05,
      "loss": 0.3589,
      "step": 533
    },
    {
      "epoch": 6.824281150159744,
      "grad_norm": 3.0176141262054443,
      "learning_rate": 1.7261538461538463e-05,
      "loss": 0.5621,
      "step": 534
    },
    {
      "epoch": 6.8370607028754,
      "grad_norm": 2.0819737911224365,
      "learning_rate": 1.725641025641026e-05,
      "loss": 0.3701,
      "step": 535
    },
    {
      "epoch": 6.8498402555910545,
      "grad_norm": 2.4570374488830566,
      "learning_rate": 1.725128205128205e-05,
      "loss": 0.463,
      "step": 536
    },
    {
      "epoch": 6.862619808306709,
      "grad_norm": 2.484651803970337,
      "learning_rate": 1.7246153846153848e-05,
      "loss": 0.4525,
      "step": 537
    },
    {
      "epoch": 6.875399361022364,
      "grad_norm": 2.1618449687957764,
      "learning_rate": 1.7241025641025642e-05,
      "loss": 0.4213,
      "step": 538
    },
    {
      "epoch": 6.888178913738019,
      "grad_norm": 2.1422417163848877,
      "learning_rate": 1.7235897435897436e-05,
      "loss": 0.4375,
      "step": 539
    },
    {
      "epoch": 6.900958466453674,
      "grad_norm": 2.6813485622406006,
      "learning_rate": 1.7230769230769234e-05,
      "loss": 0.5162,
      "step": 540
    },
    {
      "epoch": 6.913738019169329,
      "grad_norm": 2.3753178119659424,
      "learning_rate": 1.7225641025641028e-05,
      "loss": 0.4616,
      "step": 541
    },
    {
      "epoch": 6.926517571884984,
      "grad_norm": 2.3657495975494385,
      "learning_rate": 1.7220512820512822e-05,
      "loss": 0.5949,
      "step": 542
    },
    {
      "epoch": 6.939297124600639,
      "grad_norm": 2.5588560104370117,
      "learning_rate": 1.7215384615384616e-05,
      "loss": 0.5441,
      "step": 543
    },
    {
      "epoch": 6.952076677316294,
      "grad_norm": 2.5401077270507812,
      "learning_rate": 1.7210256410256413e-05,
      "loss": 0.4311,
      "step": 544
    },
    {
      "epoch": 6.964856230031949,
      "grad_norm": 2.427499532699585,
      "learning_rate": 1.7205128205128207e-05,
      "loss": 0.365,
      "step": 545
    },
    {
      "epoch": 6.977635782747604,
      "grad_norm": 2.534796714782715,
      "learning_rate": 1.72e-05,
      "loss": 0.4757,
      "step": 546
    },
    {
      "epoch": 6.9904153354632586,
      "grad_norm": 2.5496671199798584,
      "learning_rate": 1.7194871794871795e-05,
      "loss": 0.4701,
      "step": 547
    },
    {
      "epoch": 7.003194888178914,
      "grad_norm": 2.4693853855133057,
      "learning_rate": 1.718974358974359e-05,
      "loss": 0.3917,
      "step": 548
    },
    {
      "epoch": 7.015974440894569,
      "grad_norm": 2.0053486824035645,
      "learning_rate": 1.7184615384615387e-05,
      "loss": 0.356,
      "step": 549
    },
    {
      "epoch": 7.0287539936102235,
      "grad_norm": 2.299156427383423,
      "learning_rate": 1.717948717948718e-05,
      "loss": 0.3343,
      "step": 550
    },
    {
      "epoch": 7.041533546325879,
      "grad_norm": 2.448904514312744,
      "learning_rate": 1.7174358974358975e-05,
      "loss": 0.427,
      "step": 551
    },
    {
      "epoch": 7.054313099041534,
      "grad_norm": 2.3865370750427246,
      "learning_rate": 1.7169230769230772e-05,
      "loss": 0.4465,
      "step": 552
    },
    {
      "epoch": 7.067092651757188,
      "grad_norm": 2.5128276348114014,
      "learning_rate": 1.7164102564102566e-05,
      "loss": 0.4417,
      "step": 553
    },
    {
      "epoch": 7.079872204472843,
      "grad_norm": 2.5499255657196045,
      "learning_rate": 1.715897435897436e-05,
      "loss": 0.403,
      "step": 554
    },
    {
      "epoch": 7.092651757188499,
      "grad_norm": 2.176884174346924,
      "learning_rate": 1.7153846153846155e-05,
      "loss": 0.3505,
      "step": 555
    },
    {
      "epoch": 7.105431309904153,
      "grad_norm": 2.64473032951355,
      "learning_rate": 1.714871794871795e-05,
      "loss": 0.3731,
      "step": 556
    },
    {
      "epoch": 7.118210862619808,
      "grad_norm": 2.417506456375122,
      "learning_rate": 1.7143589743589746e-05,
      "loss": 0.3703,
      "step": 557
    },
    {
      "epoch": 7.1309904153354635,
      "grad_norm": 2.3878259658813477,
      "learning_rate": 1.713846153846154e-05,
      "loss": 0.3763,
      "step": 558
    },
    {
      "epoch": 7.143769968051118,
      "grad_norm": 2.473843574523926,
      "learning_rate": 1.7133333333333334e-05,
      "loss": 0.3958,
      "step": 559
    },
    {
      "epoch": 7.156549520766773,
      "grad_norm": 2.9820239543914795,
      "learning_rate": 1.7128205128205128e-05,
      "loss": 0.337,
      "step": 560
    },
    {
      "epoch": 7.169329073482428,
      "grad_norm": 2.1744096279144287,
      "learning_rate": 1.7123076923076926e-05,
      "loss": 0.4535,
      "step": 561
    },
    {
      "epoch": 7.182108626198083,
      "grad_norm": 2.5609939098358154,
      "learning_rate": 1.711794871794872e-05,
      "loss": 0.4376,
      "step": 562
    },
    {
      "epoch": 7.194888178913738,
      "grad_norm": 2.5857808589935303,
      "learning_rate": 1.7112820512820514e-05,
      "loss": 0.4187,
      "step": 563
    },
    {
      "epoch": 7.207667731629393,
      "grad_norm": 3.49450945854187,
      "learning_rate": 1.710769230769231e-05,
      "loss": 0.5547,
      "step": 564
    },
    {
      "epoch": 7.220447284345048,
      "grad_norm": 2.712728261947632,
      "learning_rate": 1.7102564102564102e-05,
      "loss": 0.3876,
      "step": 565
    },
    {
      "epoch": 7.233226837060703,
      "grad_norm": 2.9620182514190674,
      "learning_rate": 1.70974358974359e-05,
      "loss": 0.5748,
      "step": 566
    },
    {
      "epoch": 7.246006389776358,
      "grad_norm": 2.54795241355896,
      "learning_rate": 1.7092307692307693e-05,
      "loss": 0.336,
      "step": 567
    },
    {
      "epoch": 7.258785942492013,
      "grad_norm": 2.2571918964385986,
      "learning_rate": 1.7087179487179487e-05,
      "loss": 0.4416,
      "step": 568
    },
    {
      "epoch": 7.271565495207668,
      "grad_norm": 2.327420711517334,
      "learning_rate": 1.7082051282051285e-05,
      "loss": 0.3582,
      "step": 569
    },
    {
      "epoch": 7.284345047923322,
      "grad_norm": 2.4296958446502686,
      "learning_rate": 1.707692307692308e-05,
      "loss": 0.3854,
      "step": 570
    },
    {
      "epoch": 7.297124600638978,
      "grad_norm": 2.5808708667755127,
      "learning_rate": 1.7071794871794873e-05,
      "loss": 0.4507,
      "step": 571
    },
    {
      "epoch": 7.3099041533546325,
      "grad_norm": 3.1893603801727295,
      "learning_rate": 1.706666666666667e-05,
      "loss": 0.4189,
      "step": 572
    },
    {
      "epoch": 7.322683706070287,
      "grad_norm": 2.9760212898254395,
      "learning_rate": 1.706153846153846e-05,
      "loss": 0.4227,
      "step": 573
    },
    {
      "epoch": 7.335463258785943,
      "grad_norm": 1.8840742111206055,
      "learning_rate": 1.705641025641026e-05,
      "loss": 0.3461,
      "step": 574
    },
    {
      "epoch": 7.348242811501597,
      "grad_norm": 2.352691173553467,
      "learning_rate": 1.7051282051282053e-05,
      "loss": 0.4223,
      "step": 575
    },
    {
      "epoch": 7.361022364217252,
      "grad_norm": 2.5904428958892822,
      "learning_rate": 1.7046153846153847e-05,
      "loss": 0.384,
      "step": 576
    },
    {
      "epoch": 7.373801916932908,
      "grad_norm": 2.190866470336914,
      "learning_rate": 1.704102564102564e-05,
      "loss": 0.2897,
      "step": 577
    },
    {
      "epoch": 7.386581469648562,
      "grad_norm": 2.4126853942871094,
      "learning_rate": 1.7035897435897438e-05,
      "loss": 0.4496,
      "step": 578
    },
    {
      "epoch": 7.399361022364217,
      "grad_norm": 2.798590898513794,
      "learning_rate": 1.7030769230769232e-05,
      "loss": 0.432,
      "step": 579
    },
    {
      "epoch": 7.412140575079873,
      "grad_norm": 3.0337421894073486,
      "learning_rate": 1.7025641025641026e-05,
      "loss": 0.3953,
      "step": 580
    },
    {
      "epoch": 7.424920127795527,
      "grad_norm": 2.6716692447662354,
      "learning_rate": 1.7020512820512824e-05,
      "loss": 0.3673,
      "step": 581
    },
    {
      "epoch": 7.437699680511182,
      "grad_norm": 2.6297688484191895,
      "learning_rate": 1.7015384615384614e-05,
      "loss": 0.5193,
      "step": 582
    },
    {
      "epoch": 7.4504792332268375,
      "grad_norm": 3.0829811096191406,
      "learning_rate": 1.7010256410256412e-05,
      "loss": 0.5615,
      "step": 583
    },
    {
      "epoch": 7.463258785942492,
      "grad_norm": 3.34427809715271,
      "learning_rate": 1.7005128205128206e-05,
      "loss": 0.4111,
      "step": 584
    },
    {
      "epoch": 7.476038338658147,
      "grad_norm": 2.949704647064209,
      "learning_rate": 1.7e-05,
      "loss": 0.4972,
      "step": 585
    },
    {
      "epoch": 7.488817891373802,
      "grad_norm": 1.9306256771087646,
      "learning_rate": 1.6994871794871797e-05,
      "loss": 0.2485,
      "step": 586
    },
    {
      "epoch": 7.501597444089457,
      "grad_norm": 2.582099199295044,
      "learning_rate": 1.698974358974359e-05,
      "loss": 0.3855,
      "step": 587
    },
    {
      "epoch": 7.514376996805112,
      "grad_norm": 2.52205491065979,
      "learning_rate": 1.6984615384615385e-05,
      "loss": 0.3522,
      "step": 588
    },
    {
      "epoch": 7.527156549520766,
      "grad_norm": 3.256770372390747,
      "learning_rate": 1.6979487179487183e-05,
      "loss": 0.4599,
      "step": 589
    },
    {
      "epoch": 7.539936102236422,
      "grad_norm": 2.6770920753479004,
      "learning_rate": 1.6974358974358977e-05,
      "loss": 0.3897,
      "step": 590
    },
    {
      "epoch": 7.552715654952077,
      "grad_norm": 2.4803273677825928,
      "learning_rate": 1.696923076923077e-05,
      "loss": 0.4065,
      "step": 591
    },
    {
      "epoch": 7.565495207667731,
      "grad_norm": 2.484539747238159,
      "learning_rate": 1.6964102564102565e-05,
      "loss": 0.4344,
      "step": 592
    },
    {
      "epoch": 7.578274760383387,
      "grad_norm": 3.078622579574585,
      "learning_rate": 1.695897435897436e-05,
      "loss": 0.3477,
      "step": 593
    },
    {
      "epoch": 7.5910543130990416,
      "grad_norm": 2.6187045574188232,
      "learning_rate": 1.6953846153846156e-05,
      "loss": 0.4671,
      "step": 594
    },
    {
      "epoch": 7.603833865814696,
      "grad_norm": 2.429306745529175,
      "learning_rate": 1.694871794871795e-05,
      "loss": 0.4353,
      "step": 595
    },
    {
      "epoch": 7.616613418530352,
      "grad_norm": 2.7066404819488525,
      "learning_rate": 1.6943589743589745e-05,
      "loss": 0.3992,
      "step": 596
    },
    {
      "epoch": 7.6293929712460065,
      "grad_norm": 3.0803706645965576,
      "learning_rate": 1.693846153846154e-05,
      "loss": 0.5206,
      "step": 597
    },
    {
      "epoch": 7.642172523961661,
      "grad_norm": 2.8471038341522217,
      "learning_rate": 1.6933333333333336e-05,
      "loss": 0.6463,
      "step": 598
    },
    {
      "epoch": 7.654952076677317,
      "grad_norm": 2.572672128677368,
      "learning_rate": 1.692820512820513e-05,
      "loss": 0.4674,
      "step": 599
    },
    {
      "epoch": 7.667731629392971,
      "grad_norm": 2.5649611949920654,
      "learning_rate": 1.6923076923076924e-05,
      "loss": 0.4261,
      "step": 600
    },
    {
      "epoch": 7.667731629392971,
      "eval_loss": 0.4893907606601715,
      "eval_runtime": 180.8537,
      "eval_samples_per_second": 0.868,
      "eval_steps_per_second": 0.111,
      "step": 600
    },
    {
      "epoch": 7.680511182108626,
      "grad_norm": 2.6403310298919678,
      "learning_rate": 1.691794871794872e-05,
      "loss": 0.5065,
      "step": 601
    },
    {
      "epoch": 7.693290734824281,
      "grad_norm": 2.59029221534729,
      "learning_rate": 1.6912820512820512e-05,
      "loss": 0.4302,
      "step": 602
    },
    {
      "epoch": 7.706070287539936,
      "grad_norm": 2.2685883045196533,
      "learning_rate": 1.690769230769231e-05,
      "loss": 0.3629,
      "step": 603
    },
    {
      "epoch": 7.718849840255591,
      "grad_norm": 2.280118227005005,
      "learning_rate": 1.6902564102564104e-05,
      "loss": 0.5129,
      "step": 604
    },
    {
      "epoch": 7.731629392971246,
      "grad_norm": 2.507329225540161,
      "learning_rate": 1.6897435897435898e-05,
      "loss": 0.4688,
      "step": 605
    },
    {
      "epoch": 7.744408945686901,
      "grad_norm": 2.897050380706787,
      "learning_rate": 1.6892307692307695e-05,
      "loss": 0.5061,
      "step": 606
    },
    {
      "epoch": 7.757188498402556,
      "grad_norm": 2.5373477935791016,
      "learning_rate": 1.688717948717949e-05,
      "loss": 0.4823,
      "step": 607
    },
    {
      "epoch": 7.7699680511182105,
      "grad_norm": 3.002852201461792,
      "learning_rate": 1.6882051282051283e-05,
      "loss": 0.4908,
      "step": 608
    },
    {
      "epoch": 7.782747603833866,
      "grad_norm": 2.089543104171753,
      "learning_rate": 1.6876923076923077e-05,
      "loss": 0.4308,
      "step": 609
    },
    {
      "epoch": 7.795527156549521,
      "grad_norm": 2.6164510250091553,
      "learning_rate": 1.687179487179487e-05,
      "loss": 0.5141,
      "step": 610
    },
    {
      "epoch": 7.8083067092651754,
      "grad_norm": 2.691242218017578,
      "learning_rate": 1.686666666666667e-05,
      "loss": 0.4898,
      "step": 611
    },
    {
      "epoch": 7.821086261980831,
      "grad_norm": 2.664036273956299,
      "learning_rate": 1.6861538461538463e-05,
      "loss": 0.4248,
      "step": 612
    },
    {
      "epoch": 7.833865814696486,
      "grad_norm": 1.9409226179122925,
      "learning_rate": 1.6856410256410257e-05,
      "loss": 0.3507,
      "step": 613
    },
    {
      "epoch": 7.84664536741214,
      "grad_norm": 3.2386646270751953,
      "learning_rate": 1.685128205128205e-05,
      "loss": 0.3971,
      "step": 614
    },
    {
      "epoch": 7.859424920127795,
      "grad_norm": 2.29748797416687,
      "learning_rate": 1.684615384615385e-05,
      "loss": 0.3966,
      "step": 615
    },
    {
      "epoch": 7.872204472843451,
      "grad_norm": 2.7634127140045166,
      "learning_rate": 1.6841025641025643e-05,
      "loss": 0.4292,
      "step": 616
    },
    {
      "epoch": 7.884984025559105,
      "grad_norm": 2.2984588146209717,
      "learning_rate": 1.6835897435897437e-05,
      "loss": 0.3906,
      "step": 617
    },
    {
      "epoch": 7.897763578274761,
      "grad_norm": 2.8001909255981445,
      "learning_rate": 1.6830769230769234e-05,
      "loss": 0.6094,
      "step": 618
    },
    {
      "epoch": 7.9105431309904155,
      "grad_norm": 2.892528533935547,
      "learning_rate": 1.6825641025641025e-05,
      "loss": 0.4006,
      "step": 619
    },
    {
      "epoch": 7.92332268370607,
      "grad_norm": 3.403233051300049,
      "learning_rate": 1.6820512820512822e-05,
      "loss": 0.4464,
      "step": 620
    },
    {
      "epoch": 7.936102236421725,
      "grad_norm": 2.3043155670166016,
      "learning_rate": 1.6815384615384616e-05,
      "loss": 0.3331,
      "step": 621
    },
    {
      "epoch": 7.94888178913738,
      "grad_norm": 2.8333663940429688,
      "learning_rate": 1.681025641025641e-05,
      "loss": 0.415,
      "step": 622
    },
    {
      "epoch": 7.961661341853035,
      "grad_norm": 2.719902992248535,
      "learning_rate": 1.6805128205128208e-05,
      "loss": 0.5258,
      "step": 623
    },
    {
      "epoch": 7.97444089456869,
      "grad_norm": 3.271904468536377,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.4089,
      "step": 624
    },
    {
      "epoch": 7.987220447284345,
      "grad_norm": 2.5888147354125977,
      "learning_rate": 1.6794871794871796e-05,
      "loss": 0.382,
      "step": 625
    },
    {
      "epoch": 8.0,
      "grad_norm": 3.209883213043213,
      "learning_rate": 1.678974358974359e-05,
      "loss": 0.6444,
      "step": 626
    },
    {
      "epoch": 8.012779552715655,
      "grad_norm": 2.370129346847534,
      "learning_rate": 1.6784615384615387e-05,
      "loss": 0.5045,
      "step": 627
    },
    {
      "epoch": 8.02555910543131,
      "grad_norm": 2.871953010559082,
      "learning_rate": 1.677948717948718e-05,
      "loss": 0.5227,
      "step": 628
    },
    {
      "epoch": 8.038338658146964,
      "grad_norm": 2.793961763381958,
      "learning_rate": 1.6774358974358975e-05,
      "loss": 0.4616,
      "step": 629
    },
    {
      "epoch": 8.05111821086262,
      "grad_norm": 2.622335910797119,
      "learning_rate": 1.676923076923077e-05,
      "loss": 0.4296,
      "step": 630
    },
    {
      "epoch": 8.063897763578275,
      "grad_norm": 2.6651382446289062,
      "learning_rate": 1.6764102564102564e-05,
      "loss": 0.391,
      "step": 631
    },
    {
      "epoch": 8.07667731629393,
      "grad_norm": 2.927123546600342,
      "learning_rate": 1.675897435897436e-05,
      "loss": 0.3775,
      "step": 632
    },
    {
      "epoch": 8.089456869009584,
      "grad_norm": 2.3731541633605957,
      "learning_rate": 1.6753846153846155e-05,
      "loss": 0.3785,
      "step": 633
    },
    {
      "epoch": 8.10223642172524,
      "grad_norm": 2.4673285484313965,
      "learning_rate": 1.674871794871795e-05,
      "loss": 0.4422,
      "step": 634
    },
    {
      "epoch": 8.115015974440894,
      "grad_norm": 2.902761459350586,
      "learning_rate": 1.6743589743589747e-05,
      "loss": 0.3814,
      "step": 635
    },
    {
      "epoch": 8.12779552715655,
      "grad_norm": 3.2571306228637695,
      "learning_rate": 1.673846153846154e-05,
      "loss": 0.443,
      "step": 636
    },
    {
      "epoch": 8.140575079872205,
      "grad_norm": 3.122844934463501,
      "learning_rate": 1.6733333333333335e-05,
      "loss": 0.4561,
      "step": 637
    },
    {
      "epoch": 8.15335463258786,
      "grad_norm": 2.018770456314087,
      "learning_rate": 1.6728205128205132e-05,
      "loss": 0.1985,
      "step": 638
    },
    {
      "epoch": 8.166134185303514,
      "grad_norm": 2.4061696529388428,
      "learning_rate": 1.6723076923076923e-05,
      "loss": 0.37,
      "step": 639
    },
    {
      "epoch": 8.178913738019169,
      "grad_norm": 2.3851146697998047,
      "learning_rate": 1.671794871794872e-05,
      "loss": 0.3915,
      "step": 640
    },
    {
      "epoch": 8.191693290734824,
      "grad_norm": 3.250897169113159,
      "learning_rate": 1.6712820512820514e-05,
      "loss": 0.4104,
      "step": 641
    },
    {
      "epoch": 8.204472843450478,
      "grad_norm": 2.8355727195739746,
      "learning_rate": 1.6707692307692308e-05,
      "loss": 0.3437,
      "step": 642
    },
    {
      "epoch": 8.217252396166135,
      "grad_norm": 2.1854217052459717,
      "learning_rate": 1.6702564102564106e-05,
      "loss": 0.3172,
      "step": 643
    },
    {
      "epoch": 8.23003194888179,
      "grad_norm": 4.091235160827637,
      "learning_rate": 1.66974358974359e-05,
      "loss": 0.3752,
      "step": 644
    },
    {
      "epoch": 8.242811501597444,
      "grad_norm": 3.4227828979492188,
      "learning_rate": 1.6692307692307694e-05,
      "loss": 0.4983,
      "step": 645
    },
    {
      "epoch": 8.255591054313099,
      "grad_norm": 2.888979911804199,
      "learning_rate": 1.6687179487179488e-05,
      "loss": 0.5062,
      "step": 646
    },
    {
      "epoch": 8.268370607028753,
      "grad_norm": 2.7345168590545654,
      "learning_rate": 1.6682051282051285e-05,
      "loss": 0.3431,
      "step": 647
    },
    {
      "epoch": 8.281150159744408,
      "grad_norm": 3.435631036758423,
      "learning_rate": 1.6676923076923076e-05,
      "loss": 0.4714,
      "step": 648
    },
    {
      "epoch": 8.293929712460065,
      "grad_norm": 2.1894478797912598,
      "learning_rate": 1.6671794871794873e-05,
      "loss": 0.2558,
      "step": 649
    },
    {
      "epoch": 8.30670926517572,
      "grad_norm": 3.0066120624542236,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.3377,
      "step": 650
    },
    {
      "epoch": 8.319488817891374,
      "grad_norm": 2.613056182861328,
      "learning_rate": 1.666153846153846e-05,
      "loss": 0.3492,
      "step": 651
    },
    {
      "epoch": 8.332268370607029,
      "grad_norm": 2.5527093410491943,
      "learning_rate": 1.665641025641026e-05,
      "loss": 0.3209,
      "step": 652
    },
    {
      "epoch": 8.345047923322683,
      "grad_norm": 2.7275326251983643,
      "learning_rate": 1.6651282051282053e-05,
      "loss": 0.3943,
      "step": 653
    },
    {
      "epoch": 8.357827476038338,
      "grad_norm": 2.0672011375427246,
      "learning_rate": 1.6646153846153847e-05,
      "loss": 0.3674,
      "step": 654
    },
    {
      "epoch": 8.370607028753994,
      "grad_norm": 2.8069820404052734,
      "learning_rate": 1.6641025641025645e-05,
      "loss": 0.3685,
      "step": 655
    },
    {
      "epoch": 8.383386581469649,
      "grad_norm": 3.1376593112945557,
      "learning_rate": 1.6635897435897435e-05,
      "loss": 0.5658,
      "step": 656
    },
    {
      "epoch": 8.396166134185304,
      "grad_norm": 2.542144775390625,
      "learning_rate": 1.6630769230769233e-05,
      "loss": 0.3924,
      "step": 657
    },
    {
      "epoch": 8.408945686900958,
      "grad_norm": 2.6188881397247314,
      "learning_rate": 1.6625641025641027e-05,
      "loss": 0.4069,
      "step": 658
    },
    {
      "epoch": 8.421725239616613,
      "grad_norm": 2.8141064643859863,
      "learning_rate": 1.662051282051282e-05,
      "loss": 0.5237,
      "step": 659
    },
    {
      "epoch": 8.434504792332268,
      "grad_norm": 2.6490237712860107,
      "learning_rate": 1.6615384615384618e-05,
      "loss": 0.4192,
      "step": 660
    },
    {
      "epoch": 8.447284345047922,
      "grad_norm": 2.3943517208099365,
      "learning_rate": 1.6610256410256412e-05,
      "loss": 0.3519,
      "step": 661
    },
    {
      "epoch": 8.460063897763579,
      "grad_norm": 2.4590401649475098,
      "learning_rate": 1.6605128205128206e-05,
      "loss": 0.3297,
      "step": 662
    },
    {
      "epoch": 8.472843450479234,
      "grad_norm": 2.6325976848602295,
      "learning_rate": 1.66e-05,
      "loss": 0.3053,
      "step": 663
    },
    {
      "epoch": 8.485623003194888,
      "grad_norm": 3.596081495285034,
      "learning_rate": 1.6594871794871798e-05,
      "loss": 0.4897,
      "step": 664
    },
    {
      "epoch": 8.498402555910543,
      "grad_norm": 2.748485803604126,
      "learning_rate": 1.6589743589743592e-05,
      "loss": 0.4362,
      "step": 665
    },
    {
      "epoch": 8.511182108626198,
      "grad_norm": 3.381138563156128,
      "learning_rate": 1.6584615384615386e-05,
      "loss": 0.4669,
      "step": 666
    },
    {
      "epoch": 8.523961661341852,
      "grad_norm": 2.4086904525756836,
      "learning_rate": 1.657948717948718e-05,
      "loss": 0.4312,
      "step": 667
    },
    {
      "epoch": 8.536741214057509,
      "grad_norm": 2.4660682678222656,
      "learning_rate": 1.6574358974358974e-05,
      "loss": 0.4272,
      "step": 668
    },
    {
      "epoch": 8.549520766773163,
      "grad_norm": 2.776521682739258,
      "learning_rate": 1.656923076923077e-05,
      "loss": 0.4104,
      "step": 669
    },
    {
      "epoch": 8.562300319488818,
      "grad_norm": 3.0339174270629883,
      "learning_rate": 1.6564102564102565e-05,
      "loss": 0.443,
      "step": 670
    },
    {
      "epoch": 8.575079872204473,
      "grad_norm": 2.8308653831481934,
      "learning_rate": 1.655897435897436e-05,
      "loss": 0.3196,
      "step": 671
    },
    {
      "epoch": 8.587859424920127,
      "grad_norm": 2.376342296600342,
      "learning_rate": 1.6553846153846157e-05,
      "loss": 0.3005,
      "step": 672
    },
    {
      "epoch": 8.600638977635782,
      "grad_norm": 2.4938461780548096,
      "learning_rate": 1.654871794871795e-05,
      "loss": 0.3203,
      "step": 673
    },
    {
      "epoch": 8.613418530351439,
      "grad_norm": 2.8383700847625732,
      "learning_rate": 1.6543589743589745e-05,
      "loss": 0.5523,
      "step": 674
    },
    {
      "epoch": 8.626198083067093,
      "grad_norm": 2.876192331314087,
      "learning_rate": 1.653846153846154e-05,
      "loss": 0.4058,
      "step": 675
    },
    {
      "epoch": 8.638977635782748,
      "grad_norm": 2.5752999782562256,
      "learning_rate": 1.6533333333333333e-05,
      "loss": 0.5115,
      "step": 676
    },
    {
      "epoch": 8.651757188498403,
      "grad_norm": 2.703507661819458,
      "learning_rate": 1.652820512820513e-05,
      "loss": 0.4486,
      "step": 677
    },
    {
      "epoch": 8.664536741214057,
      "grad_norm": 3.2754502296447754,
      "learning_rate": 1.6523076923076925e-05,
      "loss": 0.3713,
      "step": 678
    },
    {
      "epoch": 8.677316293929712,
      "grad_norm": 3.1390693187713623,
      "learning_rate": 1.651794871794872e-05,
      "loss": 0.4578,
      "step": 679
    },
    {
      "epoch": 8.690095846645367,
      "grad_norm": 2.548980236053467,
      "learning_rate": 1.6512820512820513e-05,
      "loss": 0.38,
      "step": 680
    },
    {
      "epoch": 8.702875399361023,
      "grad_norm": 2.7647883892059326,
      "learning_rate": 1.650769230769231e-05,
      "loss": 0.4191,
      "step": 681
    },
    {
      "epoch": 8.715654952076678,
      "grad_norm": 2.330361843109131,
      "learning_rate": 1.6502564102564104e-05,
      "loss": 0.5003,
      "step": 682
    },
    {
      "epoch": 8.728434504792332,
      "grad_norm": 2.790050745010376,
      "learning_rate": 1.6497435897435898e-05,
      "loss": 0.3901,
      "step": 683
    },
    {
      "epoch": 8.741214057507987,
      "grad_norm": 3.12176775932312,
      "learning_rate": 1.6492307692307696e-05,
      "loss": 0.3778,
      "step": 684
    },
    {
      "epoch": 8.753993610223642,
      "grad_norm": 2.1713685989379883,
      "learning_rate": 1.6487179487179486e-05,
      "loss": 0.3297,
      "step": 685
    },
    {
      "epoch": 8.766773162939296,
      "grad_norm": 2.4351470470428467,
      "learning_rate": 1.6482051282051284e-05,
      "loss": 0.3086,
      "step": 686
    },
    {
      "epoch": 8.779552715654953,
      "grad_norm": 2.7927756309509277,
      "learning_rate": 1.6476923076923078e-05,
      "loss": 0.3062,
      "step": 687
    },
    {
      "epoch": 8.792332268370608,
      "grad_norm": 3.960897922515869,
      "learning_rate": 1.6471794871794872e-05,
      "loss": 0.4952,
      "step": 688
    },
    {
      "epoch": 8.805111821086262,
      "grad_norm": 2.980947971343994,
      "learning_rate": 1.646666666666667e-05,
      "loss": 0.4061,
      "step": 689
    },
    {
      "epoch": 8.817891373801917,
      "grad_norm": 3.104675769805908,
      "learning_rate": 1.6461538461538463e-05,
      "loss": 0.4165,
      "step": 690
    },
    {
      "epoch": 8.830670926517572,
      "grad_norm": 2.799107074737549,
      "learning_rate": 1.6456410256410257e-05,
      "loss": 0.3603,
      "step": 691
    },
    {
      "epoch": 8.843450479233226,
      "grad_norm": 2.9702889919281006,
      "learning_rate": 1.6451282051282055e-05,
      "loss": 0.4726,
      "step": 692
    },
    {
      "epoch": 8.856230031948883,
      "grad_norm": 3.655086040496826,
      "learning_rate": 1.6446153846153846e-05,
      "loss": 0.491,
      "step": 693
    },
    {
      "epoch": 8.869009584664537,
      "grad_norm": 3.1771860122680664,
      "learning_rate": 1.6441025641025643e-05,
      "loss": 0.3899,
      "step": 694
    },
    {
      "epoch": 8.881789137380192,
      "grad_norm": 2.3967061042785645,
      "learning_rate": 1.6435897435897437e-05,
      "loss": 0.4936,
      "step": 695
    },
    {
      "epoch": 8.894568690095847,
      "grad_norm": 3.973982095718384,
      "learning_rate": 1.643076923076923e-05,
      "loss": 0.4266,
      "step": 696
    },
    {
      "epoch": 8.907348242811501,
      "grad_norm": 2.963104248046875,
      "learning_rate": 1.6425641025641025e-05,
      "loss": 0.3856,
      "step": 697
    },
    {
      "epoch": 8.920127795527156,
      "grad_norm": 2.7379801273345947,
      "learning_rate": 1.6420512820512823e-05,
      "loss": 0.3364,
      "step": 698
    },
    {
      "epoch": 8.93290734824281,
      "grad_norm": 3.148500442504883,
      "learning_rate": 1.6415384615384617e-05,
      "loss": 0.3674,
      "step": 699
    },
    {
      "epoch": 8.945686900958467,
      "grad_norm": 2.757758140563965,
      "learning_rate": 1.641025641025641e-05,
      "loss": 0.406,
      "step": 700
    },
    {
      "epoch": 8.945686900958467,
      "eval_loss": 0.4816867411136627,
      "eval_runtime": 180.9578,
      "eval_samples_per_second": 0.868,
      "eval_steps_per_second": 0.111,
      "step": 700
    },
    {
      "epoch": 8.958466453674122,
      "grad_norm": 2.902454376220703,
      "learning_rate": 1.6405128205128208e-05,
      "loss": 0.4216,
      "step": 701
    },
    {
      "epoch": 8.971246006389777,
      "grad_norm": 2.8162240982055664,
      "learning_rate": 1.64e-05,
      "loss": 0.3975,
      "step": 702
    },
    {
      "epoch": 8.984025559105431,
      "grad_norm": 2.35843563079834,
      "learning_rate": 1.6394871794871796e-05,
      "loss": 0.4158,
      "step": 703
    },
    {
      "epoch": 8.996805111821086,
      "grad_norm": 3.153388023376465,
      "learning_rate": 1.638974358974359e-05,
      "loss": 0.5259,
      "step": 704
    },
    {
      "epoch": 9.00958466453674,
      "grad_norm": 2.3435404300689697,
      "learning_rate": 1.6384615384615384e-05,
      "loss": 0.285,
      "step": 705
    },
    {
      "epoch": 9.022364217252397,
      "grad_norm": 2.2724061012268066,
      "learning_rate": 1.6379487179487182e-05,
      "loss": 0.4115,
      "step": 706
    },
    {
      "epoch": 9.035143769968052,
      "grad_norm": 3.4185988903045654,
      "learning_rate": 1.6374358974358976e-05,
      "loss": 0.3614,
      "step": 707
    },
    {
      "epoch": 9.047923322683706,
      "grad_norm": 2.1561129093170166,
      "learning_rate": 1.636923076923077e-05,
      "loss": 0.314,
      "step": 708
    },
    {
      "epoch": 9.060702875399361,
      "grad_norm": 2.4209418296813965,
      "learning_rate": 1.6364102564102567e-05,
      "loss": 0.284,
      "step": 709
    },
    {
      "epoch": 9.073482428115016,
      "grad_norm": 2.662139892578125,
      "learning_rate": 1.635897435897436e-05,
      "loss": 0.5409,
      "step": 710
    },
    {
      "epoch": 9.08626198083067,
      "grad_norm": 2.8799054622650146,
      "learning_rate": 1.6353846153846155e-05,
      "loss": 0.3771,
      "step": 711
    },
    {
      "epoch": 9.099041533546325,
      "grad_norm": 2.704667806625366,
      "learning_rate": 1.634871794871795e-05,
      "loss": 0.3255,
      "step": 712
    },
    {
      "epoch": 9.111821086261982,
      "grad_norm": 2.934540033340454,
      "learning_rate": 1.6343589743589744e-05,
      "loss": 0.4394,
      "step": 713
    },
    {
      "epoch": 9.124600638977636,
      "grad_norm": 3.1625423431396484,
      "learning_rate": 1.633846153846154e-05,
      "loss": 0.3893,
      "step": 714
    },
    {
      "epoch": 9.13738019169329,
      "grad_norm": 3.0641441345214844,
      "learning_rate": 1.6333333333333335e-05,
      "loss": 0.375,
      "step": 715
    },
    {
      "epoch": 9.150159744408946,
      "grad_norm": 3.011550188064575,
      "learning_rate": 1.632820512820513e-05,
      "loss": 0.3112,
      "step": 716
    },
    {
      "epoch": 9.1629392971246,
      "grad_norm": 2.9342551231384277,
      "learning_rate": 1.6323076923076923e-05,
      "loss": 0.4093,
      "step": 717
    },
    {
      "epoch": 9.175718849840255,
      "grad_norm": 2.970946788787842,
      "learning_rate": 1.631794871794872e-05,
      "loss": 0.4874,
      "step": 718
    },
    {
      "epoch": 9.188498402555911,
      "grad_norm": 2.548109531402588,
      "learning_rate": 1.6312820512820515e-05,
      "loss": 0.4156,
      "step": 719
    },
    {
      "epoch": 9.201277955271566,
      "grad_norm": 3.7871553897857666,
      "learning_rate": 1.630769230769231e-05,
      "loss": 0.3074,
      "step": 720
    },
    {
      "epoch": 9.21405750798722,
      "grad_norm": 2.4763805866241455,
      "learning_rate": 1.6302564102564106e-05,
      "loss": 0.331,
      "step": 721
    },
    {
      "epoch": 9.226837060702875,
      "grad_norm": 3.3531510829925537,
      "learning_rate": 1.6297435897435897e-05,
      "loss": 0.498,
      "step": 722
    },
    {
      "epoch": 9.23961661341853,
      "grad_norm": 2.8672120571136475,
      "learning_rate": 1.6292307692307694e-05,
      "loss": 0.3971,
      "step": 723
    },
    {
      "epoch": 9.252396166134185,
      "grad_norm": 3.8358211517333984,
      "learning_rate": 1.628717948717949e-05,
      "loss": 0.3478,
      "step": 724
    },
    {
      "epoch": 9.26517571884984,
      "grad_norm": 3.4077041149139404,
      "learning_rate": 1.6282051282051282e-05,
      "loss": 0.3313,
      "step": 725
    },
    {
      "epoch": 9.277955271565496,
      "grad_norm": 3.170940637588501,
      "learning_rate": 1.627692307692308e-05,
      "loss": 0.5251,
      "step": 726
    },
    {
      "epoch": 9.29073482428115,
      "grad_norm": 2.9623332023620605,
      "learning_rate": 1.6271794871794874e-05,
      "loss": 0.5047,
      "step": 727
    },
    {
      "epoch": 9.303514376996805,
      "grad_norm": 3.793362855911255,
      "learning_rate": 1.6266666666666668e-05,
      "loss": 0.5395,
      "step": 728
    },
    {
      "epoch": 9.31629392971246,
      "grad_norm": 2.935246467590332,
      "learning_rate": 1.6261538461538462e-05,
      "loss": 0.4144,
      "step": 729
    },
    {
      "epoch": 9.329073482428115,
      "grad_norm": 2.9262139797210693,
      "learning_rate": 1.625641025641026e-05,
      "loss": 0.4187,
      "step": 730
    },
    {
      "epoch": 9.34185303514377,
      "grad_norm": 2.8220555782318115,
      "learning_rate": 1.6251282051282053e-05,
      "loss": 0.4186,
      "step": 731
    },
    {
      "epoch": 9.354632587859426,
      "grad_norm": 4.042216777801514,
      "learning_rate": 1.6246153846153848e-05,
      "loss": 0.3774,
      "step": 732
    },
    {
      "epoch": 9.36741214057508,
      "grad_norm": 3.7640748023986816,
      "learning_rate": 1.624102564102564e-05,
      "loss": 0.3001,
      "step": 733
    },
    {
      "epoch": 9.380191693290735,
      "grad_norm": 2.9387130737304688,
      "learning_rate": 1.6235897435897436e-05,
      "loss": 0.3889,
      "step": 734
    },
    {
      "epoch": 9.39297124600639,
      "grad_norm": 3.4702775478363037,
      "learning_rate": 1.6230769230769233e-05,
      "loss": 0.4282,
      "step": 735
    },
    {
      "epoch": 9.405750798722044,
      "grad_norm": 3.894277572631836,
      "learning_rate": 1.6225641025641027e-05,
      "loss": 0.5151,
      "step": 736
    },
    {
      "epoch": 9.418530351437699,
      "grad_norm": 2.918215751647949,
      "learning_rate": 1.622051282051282e-05,
      "loss": 0.4282,
      "step": 737
    },
    {
      "epoch": 9.431309904153355,
      "grad_norm": 3.2780513763427734,
      "learning_rate": 1.621538461538462e-05,
      "loss": 0.3878,
      "step": 738
    },
    {
      "epoch": 9.44408945686901,
      "grad_norm": 2.4160218238830566,
      "learning_rate": 1.621025641025641e-05,
      "loss": 0.3296,
      "step": 739
    },
    {
      "epoch": 9.456869009584665,
      "grad_norm": 2.724004030227661,
      "learning_rate": 1.6205128205128207e-05,
      "loss": 0.3507,
      "step": 740
    },
    {
      "epoch": 9.46964856230032,
      "grad_norm": 3.2544777393341064,
      "learning_rate": 1.62e-05,
      "loss": 0.5059,
      "step": 741
    },
    {
      "epoch": 9.482428115015974,
      "grad_norm": 2.500000238418579,
      "learning_rate": 1.6194871794871795e-05,
      "loss": 0.3442,
      "step": 742
    },
    {
      "epoch": 9.495207667731629,
      "grad_norm": 3.0941834449768066,
      "learning_rate": 1.6189743589743592e-05,
      "loss": 0.3219,
      "step": 743
    },
    {
      "epoch": 9.507987220447284,
      "grad_norm": 2.9002368450164795,
      "learning_rate": 1.6184615384615386e-05,
      "loss": 0.3519,
      "step": 744
    },
    {
      "epoch": 9.52076677316294,
      "grad_norm": 2.824436664581299,
      "learning_rate": 1.617948717948718e-05,
      "loss": 0.3762,
      "step": 745
    },
    {
      "epoch": 9.533546325878595,
      "grad_norm": 2.759906530380249,
      "learning_rate": 1.6174358974358974e-05,
      "loss": 0.4067,
      "step": 746
    },
    {
      "epoch": 9.54632587859425,
      "grad_norm": 2.4780590534210205,
      "learning_rate": 1.6169230769230772e-05,
      "loss": 0.3595,
      "step": 747
    },
    {
      "epoch": 9.559105431309904,
      "grad_norm": 2.7761125564575195,
      "learning_rate": 1.6164102564102566e-05,
      "loss": 0.3456,
      "step": 748
    },
    {
      "epoch": 9.571884984025559,
      "grad_norm": 4.0381951332092285,
      "learning_rate": 1.615897435897436e-05,
      "loss": 0.3862,
      "step": 749
    },
    {
      "epoch": 9.584664536741213,
      "grad_norm": 3.450500726699829,
      "learning_rate": 1.6153846153846154e-05,
      "loss": 0.3582,
      "step": 750
    },
    {
      "epoch": 9.59744408945687,
      "grad_norm": 2.6575820446014404,
      "learning_rate": 1.6148717948717948e-05,
      "loss": 0.3178,
      "step": 751
    },
    {
      "epoch": 9.610223642172524,
      "grad_norm": 2.5573747158050537,
      "learning_rate": 1.6143589743589745e-05,
      "loss": 0.3958,
      "step": 752
    },
    {
      "epoch": 9.62300319488818,
      "grad_norm": 2.849857807159424,
      "learning_rate": 1.613846153846154e-05,
      "loss": 0.3588,
      "step": 753
    },
    {
      "epoch": 9.635782747603834,
      "grad_norm": 2.322702646255493,
      "learning_rate": 1.6133333333333334e-05,
      "loss": 0.3252,
      "step": 754
    },
    {
      "epoch": 9.648562300319488,
      "grad_norm": 2.626924753189087,
      "learning_rate": 1.612820512820513e-05,
      "loss": 0.3818,
      "step": 755
    },
    {
      "epoch": 9.661341853035143,
      "grad_norm": 2.556973934173584,
      "learning_rate": 1.6123076923076925e-05,
      "loss": 0.2887,
      "step": 756
    },
    {
      "epoch": 9.6741214057508,
      "grad_norm": 3.3718507289886475,
      "learning_rate": 1.611794871794872e-05,
      "loss": 0.4669,
      "step": 757
    },
    {
      "epoch": 9.686900958466454,
      "grad_norm": 2.6675846576690674,
      "learning_rate": 1.6112820512820517e-05,
      "loss": 0.3147,
      "step": 758
    },
    {
      "epoch": 9.699680511182109,
      "grad_norm": 2.686455011367798,
      "learning_rate": 1.6107692307692307e-05,
      "loss": 0.3977,
      "step": 759
    },
    {
      "epoch": 9.712460063897764,
      "grad_norm": 3.091524839401245,
      "learning_rate": 1.6102564102564105e-05,
      "loss": 0.4134,
      "step": 760
    },
    {
      "epoch": 9.725239616613418,
      "grad_norm": 3.1435928344726562,
      "learning_rate": 1.60974358974359e-05,
      "loss": 0.4106,
      "step": 761
    },
    {
      "epoch": 9.738019169329073,
      "grad_norm": 3.289584159851074,
      "learning_rate": 1.6092307692307693e-05,
      "loss": 0.3564,
      "step": 762
    },
    {
      "epoch": 9.750798722044728,
      "grad_norm": 3.0257647037506104,
      "learning_rate": 1.608717948717949e-05,
      "loss": 0.405,
      "step": 763
    },
    {
      "epoch": 9.763578274760384,
      "grad_norm": 3.0910072326660156,
      "learning_rate": 1.6082051282051284e-05,
      "loss": 0.3332,
      "step": 764
    },
    {
      "epoch": 9.776357827476039,
      "grad_norm": 3.3785085678100586,
      "learning_rate": 1.607692307692308e-05,
      "loss": 0.3729,
      "step": 765
    },
    {
      "epoch": 9.789137380191693,
      "grad_norm": 3.3935513496398926,
      "learning_rate": 1.6071794871794872e-05,
      "loss": 0.5261,
      "step": 766
    },
    {
      "epoch": 9.801916932907348,
      "grad_norm": 3.9248788356781006,
      "learning_rate": 1.606666666666667e-05,
      "loss": 0.3136,
      "step": 767
    },
    {
      "epoch": 9.814696485623003,
      "grad_norm": 3.094261884689331,
      "learning_rate": 1.606153846153846e-05,
      "loss": 0.3301,
      "step": 768
    },
    {
      "epoch": 9.827476038338657,
      "grad_norm": 4.1841254234313965,
      "learning_rate": 1.6056410256410258e-05,
      "loss": 0.4053,
      "step": 769
    },
    {
      "epoch": 9.840255591054314,
      "grad_norm": 2.513289213180542,
      "learning_rate": 1.6051282051282052e-05,
      "loss": 0.3944,
      "step": 770
    },
    {
      "epoch": 9.853035143769969,
      "grad_norm": 3.5416557788848877,
      "learning_rate": 1.6046153846153846e-05,
      "loss": 0.3499,
      "step": 771
    },
    {
      "epoch": 9.865814696485623,
      "grad_norm": 2.6167125701904297,
      "learning_rate": 1.6041025641025643e-05,
      "loss": 0.3378,
      "step": 772
    },
    {
      "epoch": 9.878594249201278,
      "grad_norm": 2.5471789836883545,
      "learning_rate": 1.6035897435897438e-05,
      "loss": 0.346,
      "step": 773
    },
    {
      "epoch": 9.891373801916933,
      "grad_norm": 3.2733585834503174,
      "learning_rate": 1.603076923076923e-05,
      "loss": 0.4733,
      "step": 774
    },
    {
      "epoch": 9.904153354632587,
      "grad_norm": 3.283470392227173,
      "learning_rate": 1.602564102564103e-05,
      "loss": 0.3673,
      "step": 775
    },
    {
      "epoch": 9.916932907348244,
      "grad_norm": 2.748051643371582,
      "learning_rate": 1.602051282051282e-05,
      "loss": 0.3411,
      "step": 776
    },
    {
      "epoch": 9.929712460063898,
      "grad_norm": 3.7860305309295654,
      "learning_rate": 1.6015384615384617e-05,
      "loss": 0.3744,
      "step": 777
    },
    {
      "epoch": 9.942492012779553,
      "grad_norm": 2.9770305156707764,
      "learning_rate": 1.601025641025641e-05,
      "loss": 0.3485,
      "step": 778
    },
    {
      "epoch": 9.955271565495208,
      "grad_norm": 2.8799893856048584,
      "learning_rate": 1.6005128205128205e-05,
      "loss": 0.4349,
      "step": 779
    },
    {
      "epoch": 9.968051118210862,
      "grad_norm": 2.887939214706421,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.3541,
      "step": 780
    },
    {
      "epoch": 9.980830670926517,
      "grad_norm": 3.629877805709839,
      "learning_rate": 1.5994871794871797e-05,
      "loss": 0.4275,
      "step": 781
    },
    {
      "epoch": 9.993610223642172,
      "grad_norm": 3.523859739303589,
      "learning_rate": 1.598974358974359e-05,
      "loss": 0.3754,
      "step": 782
    },
    {
      "epoch": 10.006389776357828,
      "grad_norm": 3.029161214828491,
      "learning_rate": 1.5984615384615385e-05,
      "loss": 0.3925,
      "step": 783
    },
    {
      "epoch": 10.019169329073483,
      "grad_norm": 2.83622670173645,
      "learning_rate": 1.5979487179487182e-05,
      "loss": 0.3667,
      "step": 784
    },
    {
      "epoch": 10.031948881789138,
      "grad_norm": 3.071251630783081,
      "learning_rate": 1.5974358974358976e-05,
      "loss": 0.4086,
      "step": 785
    },
    {
      "epoch": 10.044728434504792,
      "grad_norm": 2.993438959121704,
      "learning_rate": 1.596923076923077e-05,
      "loss": 0.4054,
      "step": 786
    },
    {
      "epoch": 10.057507987220447,
      "grad_norm": 2.8429837226867676,
      "learning_rate": 1.5964102564102564e-05,
      "loss": 0.2923,
      "step": 787
    },
    {
      "epoch": 10.070287539936102,
      "grad_norm": 2.995058059692383,
      "learning_rate": 1.595897435897436e-05,
      "loss": 0.4094,
      "step": 788
    },
    {
      "epoch": 10.083067092651758,
      "grad_norm": 2.814988374710083,
      "learning_rate": 1.5953846153846156e-05,
      "loss": 0.2915,
      "step": 789
    },
    {
      "epoch": 10.095846645367413,
      "grad_norm": 2.83194637298584,
      "learning_rate": 1.594871794871795e-05,
      "loss": 0.2686,
      "step": 790
    },
    {
      "epoch": 10.108626198083067,
      "grad_norm": 2.4209916591644287,
      "learning_rate": 1.5943589743589744e-05,
      "loss": 0.208,
      "step": 791
    },
    {
      "epoch": 10.121405750798722,
      "grad_norm": 2.3989953994750977,
      "learning_rate": 1.593846153846154e-05,
      "loss": 0.2616,
      "step": 792
    },
    {
      "epoch": 10.134185303514377,
      "grad_norm": 3.202815294265747,
      "learning_rate": 1.5933333333333336e-05,
      "loss": 0.3878,
      "step": 793
    },
    {
      "epoch": 10.146964856230031,
      "grad_norm": 3.395106077194214,
      "learning_rate": 1.592820512820513e-05,
      "loss": 0.3824,
      "step": 794
    },
    {
      "epoch": 10.159744408945686,
      "grad_norm": 2.954267740249634,
      "learning_rate": 1.5923076923076924e-05,
      "loss": 0.2722,
      "step": 795
    },
    {
      "epoch": 10.172523961661343,
      "grad_norm": 2.9634196758270264,
      "learning_rate": 1.5917948717948718e-05,
      "loss": 0.4102,
      "step": 796
    },
    {
      "epoch": 10.185303514376997,
      "grad_norm": 3.2138140201568604,
      "learning_rate": 1.5912820512820515e-05,
      "loss": 0.5894,
      "step": 797
    },
    {
      "epoch": 10.198083067092652,
      "grad_norm": 3.678295850753784,
      "learning_rate": 1.590769230769231e-05,
      "loss": 0.4693,
      "step": 798
    },
    {
      "epoch": 10.210862619808307,
      "grad_norm": 2.467360496520996,
      "learning_rate": 1.5902564102564103e-05,
      "loss": 0.3232,
      "step": 799
    },
    {
      "epoch": 10.223642172523961,
      "grad_norm": 2.8098020553588867,
      "learning_rate": 1.5897435897435897e-05,
      "loss": 0.2776,
      "step": 800
    },
    {
      "epoch": 10.223642172523961,
      "eval_loss": 0.4912850856781006,
      "eval_runtime": 183.0248,
      "eval_samples_per_second": 0.858,
      "eval_steps_per_second": 0.109,
      "step": 800
    },
    {
      "epoch": 10.236421725239616,
      "grad_norm": 2.853611946105957,
      "learning_rate": 1.5892307692307695e-05,
      "loss": 0.2964,
      "step": 801
    },
    {
      "epoch": 10.249201277955272,
      "grad_norm": 3.0821776390075684,
      "learning_rate": 1.588717948717949e-05,
      "loss": 0.3843,
      "step": 802
    },
    {
      "epoch": 10.261980830670927,
      "grad_norm": 3.1763129234313965,
      "learning_rate": 1.5882051282051283e-05,
      "loss": 0.4039,
      "step": 803
    },
    {
      "epoch": 10.274760383386582,
      "grad_norm": 3.0195302963256836,
      "learning_rate": 1.587692307692308e-05,
      "loss": 0.232,
      "step": 804
    },
    {
      "epoch": 10.287539936102236,
      "grad_norm": 2.823012113571167,
      "learning_rate": 1.587179487179487e-05,
      "loss": 0.2895,
      "step": 805
    },
    {
      "epoch": 10.300319488817891,
      "grad_norm": 3.0618393421173096,
      "learning_rate": 1.586666666666667e-05,
      "loss": 0.3148,
      "step": 806
    },
    {
      "epoch": 10.313099041533546,
      "grad_norm": 3.9492385387420654,
      "learning_rate": 1.5861538461538462e-05,
      "loss": 0.4268,
      "step": 807
    },
    {
      "epoch": 10.3258785942492,
      "grad_norm": 4.203477382659912,
      "learning_rate": 1.5856410256410256e-05,
      "loss": 0.395,
      "step": 808
    },
    {
      "epoch": 10.338658146964857,
      "grad_norm": 3.123945951461792,
      "learning_rate": 1.5851282051282054e-05,
      "loss": 0.3183,
      "step": 809
    },
    {
      "epoch": 10.351437699680512,
      "grad_norm": 3.6256487369537354,
      "learning_rate": 1.5846153846153848e-05,
      "loss": 0.3871,
      "step": 810
    },
    {
      "epoch": 10.364217252396166,
      "grad_norm": 3.1102445125579834,
      "learning_rate": 1.5841025641025642e-05,
      "loss": 0.2778,
      "step": 811
    },
    {
      "epoch": 10.37699680511182,
      "grad_norm": 3.4523651599884033,
      "learning_rate": 1.583589743589744e-05,
      "loss": 0.3446,
      "step": 812
    },
    {
      "epoch": 10.389776357827476,
      "grad_norm": 4.347939491271973,
      "learning_rate": 1.583076923076923e-05,
      "loss": 0.46,
      "step": 813
    },
    {
      "epoch": 10.40255591054313,
      "grad_norm": 3.2706682682037354,
      "learning_rate": 1.5825641025641028e-05,
      "loss": 0.4192,
      "step": 814
    },
    {
      "epoch": 10.415335463258787,
      "grad_norm": 2.67073392868042,
      "learning_rate": 1.582051282051282e-05,
      "loss": 0.2856,
      "step": 815
    },
    {
      "epoch": 10.428115015974441,
      "grad_norm": 3.1094117164611816,
      "learning_rate": 1.5815384615384616e-05,
      "loss": 0.2858,
      "step": 816
    },
    {
      "epoch": 10.440894568690096,
      "grad_norm": 3.1258299350738525,
      "learning_rate": 1.581025641025641e-05,
      "loss": 0.394,
      "step": 817
    },
    {
      "epoch": 10.45367412140575,
      "grad_norm": 3.4687180519104004,
      "learning_rate": 1.5805128205128207e-05,
      "loss": 0.3374,
      "step": 818
    },
    {
      "epoch": 10.466453674121405,
      "grad_norm": 3.3258817195892334,
      "learning_rate": 1.58e-05,
      "loss": 0.4092,
      "step": 819
    },
    {
      "epoch": 10.47923322683706,
      "grad_norm": 3.752936840057373,
      "learning_rate": 1.5794871794871795e-05,
      "loss": 0.3828,
      "step": 820
    },
    {
      "epoch": 10.492012779552716,
      "grad_norm": 3.322557210922241,
      "learning_rate": 1.5789743589743593e-05,
      "loss": 0.3219,
      "step": 821
    },
    {
      "epoch": 10.504792332268371,
      "grad_norm": 4.089233875274658,
      "learning_rate": 1.5784615384615383e-05,
      "loss": 0.4634,
      "step": 822
    },
    {
      "epoch": 10.517571884984026,
      "grad_norm": 3.043755292892456,
      "learning_rate": 1.577948717948718e-05,
      "loss": 0.3131,
      "step": 823
    },
    {
      "epoch": 10.53035143769968,
      "grad_norm": 2.8665387630462646,
      "learning_rate": 1.5774358974358975e-05,
      "loss": 0.4076,
      "step": 824
    },
    {
      "epoch": 10.543130990415335,
      "grad_norm": 3.6322171688079834,
      "learning_rate": 1.576923076923077e-05,
      "loss": 0.3824,
      "step": 825
    },
    {
      "epoch": 10.55591054313099,
      "grad_norm": 3.296170711517334,
      "learning_rate": 1.5764102564102566e-05,
      "loss": 0.4484,
      "step": 826
    },
    {
      "epoch": 10.568690095846645,
      "grad_norm": 2.6710081100463867,
      "learning_rate": 1.575897435897436e-05,
      "loss": 0.2255,
      "step": 827
    },
    {
      "epoch": 10.581469648562301,
      "grad_norm": 3.920431137084961,
      "learning_rate": 1.5753846153846154e-05,
      "loss": 0.4387,
      "step": 828
    },
    {
      "epoch": 10.594249201277956,
      "grad_norm": 2.7028772830963135,
      "learning_rate": 1.5748717948717952e-05,
      "loss": 0.3869,
      "step": 829
    },
    {
      "epoch": 10.60702875399361,
      "grad_norm": 3.00683331489563,
      "learning_rate": 1.5743589743589746e-05,
      "loss": 0.2282,
      "step": 830
    },
    {
      "epoch": 10.619808306709265,
      "grad_norm": 3.6137325763702393,
      "learning_rate": 1.573846153846154e-05,
      "loss": 0.3368,
      "step": 831
    },
    {
      "epoch": 10.63258785942492,
      "grad_norm": 3.2704150676727295,
      "learning_rate": 1.5733333333333334e-05,
      "loss": 0.3509,
      "step": 832
    },
    {
      "epoch": 10.645367412140574,
      "grad_norm": 3.7699923515319824,
      "learning_rate": 1.5728205128205128e-05,
      "loss": 0.3801,
      "step": 833
    },
    {
      "epoch": 10.65814696485623,
      "grad_norm": 3.1907780170440674,
      "learning_rate": 1.5723076923076926e-05,
      "loss": 0.4088,
      "step": 834
    },
    {
      "epoch": 10.670926517571885,
      "grad_norm": 3.042274236679077,
      "learning_rate": 1.571794871794872e-05,
      "loss": 0.3376,
      "step": 835
    },
    {
      "epoch": 10.68370607028754,
      "grad_norm": 2.675173282623291,
      "learning_rate": 1.5712820512820514e-05,
      "loss": 0.3057,
      "step": 836
    },
    {
      "epoch": 10.696485623003195,
      "grad_norm": 3.164384365081787,
      "learning_rate": 1.5707692307692308e-05,
      "loss": 0.361,
      "step": 837
    },
    {
      "epoch": 10.70926517571885,
      "grad_norm": 3.2448647022247314,
      "learning_rate": 1.5702564102564105e-05,
      "loss": 0.2626,
      "step": 838
    },
    {
      "epoch": 10.722044728434504,
      "grad_norm": 2.900031566619873,
      "learning_rate": 1.56974358974359e-05,
      "loss": 0.296,
      "step": 839
    },
    {
      "epoch": 10.73482428115016,
      "grad_norm": 3.1761841773986816,
      "learning_rate": 1.5692307692307693e-05,
      "loss": 0.2739,
      "step": 840
    },
    {
      "epoch": 10.747603833865815,
      "grad_norm": 3.444432497024536,
      "learning_rate": 1.568717948717949e-05,
      "loss": 0.3222,
      "step": 841
    },
    {
      "epoch": 10.76038338658147,
      "grad_norm": 3.3714373111724854,
      "learning_rate": 1.568205128205128e-05,
      "loss": 0.4064,
      "step": 842
    },
    {
      "epoch": 10.773162939297125,
      "grad_norm": 3.7057511806488037,
      "learning_rate": 1.567692307692308e-05,
      "loss": 0.5211,
      "step": 843
    },
    {
      "epoch": 10.78594249201278,
      "grad_norm": 3.079998254776001,
      "learning_rate": 1.5671794871794873e-05,
      "loss": 0.3329,
      "step": 844
    },
    {
      "epoch": 10.798722044728434,
      "grad_norm": 3.1543796062469482,
      "learning_rate": 1.5666666666666667e-05,
      "loss": 0.4407,
      "step": 845
    },
    {
      "epoch": 10.811501597444089,
      "grad_norm": 2.9210171699523926,
      "learning_rate": 1.5661538461538464e-05,
      "loss": 0.346,
      "step": 846
    },
    {
      "epoch": 10.824281150159745,
      "grad_norm": 3.3310635089874268,
      "learning_rate": 1.565641025641026e-05,
      "loss": 0.4371,
      "step": 847
    },
    {
      "epoch": 10.8370607028754,
      "grad_norm": 3.584244966506958,
      "learning_rate": 1.5651282051282052e-05,
      "loss": 0.4223,
      "step": 848
    },
    {
      "epoch": 10.849840255591054,
      "grad_norm": 2.6379106044769287,
      "learning_rate": 1.5646153846153846e-05,
      "loss": 0.3967,
      "step": 849
    },
    {
      "epoch": 10.86261980830671,
      "grad_norm": 2.972970724105835,
      "learning_rate": 1.5641025641025644e-05,
      "loss": 0.3419,
      "step": 850
    },
    {
      "epoch": 10.875399361022364,
      "grad_norm": 3.2894978523254395,
      "learning_rate": 1.5635897435897438e-05,
      "loss": 0.4133,
      "step": 851
    },
    {
      "epoch": 10.888178913738018,
      "grad_norm": 3.3246779441833496,
      "learning_rate": 1.5630769230769232e-05,
      "loss": 0.4367,
      "step": 852
    },
    {
      "epoch": 10.900958466453675,
      "grad_norm": 3.2328429222106934,
      "learning_rate": 1.5625641025641026e-05,
      "loss": 0.4688,
      "step": 853
    },
    {
      "epoch": 10.91373801916933,
      "grad_norm": 3.18841552734375,
      "learning_rate": 1.562051282051282e-05,
      "loss": 0.3355,
      "step": 854
    },
    {
      "epoch": 10.926517571884984,
      "grad_norm": 3.63415265083313,
      "learning_rate": 1.5615384615384618e-05,
      "loss": 0.3284,
      "step": 855
    },
    {
      "epoch": 10.939297124600639,
      "grad_norm": 2.851058006286621,
      "learning_rate": 1.561025641025641e-05,
      "loss": 0.2644,
      "step": 856
    },
    {
      "epoch": 10.952076677316294,
      "grad_norm": 3.3312418460845947,
      "learning_rate": 1.5605128205128206e-05,
      "loss": 0.4438,
      "step": 857
    },
    {
      "epoch": 10.964856230031948,
      "grad_norm": 4.002857685089111,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.336,
      "step": 858
    },
    {
      "epoch": 10.977635782747605,
      "grad_norm": 3.100975275039673,
      "learning_rate": 1.5594871794871794e-05,
      "loss": 0.437,
      "step": 859
    },
    {
      "epoch": 10.99041533546326,
      "grad_norm": 4.84814977645874,
      "learning_rate": 1.558974358974359e-05,
      "loss": 0.4298,
      "step": 860
    },
    {
      "epoch": 11.003194888178914,
      "grad_norm": 4.237227439880371,
      "learning_rate": 1.5584615384615385e-05,
      "loss": 0.34,
      "step": 861
    },
    {
      "epoch": 11.015974440894569,
      "grad_norm": 3.8608264923095703,
      "learning_rate": 1.557948717948718e-05,
      "loss": 0.4767,
      "step": 862
    },
    {
      "epoch": 11.028753993610223,
      "grad_norm": 3.2381210327148438,
      "learning_rate": 1.5574358974358977e-05,
      "loss": 0.3816,
      "step": 863
    },
    {
      "epoch": 11.041533546325878,
      "grad_norm": 3.009047031402588,
      "learning_rate": 1.556923076923077e-05,
      "loss": 0.3754,
      "step": 864
    },
    {
      "epoch": 11.054313099041533,
      "grad_norm": 3.1122117042541504,
      "learning_rate": 1.5564102564102565e-05,
      "loss": 0.4406,
      "step": 865
    },
    {
      "epoch": 11.06709265175719,
      "grad_norm": 2.9538486003875732,
      "learning_rate": 1.555897435897436e-05,
      "loss": 0.3496,
      "step": 866
    },
    {
      "epoch": 11.079872204472844,
      "grad_norm": 4.3143157958984375,
      "learning_rate": 1.5553846153846156e-05,
      "loss": 0.3724,
      "step": 867
    },
    {
      "epoch": 11.092651757188499,
      "grad_norm": 3.1895341873168945,
      "learning_rate": 1.554871794871795e-05,
      "loss": 0.3627,
      "step": 868
    },
    {
      "epoch": 11.105431309904153,
      "grad_norm": 2.7092154026031494,
      "learning_rate": 1.5543589743589744e-05,
      "loss": 0.2933,
      "step": 869
    },
    {
      "epoch": 11.118210862619808,
      "grad_norm": 2.3844687938690186,
      "learning_rate": 1.553846153846154e-05,
      "loss": 0.2012,
      "step": 870
    },
    {
      "epoch": 11.130990415335463,
      "grad_norm": 2.5584909915924072,
      "learning_rate": 1.5533333333333333e-05,
      "loss": 0.2561,
      "step": 871
    },
    {
      "epoch": 11.143769968051119,
      "grad_norm": 3.2287497520446777,
      "learning_rate": 1.552820512820513e-05,
      "loss": 0.2891,
      "step": 872
    },
    {
      "epoch": 11.156549520766774,
      "grad_norm": 2.3542449474334717,
      "learning_rate": 1.5523076923076924e-05,
      "loss": 0.2908,
      "step": 873
    },
    {
      "epoch": 11.169329073482428,
      "grad_norm": 2.944882392883301,
      "learning_rate": 1.5517948717948718e-05,
      "loss": 0.2287,
      "step": 874
    },
    {
      "epoch": 11.182108626198083,
      "grad_norm": 2.9199087619781494,
      "learning_rate": 1.5512820512820516e-05,
      "loss": 0.3413,
      "step": 875
    },
    {
      "epoch": 11.194888178913738,
      "grad_norm": 2.931480646133423,
      "learning_rate": 1.550769230769231e-05,
      "loss": 0.2779,
      "step": 876
    },
    {
      "epoch": 11.207667731629392,
      "grad_norm": 3.4157614707946777,
      "learning_rate": 1.5502564102564104e-05,
      "loss": 0.3269,
      "step": 877
    },
    {
      "epoch": 11.220447284345047,
      "grad_norm": 3.1395103931427,
      "learning_rate": 1.54974358974359e-05,
      "loss": 0.2674,
      "step": 878
    },
    {
      "epoch": 11.233226837060704,
      "grad_norm": 4.254883289337158,
      "learning_rate": 1.5492307692307692e-05,
      "loss": 0.4352,
      "step": 879
    },
    {
      "epoch": 11.246006389776358,
      "grad_norm": 3.7788186073303223,
      "learning_rate": 1.548717948717949e-05,
      "loss": 0.2672,
      "step": 880
    },
    {
      "epoch": 11.258785942492013,
      "grad_norm": 4.6283087730407715,
      "learning_rate": 1.5482051282051283e-05,
      "loss": 0.3878,
      "step": 881
    },
    {
      "epoch": 11.271565495207668,
      "grad_norm": 2.980556011199951,
      "learning_rate": 1.5476923076923077e-05,
      "loss": 0.2345,
      "step": 882
    },
    {
      "epoch": 11.284345047923322,
      "grad_norm": 3.723423719406128,
      "learning_rate": 1.5471794871794875e-05,
      "loss": 0.411,
      "step": 883
    },
    {
      "epoch": 11.297124600638977,
      "grad_norm": 3.2626750469207764,
      "learning_rate": 1.546666666666667e-05,
      "loss": 0.3025,
      "step": 884
    },
    {
      "epoch": 11.309904153354633,
      "grad_norm": 2.825082302093506,
      "learning_rate": 1.5461538461538463e-05,
      "loss": 0.3917,
      "step": 885
    },
    {
      "epoch": 11.322683706070288,
      "grad_norm": 3.563110828399658,
      "learning_rate": 1.5456410256410257e-05,
      "loss": 0.4258,
      "step": 886
    },
    {
      "epoch": 11.335463258785943,
      "grad_norm": 3.154862880706787,
      "learning_rate": 1.5451282051282054e-05,
      "loss": 0.3543,
      "step": 887
    },
    {
      "epoch": 11.348242811501597,
      "grad_norm": 3.8026084899902344,
      "learning_rate": 1.544615384615385e-05,
      "loss": 0.4576,
      "step": 888
    },
    {
      "epoch": 11.361022364217252,
      "grad_norm": 3.8097572326660156,
      "learning_rate": 1.5441025641025642e-05,
      "loss": 0.3371,
      "step": 889
    },
    {
      "epoch": 11.373801916932907,
      "grad_norm": 3.6081221103668213,
      "learning_rate": 1.5435897435897436e-05,
      "loss": 0.3445,
      "step": 890
    },
    {
      "epoch": 11.386581469648561,
      "grad_norm": 3.46155047416687,
      "learning_rate": 1.543076923076923e-05,
      "loss": 0.3375,
      "step": 891
    },
    {
      "epoch": 11.399361022364218,
      "grad_norm": 3.3258464336395264,
      "learning_rate": 1.5425641025641028e-05,
      "loss": 0.295,
      "step": 892
    },
    {
      "epoch": 11.412140575079873,
      "grad_norm": 3.3153886795043945,
      "learning_rate": 1.5420512820512822e-05,
      "loss": 0.3332,
      "step": 893
    },
    {
      "epoch": 11.424920127795527,
      "grad_norm": 3.5005221366882324,
      "learning_rate": 1.5415384615384616e-05,
      "loss": 0.286,
      "step": 894
    },
    {
      "epoch": 11.437699680511182,
      "grad_norm": 3.0043089389801025,
      "learning_rate": 1.5410256410256414e-05,
      "loss": 0.3095,
      "step": 895
    },
    {
      "epoch": 11.450479233226837,
      "grad_norm": 3.113715410232544,
      "learning_rate": 1.5405128205128204e-05,
      "loss": 0.2693,
      "step": 896
    },
    {
      "epoch": 11.463258785942491,
      "grad_norm": 3.4853270053863525,
      "learning_rate": 1.54e-05,
      "loss": 0.3533,
      "step": 897
    },
    {
      "epoch": 11.476038338658148,
      "grad_norm": 3.3521034717559814,
      "learning_rate": 1.5394871794871796e-05,
      "loss": 0.3699,
      "step": 898
    },
    {
      "epoch": 11.488817891373802,
      "grad_norm": 3.983391046524048,
      "learning_rate": 1.538974358974359e-05,
      "loss": 0.303,
      "step": 899
    },
    {
      "epoch": 11.501597444089457,
      "grad_norm": 3.161827325820923,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 0.2873,
      "step": 900
    },
    {
      "epoch": 11.501597444089457,
      "eval_loss": 0.49615687131881714,
      "eval_runtime": 183.2354,
      "eval_samples_per_second": 0.857,
      "eval_steps_per_second": 0.109,
      "step": 900
    },
    {
      "epoch": 11.514376996805112,
      "grad_norm": 3.4103128910064697,
      "learning_rate": 1.537948717948718e-05,
      "loss": 0.3428,
      "step": 901
    },
    {
      "epoch": 11.527156549520766,
      "grad_norm": 3.620884418487549,
      "learning_rate": 1.5374358974358975e-05,
      "loss": 0.4551,
      "step": 902
    },
    {
      "epoch": 11.539936102236421,
      "grad_norm": 3.1130776405334473,
      "learning_rate": 1.536923076923077e-05,
      "loss": 0.3008,
      "step": 903
    },
    {
      "epoch": 11.552715654952078,
      "grad_norm": 3.5067083835601807,
      "learning_rate": 1.5364102564102567e-05,
      "loss": 0.4184,
      "step": 904
    },
    {
      "epoch": 11.565495207667732,
      "grad_norm": 3.9358503818511963,
      "learning_rate": 1.535897435897436e-05,
      "loss": 0.4187,
      "step": 905
    },
    {
      "epoch": 11.578274760383387,
      "grad_norm": 3.875493049621582,
      "learning_rate": 1.5353846153846155e-05,
      "loss": 0.3645,
      "step": 906
    },
    {
      "epoch": 11.591054313099042,
      "grad_norm": 3.0121140480041504,
      "learning_rate": 1.534871794871795e-05,
      "loss": 0.2874,
      "step": 907
    },
    {
      "epoch": 11.603833865814696,
      "grad_norm": 3.0029237270355225,
      "learning_rate": 1.5343589743589743e-05,
      "loss": 0.3083,
      "step": 908
    },
    {
      "epoch": 11.616613418530351,
      "grad_norm": 2.714341878890991,
      "learning_rate": 1.533846153846154e-05,
      "loss": 0.22,
      "step": 909
    },
    {
      "epoch": 11.629392971246006,
      "grad_norm": 2.5919463634490967,
      "learning_rate": 1.5333333333333334e-05,
      "loss": 0.27,
      "step": 910
    },
    {
      "epoch": 11.642172523961662,
      "grad_norm": 3.5662481784820557,
      "learning_rate": 1.532820512820513e-05,
      "loss": 0.3671,
      "step": 911
    },
    {
      "epoch": 11.654952076677317,
      "grad_norm": 3.2857398986816406,
      "learning_rate": 1.5323076923076926e-05,
      "loss": 0.3629,
      "step": 912
    },
    {
      "epoch": 11.667731629392971,
      "grad_norm": 3.2995290756225586,
      "learning_rate": 1.531794871794872e-05,
      "loss": 0.345,
      "step": 913
    },
    {
      "epoch": 11.680511182108626,
      "grad_norm": 3.5005154609680176,
      "learning_rate": 1.5312820512820514e-05,
      "loss": 0.2964,
      "step": 914
    },
    {
      "epoch": 11.69329073482428,
      "grad_norm": 4.271259307861328,
      "learning_rate": 1.5307692307692308e-05,
      "loss": 0.4868,
      "step": 915
    },
    {
      "epoch": 11.706070287539935,
      "grad_norm": 3.320885181427002,
      "learning_rate": 1.5302564102564102e-05,
      "loss": 0.3535,
      "step": 916
    },
    {
      "epoch": 11.718849840255592,
      "grad_norm": 4.391448020935059,
      "learning_rate": 1.52974358974359e-05,
      "loss": 0.4242,
      "step": 917
    },
    {
      "epoch": 11.731629392971247,
      "grad_norm": 3.9817147254943848,
      "learning_rate": 1.5292307692307694e-05,
      "loss": 0.329,
      "step": 918
    },
    {
      "epoch": 11.744408945686901,
      "grad_norm": 3.0762832164764404,
      "learning_rate": 1.5287179487179488e-05,
      "loss": 0.248,
      "step": 919
    },
    {
      "epoch": 11.757188498402556,
      "grad_norm": 3.191990613937378,
      "learning_rate": 1.5282051282051282e-05,
      "loss": 0.3691,
      "step": 920
    },
    {
      "epoch": 11.76996805111821,
      "grad_norm": 3.4746310710906982,
      "learning_rate": 1.527692307692308e-05,
      "loss": 0.3916,
      "step": 921
    },
    {
      "epoch": 11.782747603833865,
      "grad_norm": 3.9543192386627197,
      "learning_rate": 1.5271794871794873e-05,
      "loss": 0.4443,
      "step": 922
    },
    {
      "epoch": 11.795527156549522,
      "grad_norm": 3.446842670440674,
      "learning_rate": 1.5266666666666667e-05,
      "loss": 0.329,
      "step": 923
    },
    {
      "epoch": 11.808306709265176,
      "grad_norm": 3.134056806564331,
      "learning_rate": 1.5261538461538465e-05,
      "loss": 0.2734,
      "step": 924
    },
    {
      "epoch": 11.821086261980831,
      "grad_norm": 3.8701589107513428,
      "learning_rate": 1.5256410256410257e-05,
      "loss": 0.4735,
      "step": 925
    },
    {
      "epoch": 11.833865814696486,
      "grad_norm": 4.0366902351379395,
      "learning_rate": 1.5251282051282053e-05,
      "loss": 0.3053,
      "step": 926
    },
    {
      "epoch": 11.84664536741214,
      "grad_norm": 2.7000772953033447,
      "learning_rate": 1.5246153846153849e-05,
      "loss": 0.2448,
      "step": 927
    },
    {
      "epoch": 11.859424920127795,
      "grad_norm": 3.872736930847168,
      "learning_rate": 1.5241025641025643e-05,
      "loss": 0.4172,
      "step": 928
    },
    {
      "epoch": 11.87220447284345,
      "grad_norm": 2.7194664478302,
      "learning_rate": 1.5235897435897438e-05,
      "loss": 0.2179,
      "step": 929
    },
    {
      "epoch": 11.884984025559106,
      "grad_norm": 2.8944573402404785,
      "learning_rate": 1.523076923076923e-05,
      "loss": 0.2808,
      "step": 930
    },
    {
      "epoch": 11.89776357827476,
      "grad_norm": 3.091066598892212,
      "learning_rate": 1.5225641025641027e-05,
      "loss": 0.3228,
      "step": 931
    },
    {
      "epoch": 11.910543130990416,
      "grad_norm": 3.6224167346954346,
      "learning_rate": 1.5220512820512822e-05,
      "loss": 0.4068,
      "step": 932
    },
    {
      "epoch": 11.92332268370607,
      "grad_norm": 3.6418344974517822,
      "learning_rate": 1.5215384615384616e-05,
      "loss": 0.3721,
      "step": 933
    },
    {
      "epoch": 11.936102236421725,
      "grad_norm": 3.031254529953003,
      "learning_rate": 1.5210256410256412e-05,
      "loss": 0.2927,
      "step": 934
    },
    {
      "epoch": 11.94888178913738,
      "grad_norm": 4.135180950164795,
      "learning_rate": 1.5205128205128206e-05,
      "loss": 0.3205,
      "step": 935
    },
    {
      "epoch": 11.961661341853034,
      "grad_norm": 3.6757776737213135,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.3106,
      "step": 936
    },
    {
      "epoch": 11.97444089456869,
      "grad_norm": 3.8090317249298096,
      "learning_rate": 1.5194871794871798e-05,
      "loss": 0.3209,
      "step": 937
    },
    {
      "epoch": 11.987220447284345,
      "grad_norm": 2.8670780658721924,
      "learning_rate": 1.518974358974359e-05,
      "loss": 0.3085,
      "step": 938
    },
    {
      "epoch": 12.0,
      "grad_norm": 3.156921148300171,
      "learning_rate": 1.5184615384615386e-05,
      "loss": 0.3247,
      "step": 939
    },
    {
      "epoch": 12.012779552715655,
      "grad_norm": 3.98889422416687,
      "learning_rate": 1.517948717948718e-05,
      "loss": 0.2793,
      "step": 940
    },
    {
      "epoch": 12.02555910543131,
      "grad_norm": 3.3414158821105957,
      "learning_rate": 1.5174358974358976e-05,
      "loss": 0.2521,
      "step": 941
    },
    {
      "epoch": 12.038338658146964,
      "grad_norm": 3.1532747745513916,
      "learning_rate": 1.516923076923077e-05,
      "loss": 0.2792,
      "step": 942
    },
    {
      "epoch": 12.05111821086262,
      "grad_norm": 3.7512667179107666,
      "learning_rate": 1.5164102564102565e-05,
      "loss": 0.2176,
      "step": 943
    },
    {
      "epoch": 12.063897763578275,
      "grad_norm": 3.0928268432617188,
      "learning_rate": 1.5158974358974361e-05,
      "loss": 0.376,
      "step": 944
    },
    {
      "epoch": 12.07667731629393,
      "grad_norm": 2.8840041160583496,
      "learning_rate": 1.5153846153846155e-05,
      "loss": 0.2967,
      "step": 945
    },
    {
      "epoch": 12.089456869009584,
      "grad_norm": 2.912339925765991,
      "learning_rate": 1.514871794871795e-05,
      "loss": 0.2307,
      "step": 946
    },
    {
      "epoch": 12.10223642172524,
      "grad_norm": 4.07706880569458,
      "learning_rate": 1.5143589743589743e-05,
      "loss": 0.3104,
      "step": 947
    },
    {
      "epoch": 12.115015974440894,
      "grad_norm": 3.8254222869873047,
      "learning_rate": 1.5138461538461539e-05,
      "loss": 0.299,
      "step": 948
    },
    {
      "epoch": 12.12779552715655,
      "grad_norm": 3.255998134613037,
      "learning_rate": 1.5133333333333335e-05,
      "loss": 0.2724,
      "step": 949
    },
    {
      "epoch": 12.140575079872205,
      "grad_norm": 3.329000949859619,
      "learning_rate": 1.5128205128205129e-05,
      "loss": 0.3398,
      "step": 950
    },
    {
      "epoch": 12.15335463258786,
      "grad_norm": 3.314716339111328,
      "learning_rate": 1.5123076923076924e-05,
      "loss": 0.2941,
      "step": 951
    },
    {
      "epoch": 12.166134185303514,
      "grad_norm": 2.983738660812378,
      "learning_rate": 1.5117948717948719e-05,
      "loss": 0.2255,
      "step": 952
    },
    {
      "epoch": 12.178913738019169,
      "grad_norm": 3.6834664344787598,
      "learning_rate": 1.5112820512820514e-05,
      "loss": 0.2873,
      "step": 953
    },
    {
      "epoch": 12.191693290734824,
      "grad_norm": 3.5403318405151367,
      "learning_rate": 1.510769230769231e-05,
      "loss": 0.3309,
      "step": 954
    },
    {
      "epoch": 12.204472843450478,
      "grad_norm": 3.9441781044006348,
      "learning_rate": 1.5102564102564104e-05,
      "loss": 0.287,
      "step": 955
    },
    {
      "epoch": 12.217252396166135,
      "grad_norm": 2.9877588748931885,
      "learning_rate": 1.50974358974359e-05,
      "loss": 0.2715,
      "step": 956
    },
    {
      "epoch": 12.23003194888179,
      "grad_norm": 3.5199813842773438,
      "learning_rate": 1.5092307692307692e-05,
      "loss": 0.4142,
      "step": 957
    },
    {
      "epoch": 12.242811501597444,
      "grad_norm": 3.2963194847106934,
      "learning_rate": 1.5087179487179488e-05,
      "loss": 0.2854,
      "step": 958
    },
    {
      "epoch": 12.255591054313099,
      "grad_norm": 4.313882350921631,
      "learning_rate": 1.5082051282051284e-05,
      "loss": 0.4677,
      "step": 959
    },
    {
      "epoch": 12.268370607028753,
      "grad_norm": 4.474306583404541,
      "learning_rate": 1.5076923076923078e-05,
      "loss": 0.3723,
      "step": 960
    },
    {
      "epoch": 12.281150159744408,
      "grad_norm": 3.5448555946350098,
      "learning_rate": 1.5071794871794873e-05,
      "loss": 0.2871,
      "step": 961
    },
    {
      "epoch": 12.293929712460065,
      "grad_norm": 3.978175163269043,
      "learning_rate": 1.5066666666666668e-05,
      "loss": 0.3368,
      "step": 962
    },
    {
      "epoch": 12.30670926517572,
      "grad_norm": 4.108724117279053,
      "learning_rate": 1.5061538461538463e-05,
      "loss": 0.3639,
      "step": 963
    },
    {
      "epoch": 12.319488817891374,
      "grad_norm": 3.223417043685913,
      "learning_rate": 1.5056410256410257e-05,
      "loss": 0.2408,
      "step": 964
    },
    {
      "epoch": 12.332268370607029,
      "grad_norm": 3.166851043701172,
      "learning_rate": 1.5051282051282053e-05,
      "loss": 0.2552,
      "step": 965
    },
    {
      "epoch": 12.345047923322683,
      "grad_norm": 4.144366264343262,
      "learning_rate": 1.5046153846153849e-05,
      "loss": 0.344,
      "step": 966
    },
    {
      "epoch": 12.357827476038338,
      "grad_norm": 3.4670357704162598,
      "learning_rate": 1.5041025641025641e-05,
      "loss": 0.3234,
      "step": 967
    },
    {
      "epoch": 12.370607028753994,
      "grad_norm": 3.67396879196167,
      "learning_rate": 1.5035897435897437e-05,
      "loss": 0.3976,
      "step": 968
    },
    {
      "epoch": 12.383386581469649,
      "grad_norm": 7.3044867515563965,
      "learning_rate": 1.5030769230769231e-05,
      "loss": 0.2831,
      "step": 969
    },
    {
      "epoch": 12.396166134185304,
      "grad_norm": 4.498189926147461,
      "learning_rate": 1.5025641025641027e-05,
      "loss": 0.2689,
      "step": 970
    },
    {
      "epoch": 12.408945686900958,
      "grad_norm": 3.387584924697876,
      "learning_rate": 1.5020512820512822e-05,
      "loss": 0.2477,
      "step": 971
    },
    {
      "epoch": 12.421725239616613,
      "grad_norm": 4.313808917999268,
      "learning_rate": 1.5015384615384617e-05,
      "loss": 0.309,
      "step": 972
    },
    {
      "epoch": 12.434504792332268,
      "grad_norm": 3.6473467350006104,
      "learning_rate": 1.5010256410256412e-05,
      "loss": 0.2361,
      "step": 973
    },
    {
      "epoch": 12.447284345047922,
      "grad_norm": 4.041528701782227,
      "learning_rate": 1.5005128205128205e-05,
      "loss": 0.3795,
      "step": 974
    },
    {
      "epoch": 12.460063897763579,
      "grad_norm": 3.3756825923919678,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.2511,
      "step": 975
    },
    {
      "epoch": 12.472843450479234,
      "grad_norm": 3.396024703979492,
      "learning_rate": 1.4994871794871798e-05,
      "loss": 0.3154,
      "step": 976
    },
    {
      "epoch": 12.485623003194888,
      "grad_norm": 3.4328291416168213,
      "learning_rate": 1.498974358974359e-05,
      "loss": 0.2605,
      "step": 977
    },
    {
      "epoch": 12.498402555910543,
      "grad_norm": 4.163756370544434,
      "learning_rate": 1.4984615384615386e-05,
      "loss": 0.2772,
      "step": 978
    },
    {
      "epoch": 12.511182108626198,
      "grad_norm": 4.738368034362793,
      "learning_rate": 1.497948717948718e-05,
      "loss": 0.3521,
      "step": 979
    },
    {
      "epoch": 12.523961661341852,
      "grad_norm": 3.4938671588897705,
      "learning_rate": 1.4974358974358976e-05,
      "loss": 0.363,
      "step": 980
    },
    {
      "epoch": 12.536741214057509,
      "grad_norm": 3.3207099437713623,
      "learning_rate": 1.4969230769230771e-05,
      "loss": 0.2558,
      "step": 981
    },
    {
      "epoch": 12.549520766773163,
      "grad_norm": 3.4112465381622314,
      "learning_rate": 1.4964102564102566e-05,
      "loss": 0.3083,
      "step": 982
    },
    {
      "epoch": 12.562300319488818,
      "grad_norm": 3.6576359272003174,
      "learning_rate": 1.4958974358974361e-05,
      "loss": 0.3153,
      "step": 983
    },
    {
      "epoch": 12.575079872204473,
      "grad_norm": 3.1514041423797607,
      "learning_rate": 1.4953846153846154e-05,
      "loss": 0.2925,
      "step": 984
    },
    {
      "epoch": 12.587859424920127,
      "grad_norm": 3.492708206176758,
      "learning_rate": 1.494871794871795e-05,
      "loss": 0.3153,
      "step": 985
    },
    {
      "epoch": 12.600638977635782,
      "grad_norm": 3.412132740020752,
      "learning_rate": 1.4943589743589745e-05,
      "loss": 0.3022,
      "step": 986
    },
    {
      "epoch": 12.613418530351439,
      "grad_norm": 4.253443241119385,
      "learning_rate": 1.493846153846154e-05,
      "loss": 0.2909,
      "step": 987
    },
    {
      "epoch": 12.626198083067093,
      "grad_norm": 3.0681474208831787,
      "learning_rate": 1.4933333333333335e-05,
      "loss": 0.4074,
      "step": 988
    },
    {
      "epoch": 12.638977635782748,
      "grad_norm": 3.2361674308776855,
      "learning_rate": 1.4928205128205129e-05,
      "loss": 0.4487,
      "step": 989
    },
    {
      "epoch": 12.651757188498403,
      "grad_norm": 3.1446540355682373,
      "learning_rate": 1.4923076923076925e-05,
      "loss": 0.242,
      "step": 990
    },
    {
      "epoch": 12.664536741214057,
      "grad_norm": 4.866894245147705,
      "learning_rate": 1.4917948717948719e-05,
      "loss": 0.4236,
      "step": 991
    },
    {
      "epoch": 12.677316293929712,
      "grad_norm": 3.479435920715332,
      "learning_rate": 1.4912820512820515e-05,
      "loss": 0.2779,
      "step": 992
    },
    {
      "epoch": 12.690095846645367,
      "grad_norm": 3.4308862686157227,
      "learning_rate": 1.490769230769231e-05,
      "loss": 0.312,
      "step": 993
    },
    {
      "epoch": 12.702875399361023,
      "grad_norm": 4.010195732116699,
      "learning_rate": 1.4902564102564103e-05,
      "loss": 0.2753,
      "step": 994
    },
    {
      "epoch": 12.715654952076678,
      "grad_norm": 3.1232686042785645,
      "learning_rate": 1.4897435897435898e-05,
      "loss": 0.2055,
      "step": 995
    },
    {
      "epoch": 12.728434504792332,
      "grad_norm": 4.415332317352295,
      "learning_rate": 1.4892307692307692e-05,
      "loss": 0.4439,
      "step": 996
    },
    {
      "epoch": 12.741214057507987,
      "grad_norm": 4.104088306427002,
      "learning_rate": 1.4887179487179488e-05,
      "loss": 0.3345,
      "step": 997
    },
    {
      "epoch": 12.753993610223642,
      "grad_norm": 4.060264587402344,
      "learning_rate": 1.4882051282051284e-05,
      "loss": 0.2695,
      "step": 998
    },
    {
      "epoch": 12.766773162939296,
      "grad_norm": 2.8599672317504883,
      "learning_rate": 1.4876923076923078e-05,
      "loss": 0.2159,
      "step": 999
    },
    {
      "epoch": 12.779552715654953,
      "grad_norm": 4.045029640197754,
      "learning_rate": 1.4871794871794874e-05,
      "loss": 0.4147,
      "step": 1000
    },
    {
      "epoch": 12.779552715654953,
      "eval_loss": 0.50066077709198,
      "eval_runtime": 183.3606,
      "eval_samples_per_second": 0.856,
      "eval_steps_per_second": 0.109,
      "step": 1000
    },
    {
      "epoch": 12.792332268370608,
      "grad_norm": 3.8444182872772217,
      "learning_rate": 1.4866666666666668e-05,
      "loss": 0.2935,
      "step": 1001
    },
    {
      "epoch": 12.805111821086262,
      "grad_norm": 3.37121844291687,
      "learning_rate": 1.4861538461538464e-05,
      "loss": 0.3359,
      "step": 1002
    },
    {
      "epoch": 12.817891373801917,
      "grad_norm": 3.2778074741363525,
      "learning_rate": 1.485641025641026e-05,
      "loss": 0.2979,
      "step": 1003
    },
    {
      "epoch": 12.830670926517572,
      "grad_norm": 4.910933017730713,
      "learning_rate": 1.4851282051282052e-05,
      "loss": 0.4186,
      "step": 1004
    },
    {
      "epoch": 12.843450479233226,
      "grad_norm": 3.9079768657684326,
      "learning_rate": 1.4846153846153847e-05,
      "loss": 0.3384,
      "step": 1005
    },
    {
      "epoch": 12.856230031948883,
      "grad_norm": 4.07132625579834,
      "learning_rate": 1.4841025641025641e-05,
      "loss": 0.2982,
      "step": 1006
    },
    {
      "epoch": 12.869009584664537,
      "grad_norm": 3.813551664352417,
      "learning_rate": 1.4835897435897437e-05,
      "loss": 0.316,
      "step": 1007
    },
    {
      "epoch": 12.881789137380192,
      "grad_norm": 3.3355159759521484,
      "learning_rate": 1.4830769230769233e-05,
      "loss": 0.3037,
      "step": 1008
    },
    {
      "epoch": 12.894568690095847,
      "grad_norm": 3.209801197052002,
      "learning_rate": 1.4825641025641027e-05,
      "loss": 0.2607,
      "step": 1009
    },
    {
      "epoch": 12.907348242811501,
      "grad_norm": 3.589578151702881,
      "learning_rate": 1.4820512820512823e-05,
      "loss": 0.2936,
      "step": 1010
    },
    {
      "epoch": 12.920127795527156,
      "grad_norm": 3.3078291416168213,
      "learning_rate": 1.4815384615384617e-05,
      "loss": 0.3175,
      "step": 1011
    },
    {
      "epoch": 12.93290734824281,
      "grad_norm": 4.202841281890869,
      "learning_rate": 1.4810256410256412e-05,
      "loss": 0.3471,
      "step": 1012
    },
    {
      "epoch": 12.945686900958467,
      "grad_norm": 3.8688015937805176,
      "learning_rate": 1.4805128205128205e-05,
      "loss": 0.2581,
      "step": 1013
    },
    {
      "epoch": 12.958466453674122,
      "grad_norm": 3.566972255706787,
      "learning_rate": 1.48e-05,
      "loss": 0.3632,
      "step": 1014
    },
    {
      "epoch": 12.971246006389777,
      "grad_norm": 4.008389472961426,
      "learning_rate": 1.4794871794871796e-05,
      "loss": 0.3105,
      "step": 1015
    },
    {
      "epoch": 12.984025559105431,
      "grad_norm": 4.361541271209717,
      "learning_rate": 1.478974358974359e-05,
      "loss": 0.4006,
      "step": 1016
    },
    {
      "epoch": 12.996805111821086,
      "grad_norm": 4.685972213745117,
      "learning_rate": 1.4784615384615386e-05,
      "loss": 0.4876,
      "step": 1017
    },
    {
      "epoch": 13.00958466453674,
      "grad_norm": 3.5489296913146973,
      "learning_rate": 1.477948717948718e-05,
      "loss": 0.2712,
      "step": 1018
    },
    {
      "epoch": 13.022364217252397,
      "grad_norm": 3.135805606842041,
      "learning_rate": 1.4774358974358976e-05,
      "loss": 0.1518,
      "step": 1019
    },
    {
      "epoch": 13.035143769968052,
      "grad_norm": 3.1784472465515137,
      "learning_rate": 1.4769230769230772e-05,
      "loss": 0.237,
      "step": 1020
    },
    {
      "epoch": 13.047923322683706,
      "grad_norm": 3.3885161876678467,
      "learning_rate": 1.4764102564102564e-05,
      "loss": 0.3225,
      "step": 1021
    },
    {
      "epoch": 13.060702875399361,
      "grad_norm": 4.256035327911377,
      "learning_rate": 1.475897435897436e-05,
      "loss": 0.3613,
      "step": 1022
    },
    {
      "epoch": 13.073482428115016,
      "grad_norm": 4.67989444732666,
      "learning_rate": 1.4753846153846154e-05,
      "loss": 0.3487,
      "step": 1023
    },
    {
      "epoch": 13.08626198083067,
      "grad_norm": 3.1670827865600586,
      "learning_rate": 1.474871794871795e-05,
      "loss": 0.2751,
      "step": 1024
    },
    {
      "epoch": 13.099041533546325,
      "grad_norm": 2.552567481994629,
      "learning_rate": 1.4743589743589745e-05,
      "loss": 0.1818,
      "step": 1025
    },
    {
      "epoch": 13.111821086261982,
      "grad_norm": 4.138055801391602,
      "learning_rate": 1.473846153846154e-05,
      "loss": 0.314,
      "step": 1026
    },
    {
      "epoch": 13.124600638977636,
      "grad_norm": 3.173854112625122,
      "learning_rate": 1.4733333333333335e-05,
      "loss": 0.2503,
      "step": 1027
    },
    {
      "epoch": 13.13738019169329,
      "grad_norm": 4.015395164489746,
      "learning_rate": 1.472820512820513e-05,
      "loss": 0.2352,
      "step": 1028
    },
    {
      "epoch": 13.150159744408946,
      "grad_norm": 4.005403995513916,
      "learning_rate": 1.4723076923076925e-05,
      "loss": 0.364,
      "step": 1029
    },
    {
      "epoch": 13.1629392971246,
      "grad_norm": 3.498249053955078,
      "learning_rate": 1.471794871794872e-05,
      "loss": 0.2862,
      "step": 1030
    },
    {
      "epoch": 13.175718849840255,
      "grad_norm": 4.659630298614502,
      "learning_rate": 1.4712820512820513e-05,
      "loss": 0.2042,
      "step": 1031
    },
    {
      "epoch": 13.188498402555911,
      "grad_norm": 3.3224682807922363,
      "learning_rate": 1.4707692307692309e-05,
      "loss": 0.2379,
      "step": 1032
    },
    {
      "epoch": 13.201277955271566,
      "grad_norm": 3.7572021484375,
      "learning_rate": 1.4702564102564103e-05,
      "loss": 0.2609,
      "step": 1033
    },
    {
      "epoch": 13.21405750798722,
      "grad_norm": 4.294779300689697,
      "learning_rate": 1.4697435897435899e-05,
      "loss": 0.2982,
      "step": 1034
    },
    {
      "epoch": 13.226837060702875,
      "grad_norm": 3.1984715461730957,
      "learning_rate": 1.4692307692307694e-05,
      "loss": 0.2467,
      "step": 1035
    },
    {
      "epoch": 13.23961661341853,
      "grad_norm": 4.2931318283081055,
      "learning_rate": 1.4687179487179488e-05,
      "loss": 0.3132,
      "step": 1036
    },
    {
      "epoch": 13.252396166134185,
      "grad_norm": 4.280733585357666,
      "learning_rate": 1.4682051282051284e-05,
      "loss": 0.3911,
      "step": 1037
    },
    {
      "epoch": 13.26517571884984,
      "grad_norm": 3.6063473224639893,
      "learning_rate": 1.4676923076923078e-05,
      "loss": 0.2817,
      "step": 1038
    },
    {
      "epoch": 13.277955271565496,
      "grad_norm": 3.960935354232788,
      "learning_rate": 1.4671794871794874e-05,
      "loss": 0.2993,
      "step": 1039
    },
    {
      "epoch": 13.29073482428115,
      "grad_norm": 3.975257396697998,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.265,
      "step": 1040
    },
    {
      "epoch": 13.303514376996805,
      "grad_norm": 4.084619045257568,
      "learning_rate": 1.4661538461538462e-05,
      "loss": 0.2999,
      "step": 1041
    },
    {
      "epoch": 13.31629392971246,
      "grad_norm": 4.0943074226379395,
      "learning_rate": 1.4656410256410258e-05,
      "loss": 0.3043,
      "step": 1042
    },
    {
      "epoch": 13.329073482428115,
      "grad_norm": 4.909326553344727,
      "learning_rate": 1.4651282051282052e-05,
      "loss": 0.3645,
      "step": 1043
    },
    {
      "epoch": 13.34185303514377,
      "grad_norm": 3.738766670227051,
      "learning_rate": 1.4646153846153848e-05,
      "loss": 0.2897,
      "step": 1044
    },
    {
      "epoch": 13.354632587859426,
      "grad_norm": 4.270811557769775,
      "learning_rate": 1.4641025641025642e-05,
      "loss": 0.267,
      "step": 1045
    },
    {
      "epoch": 13.36741214057508,
      "grad_norm": 4.264181613922119,
      "learning_rate": 1.4635897435897437e-05,
      "loss": 0.3641,
      "step": 1046
    },
    {
      "epoch": 13.380191693290735,
      "grad_norm": 3.642995595932007,
      "learning_rate": 1.4630769230769233e-05,
      "loss": 0.2911,
      "step": 1047
    },
    {
      "epoch": 13.39297124600639,
      "grad_norm": 3.135666847229004,
      "learning_rate": 1.4625641025641027e-05,
      "loss": 0.2103,
      "step": 1048
    },
    {
      "epoch": 13.405750798722044,
      "grad_norm": 3.1206305027008057,
      "learning_rate": 1.4620512820512823e-05,
      "loss": 0.1989,
      "step": 1049
    },
    {
      "epoch": 13.418530351437699,
      "grad_norm": 3.8206746578216553,
      "learning_rate": 1.4615384615384615e-05,
      "loss": 0.3463,
      "step": 1050
    },
    {
      "epoch": 13.431309904153355,
      "grad_norm": 4.430676460266113,
      "learning_rate": 1.4610256410256411e-05,
      "loss": 0.2887,
      "step": 1051
    },
    {
      "epoch": 13.44408945686901,
      "grad_norm": 3.5906941890716553,
      "learning_rate": 1.4605128205128207e-05,
      "loss": 0.2832,
      "step": 1052
    },
    {
      "epoch": 13.456869009584665,
      "grad_norm": 3.3331151008605957,
      "learning_rate": 1.46e-05,
      "loss": 0.3086,
      "step": 1053
    },
    {
      "epoch": 13.46964856230032,
      "grad_norm": 3.582110643386841,
      "learning_rate": 1.4594871794871797e-05,
      "loss": 0.2267,
      "step": 1054
    },
    {
      "epoch": 13.482428115015974,
      "grad_norm": 3.459418535232544,
      "learning_rate": 1.458974358974359e-05,
      "loss": 0.2595,
      "step": 1055
    },
    {
      "epoch": 13.495207667731629,
      "grad_norm": 3.323029041290283,
      "learning_rate": 1.4584615384615386e-05,
      "loss": 0.287,
      "step": 1056
    },
    {
      "epoch": 13.507987220447284,
      "grad_norm": 3.314846992492676,
      "learning_rate": 1.4579487179487182e-05,
      "loss": 0.2498,
      "step": 1057
    },
    {
      "epoch": 13.52076677316294,
      "grad_norm": 3.9220614433288574,
      "learning_rate": 1.4574358974358974e-05,
      "loss": 0.2485,
      "step": 1058
    },
    {
      "epoch": 13.533546325878595,
      "grad_norm": 3.4297268390655518,
      "learning_rate": 1.4569230769230772e-05,
      "loss": 0.2627,
      "step": 1059
    },
    {
      "epoch": 13.54632587859425,
      "grad_norm": 4.226787567138672,
      "learning_rate": 1.4564102564102564e-05,
      "loss": 0.3431,
      "step": 1060
    },
    {
      "epoch": 13.559105431309904,
      "grad_norm": 4.605924606323242,
      "learning_rate": 1.455897435897436e-05,
      "loss": 0.2833,
      "step": 1061
    },
    {
      "epoch": 13.571884984025559,
      "grad_norm": 4.2034525871276855,
      "learning_rate": 1.4553846153846154e-05,
      "loss": 0.3244,
      "step": 1062
    },
    {
      "epoch": 13.584664536741213,
      "grad_norm": 3.6468238830566406,
      "learning_rate": 1.454871794871795e-05,
      "loss": 0.3597,
      "step": 1063
    },
    {
      "epoch": 13.59744408945687,
      "grad_norm": 3.9421803951263428,
      "learning_rate": 1.4543589743589746e-05,
      "loss": 0.2524,
      "step": 1064
    },
    {
      "epoch": 13.610223642172524,
      "grad_norm": 3.6966607570648193,
      "learning_rate": 1.453846153846154e-05,
      "loss": 0.255,
      "step": 1065
    },
    {
      "epoch": 13.62300319488818,
      "grad_norm": 4.093931674957275,
      "learning_rate": 1.4533333333333335e-05,
      "loss": 0.2975,
      "step": 1066
    },
    {
      "epoch": 13.635782747603834,
      "grad_norm": 3.0992209911346436,
      "learning_rate": 1.4528205128205128e-05,
      "loss": 0.3078,
      "step": 1067
    },
    {
      "epoch": 13.648562300319488,
      "grad_norm": 4.053220272064209,
      "learning_rate": 1.4523076923076923e-05,
      "loss": 0.2661,
      "step": 1068
    },
    {
      "epoch": 13.661341853035143,
      "grad_norm": 4.509278297424316,
      "learning_rate": 1.451794871794872e-05,
      "loss": 0.2897,
      "step": 1069
    },
    {
      "epoch": 13.6741214057508,
      "grad_norm": 4.291611194610596,
      "learning_rate": 1.4512820512820513e-05,
      "loss": 0.3536,
      "step": 1070
    },
    {
      "epoch": 13.686900958466454,
      "grad_norm": 4.656402111053467,
      "learning_rate": 1.4507692307692309e-05,
      "loss": 0.2744,
      "step": 1071
    },
    {
      "epoch": 13.699680511182109,
      "grad_norm": 3.606649398803711,
      "learning_rate": 1.4502564102564103e-05,
      "loss": 0.2722,
      "step": 1072
    },
    {
      "epoch": 13.712460063897764,
      "grad_norm": 3.419447422027588,
      "learning_rate": 1.4497435897435899e-05,
      "loss": 0.2498,
      "step": 1073
    },
    {
      "epoch": 13.725239616613418,
      "grad_norm": 4.817013263702393,
      "learning_rate": 1.4492307692307695e-05,
      "loss": 0.3591,
      "step": 1074
    },
    {
      "epoch": 13.738019169329073,
      "grad_norm": 3.9367573261260986,
      "learning_rate": 1.4487179487179489e-05,
      "loss": 0.4173,
      "step": 1075
    },
    {
      "epoch": 13.750798722044728,
      "grad_norm": 3.777169942855835,
      "learning_rate": 1.4482051282051284e-05,
      "loss": 0.2441,
      "step": 1076
    },
    {
      "epoch": 13.763578274760384,
      "grad_norm": 3.349952220916748,
      "learning_rate": 1.4476923076923077e-05,
      "loss": 0.2305,
      "step": 1077
    },
    {
      "epoch": 13.776357827476039,
      "grad_norm": 3.7569973468780518,
      "learning_rate": 1.4471794871794872e-05,
      "loss": 0.253,
      "step": 1078
    },
    {
      "epoch": 13.789137380191693,
      "grad_norm": 4.3344950675964355,
      "learning_rate": 1.4466666666666668e-05,
      "loss": 0.3236,
      "step": 1079
    },
    {
      "epoch": 13.801916932907348,
      "grad_norm": 4.360065937042236,
      "learning_rate": 1.4461538461538462e-05,
      "loss": 0.3244,
      "step": 1080
    },
    {
      "epoch": 13.814696485623003,
      "grad_norm": 3.8668954372406006,
      "learning_rate": 1.4456410256410258e-05,
      "loss": 0.2798,
      "step": 1081
    },
    {
      "epoch": 13.827476038338657,
      "grad_norm": 4.55317497253418,
      "learning_rate": 1.4451282051282052e-05,
      "loss": 0.322,
      "step": 1082
    },
    {
      "epoch": 13.840255591054314,
      "grad_norm": 4.994724750518799,
      "learning_rate": 1.4446153846153848e-05,
      "loss": 0.2554,
      "step": 1083
    },
    {
      "epoch": 13.853035143769969,
      "grad_norm": 3.47524094581604,
      "learning_rate": 1.4441025641025644e-05,
      "loss": 0.3137,
      "step": 1084
    },
    {
      "epoch": 13.865814696485623,
      "grad_norm": 3.828433036804199,
      "learning_rate": 1.4435897435897438e-05,
      "loss": 0.2692,
      "step": 1085
    },
    {
      "epoch": 13.878594249201278,
      "grad_norm": 3.433011054992676,
      "learning_rate": 1.4430769230769233e-05,
      "loss": 0.3363,
      "step": 1086
    },
    {
      "epoch": 13.891373801916933,
      "grad_norm": 3.745716094970703,
      "learning_rate": 1.4425641025641026e-05,
      "loss": 0.2724,
      "step": 1087
    },
    {
      "epoch": 13.904153354632587,
      "grad_norm": 4.244868755340576,
      "learning_rate": 1.4420512820512821e-05,
      "loss": 0.2699,
      "step": 1088
    },
    {
      "epoch": 13.916932907348244,
      "grad_norm": 3.151378631591797,
      "learning_rate": 1.4415384615384615e-05,
      "loss": 0.2184,
      "step": 1089
    },
    {
      "epoch": 13.929712460063898,
      "grad_norm": 3.5328688621520996,
      "learning_rate": 1.4410256410256411e-05,
      "loss": 0.3188,
      "step": 1090
    },
    {
      "epoch": 13.942492012779553,
      "grad_norm": 3.9043216705322266,
      "learning_rate": 1.4405128205128207e-05,
      "loss": 0.4269,
      "step": 1091
    },
    {
      "epoch": 13.955271565495208,
      "grad_norm": 4.665742874145508,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.3837,
      "step": 1092
    },
    {
      "epoch": 13.968051118210862,
      "grad_norm": 3.9148755073547363,
      "learning_rate": 1.4394871794871797e-05,
      "loss": 0.274,
      "step": 1093
    },
    {
      "epoch": 13.980830670926517,
      "grad_norm": 4.3452630043029785,
      "learning_rate": 1.4389743589743589e-05,
      "loss": 0.3916,
      "step": 1094
    },
    {
      "epoch": 13.993610223642172,
      "grad_norm": 3.2570807933807373,
      "learning_rate": 1.4384615384615387e-05,
      "loss": 0.225,
      "step": 1095
    },
    {
      "epoch": 14.006389776357828,
      "grad_norm": 5.40920352935791,
      "learning_rate": 1.4379487179487182e-05,
      "loss": 0.4157,
      "step": 1096
    },
    {
      "epoch": 14.019169329073483,
      "grad_norm": 3.3941538333892822,
      "learning_rate": 1.4374358974358975e-05,
      "loss": 0.2426,
      "step": 1097
    },
    {
      "epoch": 14.031948881789138,
      "grad_norm": 3.874727964401245,
      "learning_rate": 1.436923076923077e-05,
      "loss": 0.2408,
      "step": 1098
    },
    {
      "epoch": 14.044728434504792,
      "grad_norm": 3.6664178371429443,
      "learning_rate": 1.4364102564102564e-05,
      "loss": 0.3009,
      "step": 1099
    },
    {
      "epoch": 14.057507987220447,
      "grad_norm": 4.098280429840088,
      "learning_rate": 1.435897435897436e-05,
      "loss": 0.2583,
      "step": 1100
    },
    {
      "epoch": 14.057507987220447,
      "eval_loss": 0.5211612582206726,
      "eval_runtime": 183.7005,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 1100
    },
    {
      "epoch": 14.070287539936102,
      "grad_norm": 3.162128210067749,
      "learning_rate": 1.4353846153846156e-05,
      "loss": 0.1974,
      "step": 1101
    },
    {
      "epoch": 14.083067092651758,
      "grad_norm": 3.343733787536621,
      "learning_rate": 1.434871794871795e-05,
      "loss": 0.2123,
      "step": 1102
    },
    {
      "epoch": 14.095846645367413,
      "grad_norm": 4.774056911468506,
      "learning_rate": 1.4343589743589746e-05,
      "loss": 0.3359,
      "step": 1103
    },
    {
      "epoch": 14.108626198083067,
      "grad_norm": 3.923760175704956,
      "learning_rate": 1.4338461538461538e-05,
      "loss": 0.2391,
      "step": 1104
    },
    {
      "epoch": 14.121405750798722,
      "grad_norm": 4.176171779632568,
      "learning_rate": 1.4333333333333334e-05,
      "loss": 0.2406,
      "step": 1105
    },
    {
      "epoch": 14.134185303514377,
      "grad_norm": 4.3658246994018555,
      "learning_rate": 1.432820512820513e-05,
      "loss": 0.2898,
      "step": 1106
    },
    {
      "epoch": 14.146964856230031,
      "grad_norm": 4.221414566040039,
      "learning_rate": 1.4323076923076924e-05,
      "loss": 0.298,
      "step": 1107
    },
    {
      "epoch": 14.159744408945686,
      "grad_norm": 4.855254173278809,
      "learning_rate": 1.431794871794872e-05,
      "loss": 0.2802,
      "step": 1108
    },
    {
      "epoch": 14.172523961661343,
      "grad_norm": 4.903701305389404,
      "learning_rate": 1.4312820512820513e-05,
      "loss": 0.3151,
      "step": 1109
    },
    {
      "epoch": 14.185303514376997,
      "grad_norm": 2.996689796447754,
      "learning_rate": 1.430769230769231e-05,
      "loss": 0.1835,
      "step": 1110
    },
    {
      "epoch": 14.198083067092652,
      "grad_norm": 4.175163745880127,
      "learning_rate": 1.4302564102564103e-05,
      "loss": 0.2826,
      "step": 1111
    },
    {
      "epoch": 14.210862619808307,
      "grad_norm": 4.538963317871094,
      "learning_rate": 1.4297435897435899e-05,
      "loss": 0.2523,
      "step": 1112
    },
    {
      "epoch": 14.223642172523961,
      "grad_norm": 4.176215648651123,
      "learning_rate": 1.4292307692307695e-05,
      "loss": 0.2728,
      "step": 1113
    },
    {
      "epoch": 14.236421725239616,
      "grad_norm": 4.221230983734131,
      "learning_rate": 1.4287179487179487e-05,
      "loss": 0.3101,
      "step": 1114
    },
    {
      "epoch": 14.249201277955272,
      "grad_norm": 5.207468509674072,
      "learning_rate": 1.4282051282051283e-05,
      "loss": 0.2379,
      "step": 1115
    },
    {
      "epoch": 14.261980830670927,
      "grad_norm": 2.8653461933135986,
      "learning_rate": 1.4276923076923077e-05,
      "loss": 0.2439,
      "step": 1116
    },
    {
      "epoch": 14.274760383386582,
      "grad_norm": 3.800055742263794,
      "learning_rate": 1.4271794871794873e-05,
      "loss": 0.2981,
      "step": 1117
    },
    {
      "epoch": 14.287539936102236,
      "grad_norm": 3.1056973934173584,
      "learning_rate": 1.4266666666666668e-05,
      "loss": 0.1567,
      "step": 1118
    },
    {
      "epoch": 14.300319488817891,
      "grad_norm": 3.552155017852783,
      "learning_rate": 1.4261538461538462e-05,
      "loss": 0.2754,
      "step": 1119
    },
    {
      "epoch": 14.313099041533546,
      "grad_norm": 4.416917324066162,
      "learning_rate": 1.4256410256410258e-05,
      "loss": 0.3312,
      "step": 1120
    },
    {
      "epoch": 14.3258785942492,
      "grad_norm": 3.600459098815918,
      "learning_rate": 1.4251282051282052e-05,
      "loss": 0.2085,
      "step": 1121
    },
    {
      "epoch": 14.338658146964857,
      "grad_norm": 3.768317699432373,
      "learning_rate": 1.4246153846153848e-05,
      "loss": 0.2848,
      "step": 1122
    },
    {
      "epoch": 14.351437699680512,
      "grad_norm": 3.480001211166382,
      "learning_rate": 1.4241025641025644e-05,
      "loss": 0.1951,
      "step": 1123
    },
    {
      "epoch": 14.364217252396166,
      "grad_norm": 4.8978495597839355,
      "learning_rate": 1.4235897435897436e-05,
      "loss": 0.3184,
      "step": 1124
    },
    {
      "epoch": 14.37699680511182,
      "grad_norm": 3.9940297603607178,
      "learning_rate": 1.4230769230769232e-05,
      "loss": 0.2288,
      "step": 1125
    },
    {
      "epoch": 14.389776357827476,
      "grad_norm": 3.5290699005126953,
      "learning_rate": 1.4225641025641026e-05,
      "loss": 0.1906,
      "step": 1126
    },
    {
      "epoch": 14.40255591054313,
      "grad_norm": 4.18889856338501,
      "learning_rate": 1.4220512820512822e-05,
      "loss": 0.1804,
      "step": 1127
    },
    {
      "epoch": 14.415335463258787,
      "grad_norm": 4.013011455535889,
      "learning_rate": 1.4215384615384617e-05,
      "loss": 0.3671,
      "step": 1128
    },
    {
      "epoch": 14.428115015974441,
      "grad_norm": 4.15484619140625,
      "learning_rate": 1.4210256410256411e-05,
      "loss": 0.2944,
      "step": 1129
    },
    {
      "epoch": 14.440894568690096,
      "grad_norm": 4.773779392242432,
      "learning_rate": 1.4205128205128207e-05,
      "loss": 0.3174,
      "step": 1130
    },
    {
      "epoch": 14.45367412140575,
      "grad_norm": 5.463952541351318,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 0.2962,
      "step": 1131
    },
    {
      "epoch": 14.466453674121405,
      "grad_norm": 4.950415134429932,
      "learning_rate": 1.4194871794871797e-05,
      "loss": 0.2809,
      "step": 1132
    },
    {
      "epoch": 14.47923322683706,
      "grad_norm": 3.709376096725464,
      "learning_rate": 1.4189743589743593e-05,
      "loss": 0.2346,
      "step": 1133
    },
    {
      "epoch": 14.492012779552716,
      "grad_norm": 3.465916395187378,
      "learning_rate": 1.4184615384615385e-05,
      "loss": 0.212,
      "step": 1134
    },
    {
      "epoch": 14.504792332268371,
      "grad_norm": 4.268784046173096,
      "learning_rate": 1.4179487179487181e-05,
      "loss": 0.2537,
      "step": 1135
    },
    {
      "epoch": 14.517571884984026,
      "grad_norm": 3.6786417961120605,
      "learning_rate": 1.4174358974358975e-05,
      "loss": 0.2346,
      "step": 1136
    },
    {
      "epoch": 14.53035143769968,
      "grad_norm": 4.235080718994141,
      "learning_rate": 1.416923076923077e-05,
      "loss": 0.2716,
      "step": 1137
    },
    {
      "epoch": 14.543130990415335,
      "grad_norm": 4.019659519195557,
      "learning_rate": 1.4164102564102565e-05,
      "loss": 0.1886,
      "step": 1138
    },
    {
      "epoch": 14.55591054313099,
      "grad_norm": 3.9104185104370117,
      "learning_rate": 1.415897435897436e-05,
      "loss": 0.2084,
      "step": 1139
    },
    {
      "epoch": 14.568690095846645,
      "grad_norm": 4.124183177947998,
      "learning_rate": 1.4153846153846156e-05,
      "loss": 0.2415,
      "step": 1140
    },
    {
      "epoch": 14.581469648562301,
      "grad_norm": 3.432605504989624,
      "learning_rate": 1.4148717948717949e-05,
      "loss": 0.2125,
      "step": 1141
    },
    {
      "epoch": 14.594249201277956,
      "grad_norm": 4.036380290985107,
      "learning_rate": 1.4143589743589744e-05,
      "loss": 0.2915,
      "step": 1142
    },
    {
      "epoch": 14.60702875399361,
      "grad_norm": 4.067980766296387,
      "learning_rate": 1.4138461538461538e-05,
      "loss": 0.2747,
      "step": 1143
    },
    {
      "epoch": 14.619808306709265,
      "grad_norm": 3.7604496479034424,
      "learning_rate": 1.4133333333333334e-05,
      "loss": 0.1984,
      "step": 1144
    },
    {
      "epoch": 14.63258785942492,
      "grad_norm": 3.815758466720581,
      "learning_rate": 1.412820512820513e-05,
      "loss": 0.2544,
      "step": 1145
    },
    {
      "epoch": 14.645367412140574,
      "grad_norm": 4.104861259460449,
      "learning_rate": 1.4123076923076924e-05,
      "loss": 0.261,
      "step": 1146
    },
    {
      "epoch": 14.65814696485623,
      "grad_norm": 4.117253303527832,
      "learning_rate": 1.411794871794872e-05,
      "loss": 0.2269,
      "step": 1147
    },
    {
      "epoch": 14.670926517571885,
      "grad_norm": 4.8703837394714355,
      "learning_rate": 1.4112820512820514e-05,
      "loss": 0.3065,
      "step": 1148
    },
    {
      "epoch": 14.68370607028754,
      "grad_norm": 3.6039960384368896,
      "learning_rate": 1.410769230769231e-05,
      "loss": 0.2675,
      "step": 1149
    },
    {
      "epoch": 14.696485623003195,
      "grad_norm": 4.1222615242004395,
      "learning_rate": 1.4102564102564105e-05,
      "loss": 0.2527,
      "step": 1150
    },
    {
      "epoch": 14.70926517571885,
      "grad_norm": 4.146384239196777,
      "learning_rate": 1.4097435897435898e-05,
      "loss": 0.2932,
      "step": 1151
    },
    {
      "epoch": 14.722044728434504,
      "grad_norm": 3.5116586685180664,
      "learning_rate": 1.4092307692307693e-05,
      "loss": 0.1855,
      "step": 1152
    },
    {
      "epoch": 14.73482428115016,
      "grad_norm": 4.162652492523193,
      "learning_rate": 1.4087179487179487e-05,
      "loss": 0.3125,
      "step": 1153
    },
    {
      "epoch": 14.747603833865815,
      "grad_norm": 4.2060747146606445,
      "learning_rate": 1.4082051282051283e-05,
      "loss": 0.2546,
      "step": 1154
    },
    {
      "epoch": 14.76038338658147,
      "grad_norm": 4.2779765129089355,
      "learning_rate": 1.4076923076923079e-05,
      "loss": 0.2141,
      "step": 1155
    },
    {
      "epoch": 14.773162939297125,
      "grad_norm": 3.6725728511810303,
      "learning_rate": 1.4071794871794873e-05,
      "loss": 0.2514,
      "step": 1156
    },
    {
      "epoch": 14.78594249201278,
      "grad_norm": 4.1309356689453125,
      "learning_rate": 1.4066666666666669e-05,
      "loss": 0.2159,
      "step": 1157
    },
    {
      "epoch": 14.798722044728434,
      "grad_norm": 4.970953941345215,
      "learning_rate": 1.4061538461538463e-05,
      "loss": 0.3516,
      "step": 1158
    },
    {
      "epoch": 14.811501597444089,
      "grad_norm": 4.282907009124756,
      "learning_rate": 1.4056410256410258e-05,
      "loss": 0.2957,
      "step": 1159
    },
    {
      "epoch": 14.824281150159745,
      "grad_norm": 4.15264368057251,
      "learning_rate": 1.405128205128205e-05,
      "loss": 0.3114,
      "step": 1160
    },
    {
      "epoch": 14.8370607028754,
      "grad_norm": 6.291984558105469,
      "learning_rate": 1.4046153846153847e-05,
      "loss": 0.3656,
      "step": 1161
    },
    {
      "epoch": 14.849840255591054,
      "grad_norm": 4.7016520500183105,
      "learning_rate": 1.4041025641025642e-05,
      "loss": 0.2923,
      "step": 1162
    },
    {
      "epoch": 14.86261980830671,
      "grad_norm": 4.363107204437256,
      "learning_rate": 1.4035897435897436e-05,
      "loss": 0.2412,
      "step": 1163
    },
    {
      "epoch": 14.875399361022364,
      "grad_norm": 4.666199207305908,
      "learning_rate": 1.4030769230769232e-05,
      "loss": 0.3477,
      "step": 1164
    },
    {
      "epoch": 14.888178913738018,
      "grad_norm": 5.607475280761719,
      "learning_rate": 1.4025641025641026e-05,
      "loss": 0.4089,
      "step": 1165
    },
    {
      "epoch": 14.900958466453675,
      "grad_norm": 4.675595760345459,
      "learning_rate": 1.4020512820512822e-05,
      "loss": 0.247,
      "step": 1166
    },
    {
      "epoch": 14.91373801916933,
      "grad_norm": 4.405561447143555,
      "learning_rate": 1.4015384615384618e-05,
      "loss": 0.2161,
      "step": 1167
    },
    {
      "epoch": 14.926517571884984,
      "grad_norm": 4.1602983474731445,
      "learning_rate": 1.4010256410256412e-05,
      "loss": 0.3375,
      "step": 1168
    },
    {
      "epoch": 14.939297124600639,
      "grad_norm": 4.212582111358643,
      "learning_rate": 1.4005128205128207e-05,
      "loss": 0.2899,
      "step": 1169
    },
    {
      "epoch": 14.952076677316294,
      "grad_norm": 3.6928679943084717,
      "learning_rate": 1.4e-05,
      "loss": 0.2195,
      "step": 1170
    },
    {
      "epoch": 14.964856230031948,
      "grad_norm": 3.720360040664673,
      "learning_rate": 1.3994871794871796e-05,
      "loss": 0.2048,
      "step": 1171
    },
    {
      "epoch": 14.977635782747605,
      "grad_norm": 4.18305778503418,
      "learning_rate": 1.3989743589743591e-05,
      "loss": 0.2354,
      "step": 1172
    },
    {
      "epoch": 14.99041533546326,
      "grad_norm": 3.789397716522217,
      "learning_rate": 1.3984615384615385e-05,
      "loss": 0.2287,
      "step": 1173
    },
    {
      "epoch": 15.003194888178914,
      "grad_norm": 3.868288516998291,
      "learning_rate": 1.3979487179487181e-05,
      "loss": 0.3126,
      "step": 1174
    },
    {
      "epoch": 15.015974440894569,
      "grad_norm": 3.489731550216675,
      "learning_rate": 1.3974358974358975e-05,
      "loss": 0.2087,
      "step": 1175
    },
    {
      "epoch": 15.028753993610223,
      "grad_norm": 3.4861392974853516,
      "learning_rate": 1.3969230769230771e-05,
      "loss": 0.2221,
      "step": 1176
    },
    {
      "epoch": 15.041533546325878,
      "grad_norm": 4.025823593139648,
      "learning_rate": 1.3964102564102567e-05,
      "loss": 0.2304,
      "step": 1177
    },
    {
      "epoch": 15.054313099041533,
      "grad_norm": 3.0816750526428223,
      "learning_rate": 1.3958974358974359e-05,
      "loss": 0.2452,
      "step": 1178
    },
    {
      "epoch": 15.06709265175719,
      "grad_norm": 3.171478033065796,
      "learning_rate": 1.3953846153846156e-05,
      "loss": 0.159,
      "step": 1179
    },
    {
      "epoch": 15.079872204472844,
      "grad_norm": 3.781932830810547,
      "learning_rate": 1.3948717948717949e-05,
      "loss": 0.2649,
      "step": 1180
    },
    {
      "epoch": 15.092651757188499,
      "grad_norm": 3.238219738006592,
      "learning_rate": 1.3943589743589745e-05,
      "loss": 0.2139,
      "step": 1181
    },
    {
      "epoch": 15.105431309904153,
      "grad_norm": 4.112623691558838,
      "learning_rate": 1.393846153846154e-05,
      "loss": 0.2313,
      "step": 1182
    },
    {
      "epoch": 15.118210862619808,
      "grad_norm": 4.599632263183594,
      "learning_rate": 1.3933333333333334e-05,
      "loss": 0.2026,
      "step": 1183
    },
    {
      "epoch": 15.130990415335463,
      "grad_norm": 4.236635684967041,
      "learning_rate": 1.392820512820513e-05,
      "loss": 0.2559,
      "step": 1184
    },
    {
      "epoch": 15.143769968051119,
      "grad_norm": 3.911651134490967,
      "learning_rate": 1.3923076923076924e-05,
      "loss": 0.2067,
      "step": 1185
    },
    {
      "epoch": 15.156549520766774,
      "grad_norm": 3.7214670181274414,
      "learning_rate": 1.391794871794872e-05,
      "loss": 0.1649,
      "step": 1186
    },
    {
      "epoch": 15.169329073482428,
      "grad_norm": 4.420977592468262,
      "learning_rate": 1.3912820512820512e-05,
      "loss": 0.2361,
      "step": 1187
    },
    {
      "epoch": 15.182108626198083,
      "grad_norm": 4.661736011505127,
      "learning_rate": 1.3907692307692308e-05,
      "loss": 0.1881,
      "step": 1188
    },
    {
      "epoch": 15.194888178913738,
      "grad_norm": 4.55980110168457,
      "learning_rate": 1.3902564102564104e-05,
      "loss": 0.2297,
      "step": 1189
    },
    {
      "epoch": 15.207667731629392,
      "grad_norm": 3.6603283882141113,
      "learning_rate": 1.3897435897435898e-05,
      "loss": 0.1636,
      "step": 1190
    },
    {
      "epoch": 15.220447284345047,
      "grad_norm": 4.215869426727295,
      "learning_rate": 1.3892307692307694e-05,
      "loss": 0.2806,
      "step": 1191
    },
    {
      "epoch": 15.233226837060704,
      "grad_norm": 3.951566696166992,
      "learning_rate": 1.3887179487179488e-05,
      "loss": 0.308,
      "step": 1192
    },
    {
      "epoch": 15.246006389776358,
      "grad_norm": 4.275181770324707,
      "learning_rate": 1.3882051282051283e-05,
      "loss": 0.2921,
      "step": 1193
    },
    {
      "epoch": 15.258785942492013,
      "grad_norm": 3.9310081005096436,
      "learning_rate": 1.3876923076923079e-05,
      "loss": 0.2026,
      "step": 1194
    },
    {
      "epoch": 15.271565495207668,
      "grad_norm": 4.194605350494385,
      "learning_rate": 1.3871794871794873e-05,
      "loss": 0.2177,
      "step": 1195
    },
    {
      "epoch": 15.284345047923322,
      "grad_norm": 3.8993043899536133,
      "learning_rate": 1.3866666666666669e-05,
      "loss": 0.2069,
      "step": 1196
    },
    {
      "epoch": 15.297124600638977,
      "grad_norm": 4.589502334594727,
      "learning_rate": 1.3861538461538461e-05,
      "loss": 0.2307,
      "step": 1197
    },
    {
      "epoch": 15.309904153354633,
      "grad_norm": 3.8216519355773926,
      "learning_rate": 1.3856410256410257e-05,
      "loss": 0.2491,
      "step": 1198
    },
    {
      "epoch": 15.322683706070288,
      "grad_norm": 3.8244547843933105,
      "learning_rate": 1.3851282051282053e-05,
      "loss": 0.197,
      "step": 1199
    },
    {
      "epoch": 15.335463258785943,
      "grad_norm": 3.713712215423584,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 0.2318,
      "step": 1200
    },
    {
      "epoch": 15.335463258785943,
      "eval_loss": 0.5259692668914795,
      "eval_runtime": 183.2489,
      "eval_samples_per_second": 0.857,
      "eval_steps_per_second": 0.109,
      "step": 1200
    },
    {
      "epoch": 15.348242811501597,
      "grad_norm": 3.2528011798858643,
      "learning_rate": 1.3841025641025643e-05,
      "loss": 0.2092,
      "step": 1201
    },
    {
      "epoch": 15.361022364217252,
      "grad_norm": 3.8089711666107178,
      "learning_rate": 1.3835897435897437e-05,
      "loss": 0.2623,
      "step": 1202
    },
    {
      "epoch": 15.373801916932907,
      "grad_norm": 3.917395830154419,
      "learning_rate": 1.3830769230769232e-05,
      "loss": 0.2547,
      "step": 1203
    },
    {
      "epoch": 15.386581469648561,
      "grad_norm": 3.8710122108459473,
      "learning_rate": 1.3825641025641028e-05,
      "loss": 0.2426,
      "step": 1204
    },
    {
      "epoch": 15.399361022364218,
      "grad_norm": 3.7556984424591064,
      "learning_rate": 1.3820512820512822e-05,
      "loss": 0.2428,
      "step": 1205
    },
    {
      "epoch": 15.412140575079873,
      "grad_norm": 3.3744430541992188,
      "learning_rate": 1.3815384615384618e-05,
      "loss": 0.2461,
      "step": 1206
    },
    {
      "epoch": 15.424920127795527,
      "grad_norm": 3.5001065731048584,
      "learning_rate": 1.381025641025641e-05,
      "loss": 0.1807,
      "step": 1207
    },
    {
      "epoch": 15.437699680511182,
      "grad_norm": 3.866056442260742,
      "learning_rate": 1.3805128205128206e-05,
      "loss": 0.1739,
      "step": 1208
    },
    {
      "epoch": 15.450479233226837,
      "grad_norm": 4.576041221618652,
      "learning_rate": 1.38e-05,
      "loss": 0.2128,
      "step": 1209
    },
    {
      "epoch": 15.463258785942491,
      "grad_norm": 4.190147876739502,
      "learning_rate": 1.3794871794871796e-05,
      "loss": 0.1893,
      "step": 1210
    },
    {
      "epoch": 15.476038338658148,
      "grad_norm": 4.4232940673828125,
      "learning_rate": 1.3789743589743592e-05,
      "loss": 0.2429,
      "step": 1211
    },
    {
      "epoch": 15.488817891373802,
      "grad_norm": 4.887746810913086,
      "learning_rate": 1.3784615384615386e-05,
      "loss": 0.193,
      "step": 1212
    },
    {
      "epoch": 15.501597444089457,
      "grad_norm": 4.072628498077393,
      "learning_rate": 1.3779487179487181e-05,
      "loss": 0.2494,
      "step": 1213
    },
    {
      "epoch": 15.514376996805112,
      "grad_norm": 4.401451110839844,
      "learning_rate": 1.3774358974358975e-05,
      "loss": 0.2221,
      "step": 1214
    },
    {
      "epoch": 15.527156549520766,
      "grad_norm": 4.5076422691345215,
      "learning_rate": 1.3769230769230771e-05,
      "loss": 0.2881,
      "step": 1215
    },
    {
      "epoch": 15.539936102236421,
      "grad_norm": 4.133837699890137,
      "learning_rate": 1.3764102564102567e-05,
      "loss": 0.1981,
      "step": 1216
    },
    {
      "epoch": 15.552715654952078,
      "grad_norm": 4.182383060455322,
      "learning_rate": 1.375897435897436e-05,
      "loss": 0.2526,
      "step": 1217
    },
    {
      "epoch": 15.565495207667732,
      "grad_norm": 4.154629707336426,
      "learning_rate": 1.3753846153846155e-05,
      "loss": 0.2917,
      "step": 1218
    },
    {
      "epoch": 15.578274760383387,
      "grad_norm": 4.1074347496032715,
      "learning_rate": 1.3748717948717949e-05,
      "loss": 0.181,
      "step": 1219
    },
    {
      "epoch": 15.591054313099042,
      "grad_norm": 4.3682403564453125,
      "learning_rate": 1.3743589743589745e-05,
      "loss": 0.1997,
      "step": 1220
    },
    {
      "epoch": 15.603833865814696,
      "grad_norm": 4.54557466506958,
      "learning_rate": 1.373846153846154e-05,
      "loss": 0.1816,
      "step": 1221
    },
    {
      "epoch": 15.616613418530351,
      "grad_norm": 5.212317943572998,
      "learning_rate": 1.3733333333333335e-05,
      "loss": 0.2666,
      "step": 1222
    },
    {
      "epoch": 15.629392971246006,
      "grad_norm": 3.5865111351013184,
      "learning_rate": 1.372820512820513e-05,
      "loss": 0.1983,
      "step": 1223
    },
    {
      "epoch": 15.642172523961662,
      "grad_norm": 4.672217845916748,
      "learning_rate": 1.3723076923076923e-05,
      "loss": 0.3151,
      "step": 1224
    },
    {
      "epoch": 15.654952076677317,
      "grad_norm": 3.933727741241455,
      "learning_rate": 1.3717948717948718e-05,
      "loss": 0.2512,
      "step": 1225
    },
    {
      "epoch": 15.667731629392971,
      "grad_norm": 3.665114164352417,
      "learning_rate": 1.3712820512820514e-05,
      "loss": 0.1774,
      "step": 1226
    },
    {
      "epoch": 15.680511182108626,
      "grad_norm": 4.567009925842285,
      "learning_rate": 1.3707692307692308e-05,
      "loss": 0.2918,
      "step": 1227
    },
    {
      "epoch": 15.69329073482428,
      "grad_norm": 3.781317949295044,
      "learning_rate": 1.3702564102564104e-05,
      "loss": 0.2409,
      "step": 1228
    },
    {
      "epoch": 15.706070287539935,
      "grad_norm": 4.618795394897461,
      "learning_rate": 1.3697435897435898e-05,
      "loss": 0.29,
      "step": 1229
    },
    {
      "epoch": 15.718849840255592,
      "grad_norm": 4.268044471740723,
      "learning_rate": 1.3692307692307694e-05,
      "loss": 0.2358,
      "step": 1230
    },
    {
      "epoch": 15.731629392971247,
      "grad_norm": 4.229768753051758,
      "learning_rate": 1.3687179487179488e-05,
      "loss": 0.221,
      "step": 1231
    },
    {
      "epoch": 15.744408945686901,
      "grad_norm": 4.791956901550293,
      "learning_rate": 1.3682051282051284e-05,
      "loss": 0.3294,
      "step": 1232
    },
    {
      "epoch": 15.757188498402556,
      "grad_norm": 4.449877738952637,
      "learning_rate": 1.367692307692308e-05,
      "loss": 0.2586,
      "step": 1233
    },
    {
      "epoch": 15.76996805111821,
      "grad_norm": 4.792293071746826,
      "learning_rate": 1.3671794871794872e-05,
      "loss": 0.2943,
      "step": 1234
    },
    {
      "epoch": 15.782747603833865,
      "grad_norm": 3.772984743118286,
      "learning_rate": 1.3666666666666667e-05,
      "loss": 0.2267,
      "step": 1235
    },
    {
      "epoch": 15.795527156549522,
      "grad_norm": 3.729710578918457,
      "learning_rate": 1.3661538461538461e-05,
      "loss": 0.223,
      "step": 1236
    },
    {
      "epoch": 15.808306709265176,
      "grad_norm": 5.369871616363525,
      "learning_rate": 1.3656410256410257e-05,
      "loss": 0.2562,
      "step": 1237
    },
    {
      "epoch": 15.821086261980831,
      "grad_norm": 4.903266906738281,
      "learning_rate": 1.3651282051282053e-05,
      "loss": 0.2558,
      "step": 1238
    },
    {
      "epoch": 15.833865814696486,
      "grad_norm": 3.3962509632110596,
      "learning_rate": 1.3646153846153847e-05,
      "loss": 0.1516,
      "step": 1239
    },
    {
      "epoch": 15.84664536741214,
      "grad_norm": 5.115278244018555,
      "learning_rate": 1.3641025641025643e-05,
      "loss": 0.1924,
      "step": 1240
    },
    {
      "epoch": 15.859424920127795,
      "grad_norm": 4.458291053771973,
      "learning_rate": 1.3635897435897437e-05,
      "loss": 0.3075,
      "step": 1241
    },
    {
      "epoch": 15.87220447284345,
      "grad_norm": 3.402921199798584,
      "learning_rate": 1.3630769230769233e-05,
      "loss": 0.1319,
      "step": 1242
    },
    {
      "epoch": 15.884984025559106,
      "grad_norm": 4.096307277679443,
      "learning_rate": 1.3625641025641028e-05,
      "loss": 0.2711,
      "step": 1243
    },
    {
      "epoch": 15.89776357827476,
      "grad_norm": 4.215847969055176,
      "learning_rate": 1.362051282051282e-05,
      "loss": 0.2353,
      "step": 1244
    },
    {
      "epoch": 15.910543130990416,
      "grad_norm": 4.571593761444092,
      "learning_rate": 1.3615384615384616e-05,
      "loss": 0.2092,
      "step": 1245
    },
    {
      "epoch": 15.92332268370607,
      "grad_norm": 4.602342128753662,
      "learning_rate": 1.361025641025641e-05,
      "loss": 0.351,
      "step": 1246
    },
    {
      "epoch": 15.936102236421725,
      "grad_norm": 4.611629009246826,
      "learning_rate": 1.3605128205128206e-05,
      "loss": 0.1752,
      "step": 1247
    },
    {
      "epoch": 15.94888178913738,
      "grad_norm": 5.004614353179932,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.2447,
      "step": 1248
    },
    {
      "epoch": 15.961661341853034,
      "grad_norm": 4.196335315704346,
      "learning_rate": 1.3594871794871796e-05,
      "loss": 0.2436,
      "step": 1249
    },
    {
      "epoch": 15.97444089456869,
      "grad_norm": 3.1841630935668945,
      "learning_rate": 1.3589743589743592e-05,
      "loss": 0.1514,
      "step": 1250
    },
    {
      "epoch": 15.987220447284345,
      "grad_norm": 3.8978161811828613,
      "learning_rate": 1.3584615384615386e-05,
      "loss": 0.3935,
      "step": 1251
    },
    {
      "epoch": 16.0,
      "grad_norm": 4.464201927185059,
      "learning_rate": 1.3579487179487182e-05,
      "loss": 0.2584,
      "step": 1252
    },
    {
      "epoch": 16.012779552715656,
      "grad_norm": 3.4007415771484375,
      "learning_rate": 1.3574358974358977e-05,
      "loss": 0.2057,
      "step": 1253
    },
    {
      "epoch": 16.02555910543131,
      "grad_norm": 3.3608012199401855,
      "learning_rate": 1.356923076923077e-05,
      "loss": 0.2504,
      "step": 1254
    },
    {
      "epoch": 16.038338658146966,
      "grad_norm": 3.8571906089782715,
      "learning_rate": 1.3564102564102565e-05,
      "loss": 0.1662,
      "step": 1255
    },
    {
      "epoch": 16.05111821086262,
      "grad_norm": 4.310530185699463,
      "learning_rate": 1.355897435897436e-05,
      "loss": 0.2197,
      "step": 1256
    },
    {
      "epoch": 16.063897763578275,
      "grad_norm": 4.129376411437988,
      "learning_rate": 1.3553846153846155e-05,
      "loss": 0.1793,
      "step": 1257
    },
    {
      "epoch": 16.076677316293928,
      "grad_norm": 4.5880045890808105,
      "learning_rate": 1.354871794871795e-05,
      "loss": 0.1637,
      "step": 1258
    },
    {
      "epoch": 16.089456869009584,
      "grad_norm": 34.26468276977539,
      "learning_rate": 1.3543589743589745e-05,
      "loss": 0.2003,
      "step": 1259
    },
    {
      "epoch": 16.10223642172524,
      "grad_norm": 4.134389877319336,
      "learning_rate": 1.353846153846154e-05,
      "loss": 0.191,
      "step": 1260
    },
    {
      "epoch": 16.115015974440894,
      "grad_norm": 5.159793376922607,
      "learning_rate": 1.3533333333333333e-05,
      "loss": 0.1985,
      "step": 1261
    },
    {
      "epoch": 16.12779552715655,
      "grad_norm": 4.2011003494262695,
      "learning_rate": 1.3528205128205129e-05,
      "loss": 0.1624,
      "step": 1262
    },
    {
      "epoch": 16.140575079872203,
      "grad_norm": 4.127978801727295,
      "learning_rate": 1.3523076923076923e-05,
      "loss": 0.1746,
      "step": 1263
    },
    {
      "epoch": 16.15335463258786,
      "grad_norm": 4.027311325073242,
      "learning_rate": 1.3517948717948719e-05,
      "loss": 0.1958,
      "step": 1264
    },
    {
      "epoch": 16.166134185303516,
      "grad_norm": 4.948392868041992,
      "learning_rate": 1.3512820512820514e-05,
      "loss": 0.2804,
      "step": 1265
    },
    {
      "epoch": 16.17891373801917,
      "grad_norm": 3.5425753593444824,
      "learning_rate": 1.3507692307692308e-05,
      "loss": 0.132,
      "step": 1266
    },
    {
      "epoch": 16.191693290734825,
      "grad_norm": 4.664565563201904,
      "learning_rate": 1.3502564102564104e-05,
      "loss": 0.2148,
      "step": 1267
    },
    {
      "epoch": 16.20447284345048,
      "grad_norm": 4.611464977264404,
      "learning_rate": 1.3497435897435898e-05,
      "loss": 0.2382,
      "step": 1268
    },
    {
      "epoch": 16.217252396166135,
      "grad_norm": 3.5738282203674316,
      "learning_rate": 1.3492307692307694e-05,
      "loss": 0.1605,
      "step": 1269
    },
    {
      "epoch": 16.230031948881788,
      "grad_norm": 3.991009473800659,
      "learning_rate": 1.348717948717949e-05,
      "loss": 0.1839,
      "step": 1270
    },
    {
      "epoch": 16.242811501597444,
      "grad_norm": 4.211781024932861,
      "learning_rate": 1.3482051282051282e-05,
      "loss": 0.2611,
      "step": 1271
    },
    {
      "epoch": 16.2555910543131,
      "grad_norm": 3.7441012859344482,
      "learning_rate": 1.3476923076923078e-05,
      "loss": 0.1583,
      "step": 1272
    },
    {
      "epoch": 16.268370607028753,
      "grad_norm": 5.4371490478515625,
      "learning_rate": 1.3471794871794872e-05,
      "loss": 0.3684,
      "step": 1273
    },
    {
      "epoch": 16.28115015974441,
      "grad_norm": 3.0008156299591064,
      "learning_rate": 1.3466666666666668e-05,
      "loss": 0.1685,
      "step": 1274
    },
    {
      "epoch": 16.293929712460063,
      "grad_norm": 4.355956554412842,
      "learning_rate": 1.3461538461538463e-05,
      "loss": 0.1921,
      "step": 1275
    },
    {
      "epoch": 16.30670926517572,
      "grad_norm": 3.5138537883758545,
      "learning_rate": 1.3456410256410257e-05,
      "loss": 0.136,
      "step": 1276
    },
    {
      "epoch": 16.319488817891372,
      "grad_norm": 3.696995258331299,
      "learning_rate": 1.3451282051282053e-05,
      "loss": 0.1814,
      "step": 1277
    },
    {
      "epoch": 16.33226837060703,
      "grad_norm": 4.32089376449585,
      "learning_rate": 1.3446153846153847e-05,
      "loss": 0.2155,
      "step": 1278
    },
    {
      "epoch": 16.345047923322685,
      "grad_norm": 3.804948568344116,
      "learning_rate": 1.3441025641025643e-05,
      "loss": 0.1293,
      "step": 1279
    },
    {
      "epoch": 16.357827476038338,
      "grad_norm": 3.8062856197357178,
      "learning_rate": 1.3435897435897435e-05,
      "loss": 0.1458,
      "step": 1280
    },
    {
      "epoch": 16.370607028753994,
      "grad_norm": 4.169013500213623,
      "learning_rate": 1.3430769230769231e-05,
      "loss": 0.2107,
      "step": 1281
    },
    {
      "epoch": 16.383386581469647,
      "grad_norm": 5.1493306159973145,
      "learning_rate": 1.3425641025641027e-05,
      "loss": 0.1938,
      "step": 1282
    },
    {
      "epoch": 16.396166134185304,
      "grad_norm": 4.721076011657715,
      "learning_rate": 1.3420512820512821e-05,
      "loss": 0.3148,
      "step": 1283
    },
    {
      "epoch": 16.408945686900957,
      "grad_norm": 4.6146321296691895,
      "learning_rate": 1.3415384615384617e-05,
      "loss": 0.2838,
      "step": 1284
    },
    {
      "epoch": 16.421725239616613,
      "grad_norm": 3.725146770477295,
      "learning_rate": 1.341025641025641e-05,
      "loss": 0.1809,
      "step": 1285
    },
    {
      "epoch": 16.43450479233227,
      "grad_norm": 4.293447017669678,
      "learning_rate": 1.3405128205128206e-05,
      "loss": 0.2271,
      "step": 1286
    },
    {
      "epoch": 16.447284345047922,
      "grad_norm": 4.092364311218262,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.1976,
      "step": 1287
    },
    {
      "epoch": 16.46006389776358,
      "grad_norm": 5.2461161613464355,
      "learning_rate": 1.3394871794871796e-05,
      "loss": 0.2599,
      "step": 1288
    },
    {
      "epoch": 16.472843450479232,
      "grad_norm": 3.7622294425964355,
      "learning_rate": 1.3389743589743592e-05,
      "loss": 0.193,
      "step": 1289
    },
    {
      "epoch": 16.48562300319489,
      "grad_norm": 3.409365653991699,
      "learning_rate": 1.3384615384615384e-05,
      "loss": 0.1305,
      "step": 1290
    },
    {
      "epoch": 16.498402555910545,
      "grad_norm": 3.733654737472534,
      "learning_rate": 1.337948717948718e-05,
      "loss": 0.2246,
      "step": 1291
    },
    {
      "epoch": 16.511182108626198,
      "grad_norm": 4.387030124664307,
      "learning_rate": 1.3374358974358976e-05,
      "loss": 0.2397,
      "step": 1292
    },
    {
      "epoch": 16.523961661341854,
      "grad_norm": 3.9117209911346436,
      "learning_rate": 1.336923076923077e-05,
      "loss": 0.1676,
      "step": 1293
    },
    {
      "epoch": 16.536741214057507,
      "grad_norm": 4.131397724151611,
      "learning_rate": 1.3364102564102566e-05,
      "loss": 0.2039,
      "step": 1294
    },
    {
      "epoch": 16.549520766773163,
      "grad_norm": 3.769676446914673,
      "learning_rate": 1.335897435897436e-05,
      "loss": 0.1712,
      "step": 1295
    },
    {
      "epoch": 16.562300319488816,
      "grad_norm": 4.4638848304748535,
      "learning_rate": 1.3353846153846155e-05,
      "loss": 0.1844,
      "step": 1296
    },
    {
      "epoch": 16.575079872204473,
      "grad_norm": 4.5951361656188965,
      "learning_rate": 1.3348717948717951e-05,
      "loss": 0.1891,
      "step": 1297
    },
    {
      "epoch": 16.58785942492013,
      "grad_norm": 4.4006667137146,
      "learning_rate": 1.3343589743589745e-05,
      "loss": 0.2095,
      "step": 1298
    },
    {
      "epoch": 16.600638977635782,
      "grad_norm": 5.249375820159912,
      "learning_rate": 1.3338461538461541e-05,
      "loss": 0.2312,
      "step": 1299
    },
    {
      "epoch": 16.61341853035144,
      "grad_norm": 4.6556243896484375,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.297,
      "step": 1300
    },
    {
      "epoch": 16.61341853035144,
      "eval_loss": 0.5607043504714966,
      "eval_runtime": 183.5728,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 1300
    },
    {
      "epoch": 16.62619808306709,
      "grad_norm": 4.347846031188965,
      "learning_rate": 1.3328205128205129e-05,
      "loss": 0.2292,
      "step": 1301
    },
    {
      "epoch": 16.638977635782748,
      "grad_norm": 4.716831684112549,
      "learning_rate": 1.3323076923076925e-05,
      "loss": 0.1874,
      "step": 1302
    },
    {
      "epoch": 16.6517571884984,
      "grad_norm": 3.6589035987854004,
      "learning_rate": 1.3317948717948719e-05,
      "loss": 0.1601,
      "step": 1303
    },
    {
      "epoch": 16.664536741214057,
      "grad_norm": 3.7984180450439453,
      "learning_rate": 1.3312820512820515e-05,
      "loss": 0.2677,
      "step": 1304
    },
    {
      "epoch": 16.677316293929714,
      "grad_norm": 3.922208309173584,
      "learning_rate": 1.3307692307692309e-05,
      "loss": 0.1818,
      "step": 1305
    },
    {
      "epoch": 16.690095846645367,
      "grad_norm": 4.426130294799805,
      "learning_rate": 1.3302564102564104e-05,
      "loss": 0.2011,
      "step": 1306
    },
    {
      "epoch": 16.702875399361023,
      "grad_norm": 4.943461894989014,
      "learning_rate": 1.3297435897435897e-05,
      "loss": 0.3119,
      "step": 1307
    },
    {
      "epoch": 16.715654952076676,
      "grad_norm": 3.7821176052093506,
      "learning_rate": 1.3292307692307692e-05,
      "loss": 0.13,
      "step": 1308
    },
    {
      "epoch": 16.728434504792332,
      "grad_norm": 3.5825772285461426,
      "learning_rate": 1.3287179487179488e-05,
      "loss": 0.1673,
      "step": 1309
    },
    {
      "epoch": 16.74121405750799,
      "grad_norm": 5.194157600402832,
      "learning_rate": 1.3282051282051282e-05,
      "loss": 0.1959,
      "step": 1310
    },
    {
      "epoch": 16.75399361022364,
      "grad_norm": 5.931427001953125,
      "learning_rate": 1.3276923076923078e-05,
      "loss": 0.3577,
      "step": 1311
    },
    {
      "epoch": 16.766773162939298,
      "grad_norm": 4.4217071533203125,
      "learning_rate": 1.3271794871794872e-05,
      "loss": 0.2588,
      "step": 1312
    },
    {
      "epoch": 16.77955271565495,
      "grad_norm": 4.538804054260254,
      "learning_rate": 1.3266666666666668e-05,
      "loss": 0.2171,
      "step": 1313
    },
    {
      "epoch": 16.792332268370608,
      "grad_norm": 5.510791778564453,
      "learning_rate": 1.3261538461538464e-05,
      "loss": 0.345,
      "step": 1314
    },
    {
      "epoch": 16.80511182108626,
      "grad_norm": 4.038776874542236,
      "learning_rate": 1.3256410256410258e-05,
      "loss": 0.1822,
      "step": 1315
    },
    {
      "epoch": 16.817891373801917,
      "grad_norm": 4.527102947235107,
      "learning_rate": 1.3251282051282053e-05,
      "loss": 0.2724,
      "step": 1316
    },
    {
      "epoch": 16.830670926517573,
      "grad_norm": 4.719036102294922,
      "learning_rate": 1.3246153846153846e-05,
      "loss": 0.2073,
      "step": 1317
    },
    {
      "epoch": 16.843450479233226,
      "grad_norm": 5.0867791175842285,
      "learning_rate": 1.3241025641025641e-05,
      "loss": 0.2437,
      "step": 1318
    },
    {
      "epoch": 16.856230031948883,
      "grad_norm": 3.9539568424224854,
      "learning_rate": 1.3235897435897437e-05,
      "loss": 0.1762,
      "step": 1319
    },
    {
      "epoch": 16.869009584664536,
      "grad_norm": 3.6446242332458496,
      "learning_rate": 1.3230769230769231e-05,
      "loss": 0.1671,
      "step": 1320
    },
    {
      "epoch": 16.881789137380192,
      "grad_norm": 4.763880729675293,
      "learning_rate": 1.3225641025641027e-05,
      "loss": 0.2316,
      "step": 1321
    },
    {
      "epoch": 16.894568690095845,
      "grad_norm": 3.9518229961395264,
      "learning_rate": 1.3220512820512821e-05,
      "loss": 0.1966,
      "step": 1322
    },
    {
      "epoch": 16.9073482428115,
      "grad_norm": 3.724243402481079,
      "learning_rate": 1.3215384615384617e-05,
      "loss": 0.1164,
      "step": 1323
    },
    {
      "epoch": 16.920127795527158,
      "grad_norm": 3.9408881664276123,
      "learning_rate": 1.3210256410256413e-05,
      "loss": 0.1555,
      "step": 1324
    },
    {
      "epoch": 16.93290734824281,
      "grad_norm": 4.742832183837891,
      "learning_rate": 1.3205128205128207e-05,
      "loss": 0.256,
      "step": 1325
    },
    {
      "epoch": 16.945686900958467,
      "grad_norm": 4.283969879150391,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.19,
      "step": 1326
    },
    {
      "epoch": 16.95846645367412,
      "grad_norm": 4.325873374938965,
      "learning_rate": 1.3194871794871795e-05,
      "loss": 0.2203,
      "step": 1327
    },
    {
      "epoch": 16.971246006389777,
      "grad_norm": 4.042814254760742,
      "learning_rate": 1.318974358974359e-05,
      "loss": 0.2201,
      "step": 1328
    },
    {
      "epoch": 16.984025559105433,
      "grad_norm": 4.018223285675049,
      "learning_rate": 1.3184615384615385e-05,
      "loss": 0.1951,
      "step": 1329
    },
    {
      "epoch": 16.996805111821086,
      "grad_norm": 4.505209922790527,
      "learning_rate": 1.317948717948718e-05,
      "loss": 0.234,
      "step": 1330
    },
    {
      "epoch": 17.009584664536742,
      "grad_norm": 4.061479091644287,
      "learning_rate": 1.3174358974358976e-05,
      "loss": 0.2135,
      "step": 1331
    },
    {
      "epoch": 17.022364217252395,
      "grad_norm": 4.851925373077393,
      "learning_rate": 1.316923076923077e-05,
      "loss": 0.2187,
      "step": 1332
    },
    {
      "epoch": 17.03514376996805,
      "grad_norm": 3.1499011516571045,
      "learning_rate": 1.3164102564102566e-05,
      "loss": 0.1274,
      "step": 1333
    },
    {
      "epoch": 17.047923322683705,
      "grad_norm": 4.1169233322143555,
      "learning_rate": 1.315897435897436e-05,
      "loss": 0.1484,
      "step": 1334
    },
    {
      "epoch": 17.06070287539936,
      "grad_norm": 4.023073673248291,
      "learning_rate": 1.3153846153846156e-05,
      "loss": 0.1449,
      "step": 1335
    },
    {
      "epoch": 17.073482428115017,
      "grad_norm": 3.9548263549804688,
      "learning_rate": 1.3148717948717951e-05,
      "loss": 0.11,
      "step": 1336
    },
    {
      "epoch": 17.08626198083067,
      "grad_norm": 3.6275691986083984,
      "learning_rate": 1.3143589743589744e-05,
      "loss": 0.1581,
      "step": 1337
    },
    {
      "epoch": 17.099041533546327,
      "grad_norm": 3.9273226261138916,
      "learning_rate": 1.313846153846154e-05,
      "loss": 0.137,
      "step": 1338
    },
    {
      "epoch": 17.11182108626198,
      "grad_norm": 3.3632616996765137,
      "learning_rate": 1.3133333333333334e-05,
      "loss": 0.1551,
      "step": 1339
    },
    {
      "epoch": 17.124600638977636,
      "grad_norm": 4.732468605041504,
      "learning_rate": 1.312820512820513e-05,
      "loss": 0.2677,
      "step": 1340
    },
    {
      "epoch": 17.13738019169329,
      "grad_norm": 4.406965255737305,
      "learning_rate": 1.3123076923076925e-05,
      "loss": 0.1712,
      "step": 1341
    },
    {
      "epoch": 17.150159744408946,
      "grad_norm": 4.088700294494629,
      "learning_rate": 1.3117948717948719e-05,
      "loss": 0.2075,
      "step": 1342
    },
    {
      "epoch": 17.162939297124602,
      "grad_norm": 3.2795867919921875,
      "learning_rate": 1.3112820512820515e-05,
      "loss": 0.132,
      "step": 1343
    },
    {
      "epoch": 17.175718849840255,
      "grad_norm": 4.069875240325928,
      "learning_rate": 1.3107692307692307e-05,
      "loss": 0.1838,
      "step": 1344
    },
    {
      "epoch": 17.18849840255591,
      "grad_norm": 4.887452602386475,
      "learning_rate": 1.3102564102564103e-05,
      "loss": 0.2012,
      "step": 1345
    },
    {
      "epoch": 17.201277955271564,
      "grad_norm": 5.204954147338867,
      "learning_rate": 1.3097435897435899e-05,
      "loss": 0.1768,
      "step": 1346
    },
    {
      "epoch": 17.21405750798722,
      "grad_norm": 4.781338691711426,
      "learning_rate": 1.3092307692307693e-05,
      "loss": 0.2594,
      "step": 1347
    },
    {
      "epoch": 17.226837060702877,
      "grad_norm": 4.1371870040893555,
      "learning_rate": 1.3087179487179488e-05,
      "loss": 0.1797,
      "step": 1348
    },
    {
      "epoch": 17.23961661341853,
      "grad_norm": 4.221982002258301,
      "learning_rate": 1.3082051282051283e-05,
      "loss": 0.1579,
      "step": 1349
    },
    {
      "epoch": 17.252396166134186,
      "grad_norm": 4.641051769256592,
      "learning_rate": 1.3076923076923078e-05,
      "loss": 0.2327,
      "step": 1350
    },
    {
      "epoch": 17.26517571884984,
      "grad_norm": 4.150181293487549,
      "learning_rate": 1.3071794871794874e-05,
      "loss": 0.1501,
      "step": 1351
    },
    {
      "epoch": 17.277955271565496,
      "grad_norm": 4.188239097595215,
      "learning_rate": 1.3066666666666668e-05,
      "loss": 0.1982,
      "step": 1352
    },
    {
      "epoch": 17.29073482428115,
      "grad_norm": 3.598334312438965,
      "learning_rate": 1.3061538461538464e-05,
      "loss": 0.1179,
      "step": 1353
    },
    {
      "epoch": 17.303514376996805,
      "grad_norm": 4.53875732421875,
      "learning_rate": 1.3056410256410256e-05,
      "loss": 0.1706,
      "step": 1354
    },
    {
      "epoch": 17.31629392971246,
      "grad_norm": 3.595996379852295,
      "learning_rate": 1.3051282051282052e-05,
      "loss": 0.1708,
      "step": 1355
    },
    {
      "epoch": 17.329073482428115,
      "grad_norm": 3.832343578338623,
      "learning_rate": 1.3046153846153846e-05,
      "loss": 0.1785,
      "step": 1356
    },
    {
      "epoch": 17.34185303514377,
      "grad_norm": 5.0454583168029785,
      "learning_rate": 1.3041025641025642e-05,
      "loss": 0.1427,
      "step": 1357
    },
    {
      "epoch": 17.354632587859424,
      "grad_norm": 4.708250999450684,
      "learning_rate": 1.3035897435897437e-05,
      "loss": 0.1731,
      "step": 1358
    },
    {
      "epoch": 17.36741214057508,
      "grad_norm": 3.5192267894744873,
      "learning_rate": 1.3030769230769231e-05,
      "loss": 0.1117,
      "step": 1359
    },
    {
      "epoch": 17.380191693290733,
      "grad_norm": 4.695788860321045,
      "learning_rate": 1.3025641025641027e-05,
      "loss": 0.2105,
      "step": 1360
    },
    {
      "epoch": 17.39297124600639,
      "grad_norm": 4.710606098175049,
      "learning_rate": 1.3020512820512821e-05,
      "loss": 0.1613,
      "step": 1361
    },
    {
      "epoch": 17.405750798722046,
      "grad_norm": 3.6666271686553955,
      "learning_rate": 1.3015384615384617e-05,
      "loss": 0.179,
      "step": 1362
    },
    {
      "epoch": 17.4185303514377,
      "grad_norm": 3.9461636543273926,
      "learning_rate": 1.3010256410256413e-05,
      "loss": 0.1426,
      "step": 1363
    },
    {
      "epoch": 17.431309904153355,
      "grad_norm": 3.5908493995666504,
      "learning_rate": 1.3005128205128205e-05,
      "loss": 0.1152,
      "step": 1364
    },
    {
      "epoch": 17.44408945686901,
      "grad_norm": 4.1953959465026855,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.1717,
      "step": 1365
    },
    {
      "epoch": 17.456869009584665,
      "grad_norm": 5.178595542907715,
      "learning_rate": 1.2994871794871795e-05,
      "loss": 0.1723,
      "step": 1366
    },
    {
      "epoch": 17.46964856230032,
      "grad_norm": 4.45409631729126,
      "learning_rate": 1.298974358974359e-05,
      "loss": 0.2075,
      "step": 1367
    },
    {
      "epoch": 17.482428115015974,
      "grad_norm": 6.469843864440918,
      "learning_rate": 1.2984615384615386e-05,
      "loss": 0.2422,
      "step": 1368
    },
    {
      "epoch": 17.49520766773163,
      "grad_norm": 3.8967084884643555,
      "learning_rate": 1.297948717948718e-05,
      "loss": 0.1658,
      "step": 1369
    },
    {
      "epoch": 17.507987220447284,
      "grad_norm": 4.227680683135986,
      "learning_rate": 1.2974358974358976e-05,
      "loss": 0.1941,
      "step": 1370
    },
    {
      "epoch": 17.52076677316294,
      "grad_norm": 4.642873287200928,
      "learning_rate": 1.296923076923077e-05,
      "loss": 0.1918,
      "step": 1371
    },
    {
      "epoch": 17.533546325878593,
      "grad_norm": 4.665454864501953,
      "learning_rate": 1.2964102564102566e-05,
      "loss": 0.2758,
      "step": 1372
    },
    {
      "epoch": 17.54632587859425,
      "grad_norm": 4.068156719207764,
      "learning_rate": 1.2958974358974362e-05,
      "loss": 0.1334,
      "step": 1373
    },
    {
      "epoch": 17.559105431309906,
      "grad_norm": 4.479705810546875,
      "learning_rate": 1.2953846153846154e-05,
      "loss": 0.1793,
      "step": 1374
    },
    {
      "epoch": 17.57188498402556,
      "grad_norm": 4.446591377258301,
      "learning_rate": 1.294871794871795e-05,
      "loss": 0.1761,
      "step": 1375
    },
    {
      "epoch": 17.584664536741215,
      "grad_norm": 5.095440864562988,
      "learning_rate": 1.2943589743589744e-05,
      "loss": 0.1738,
      "step": 1376
    },
    {
      "epoch": 17.597444089456868,
      "grad_norm": 5.519388675689697,
      "learning_rate": 1.293846153846154e-05,
      "loss": 0.2111,
      "step": 1377
    },
    {
      "epoch": 17.610223642172524,
      "grad_norm": 4.9799675941467285,
      "learning_rate": 1.2933333333333334e-05,
      "loss": 0.2324,
      "step": 1378
    },
    {
      "epoch": 17.623003194888177,
      "grad_norm": 3.9601807594299316,
      "learning_rate": 1.292820512820513e-05,
      "loss": 0.1808,
      "step": 1379
    },
    {
      "epoch": 17.635782747603834,
      "grad_norm": 2.8730149269104004,
      "learning_rate": 1.2923076923076925e-05,
      "loss": 0.0883,
      "step": 1380
    },
    {
      "epoch": 17.64856230031949,
      "grad_norm": 3.890813112258911,
      "learning_rate": 1.2917948717948718e-05,
      "loss": 0.1648,
      "step": 1381
    },
    {
      "epoch": 17.661341853035143,
      "grad_norm": 4.435185432434082,
      "learning_rate": 1.2912820512820515e-05,
      "loss": 0.2343,
      "step": 1382
    },
    {
      "epoch": 17.6741214057508,
      "grad_norm": 4.9541096687316895,
      "learning_rate": 1.2907692307692307e-05,
      "loss": 0.156,
      "step": 1383
    },
    {
      "epoch": 17.686900958466452,
      "grad_norm": 4.387731552124023,
      "learning_rate": 1.2902564102564103e-05,
      "loss": 0.2392,
      "step": 1384
    },
    {
      "epoch": 17.69968051118211,
      "grad_norm": 5.2987284660339355,
      "learning_rate": 1.2897435897435899e-05,
      "loss": 0.1897,
      "step": 1385
    },
    {
      "epoch": 17.712460063897765,
      "grad_norm": 3.6852712631225586,
      "learning_rate": 1.2892307692307693e-05,
      "loss": 0.1852,
      "step": 1386
    },
    {
      "epoch": 17.72523961661342,
      "grad_norm": 4.723902702331543,
      "learning_rate": 1.2887179487179489e-05,
      "loss": 0.1622,
      "step": 1387
    },
    {
      "epoch": 17.738019169329075,
      "grad_norm": 4.313000202178955,
      "learning_rate": 1.2882051282051283e-05,
      "loss": 0.1296,
      "step": 1388
    },
    {
      "epoch": 17.750798722044728,
      "grad_norm": 4.494014263153076,
      "learning_rate": 1.2876923076923078e-05,
      "loss": 0.1799,
      "step": 1389
    },
    {
      "epoch": 17.763578274760384,
      "grad_norm": 4.625412940979004,
      "learning_rate": 1.2871794871794874e-05,
      "loss": 0.207,
      "step": 1390
    },
    {
      "epoch": 17.776357827476037,
      "grad_norm": 4.199818134307861,
      "learning_rate": 1.2866666666666667e-05,
      "loss": 0.1388,
      "step": 1391
    },
    {
      "epoch": 17.789137380191693,
      "grad_norm": 4.411600112915039,
      "learning_rate": 1.2861538461538462e-05,
      "loss": 0.2175,
      "step": 1392
    },
    {
      "epoch": 17.80191693290735,
      "grad_norm": 4.975483417510986,
      "learning_rate": 1.2856410256410256e-05,
      "loss": 0.2334,
      "step": 1393
    },
    {
      "epoch": 17.814696485623003,
      "grad_norm": 4.207812786102295,
      "learning_rate": 1.2851282051282052e-05,
      "loss": 0.1924,
      "step": 1394
    },
    {
      "epoch": 17.82747603833866,
      "grad_norm": 5.613093852996826,
      "learning_rate": 1.2846153846153848e-05,
      "loss": 0.2351,
      "step": 1395
    },
    {
      "epoch": 17.840255591054312,
      "grad_norm": 5.09583854675293,
      "learning_rate": 1.2841025641025642e-05,
      "loss": 0.2433,
      "step": 1396
    },
    {
      "epoch": 17.85303514376997,
      "grad_norm": 5.236515522003174,
      "learning_rate": 1.2835897435897438e-05,
      "loss": 0.2328,
      "step": 1397
    },
    {
      "epoch": 17.86581469648562,
      "grad_norm": 4.290408611297607,
      "learning_rate": 1.2830769230769232e-05,
      "loss": 0.1266,
      "step": 1398
    },
    {
      "epoch": 17.878594249201278,
      "grad_norm": 4.449280261993408,
      "learning_rate": 1.2825641025641027e-05,
      "loss": 0.1595,
      "step": 1399
    },
    {
      "epoch": 17.891373801916934,
      "grad_norm": 5.017588138580322,
      "learning_rate": 1.2820512820512823e-05,
      "loss": 0.1671,
      "step": 1400
    },
    {
      "epoch": 17.891373801916934,
      "eval_loss": 0.5858176946640015,
      "eval_runtime": 183.6009,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 1400
    },
    {
      "epoch": 17.904153354632587,
      "grad_norm": 3.8054473400115967,
      "learning_rate": 1.2815384615384616e-05,
      "loss": 0.1687,
      "step": 1401
    },
    {
      "epoch": 17.916932907348244,
      "grad_norm": 4.932668685913086,
      "learning_rate": 1.2810256410256411e-05,
      "loss": 0.1664,
      "step": 1402
    },
    {
      "epoch": 17.929712460063897,
      "grad_norm": 4.4709391593933105,
      "learning_rate": 1.2805128205128205e-05,
      "loss": 0.2496,
      "step": 1403
    },
    {
      "epoch": 17.942492012779553,
      "grad_norm": 5.316714286804199,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.2514,
      "step": 1404
    },
    {
      "epoch": 17.955271565495206,
      "grad_norm": 6.395861625671387,
      "learning_rate": 1.2794871794871795e-05,
      "loss": 0.2261,
      "step": 1405
    },
    {
      "epoch": 17.968051118210862,
      "grad_norm": 5.766951560974121,
      "learning_rate": 1.2789743589743591e-05,
      "loss": 0.156,
      "step": 1406
    },
    {
      "epoch": 17.98083067092652,
      "grad_norm": 4.028629302978516,
      "learning_rate": 1.2784615384615387e-05,
      "loss": 0.1327,
      "step": 1407
    },
    {
      "epoch": 17.99361022364217,
      "grad_norm": 4.956418991088867,
      "learning_rate": 1.277948717948718e-05,
      "loss": 0.1746,
      "step": 1408
    },
    {
      "epoch": 18.00638977635783,
      "grad_norm": 4.1461052894592285,
      "learning_rate": 1.2774358974358976e-05,
      "loss": 0.1636,
      "step": 1409
    },
    {
      "epoch": 18.01916932907348,
      "grad_norm": 3.723752021789551,
      "learning_rate": 1.2769230769230769e-05,
      "loss": 0.2003,
      "step": 1410
    },
    {
      "epoch": 18.031948881789138,
      "grad_norm": 3.700064182281494,
      "learning_rate": 1.2764102564102565e-05,
      "loss": 0.1339,
      "step": 1411
    },
    {
      "epoch": 18.044728434504794,
      "grad_norm": 3.3118534088134766,
      "learning_rate": 1.275897435897436e-05,
      "loss": 0.1082,
      "step": 1412
    },
    {
      "epoch": 18.057507987220447,
      "grad_norm": 3.8128724098205566,
      "learning_rate": 1.2753846153846154e-05,
      "loss": 0.1564,
      "step": 1413
    },
    {
      "epoch": 18.070287539936103,
      "grad_norm": 3.9735348224639893,
      "learning_rate": 1.274871794871795e-05,
      "loss": 0.1304,
      "step": 1414
    },
    {
      "epoch": 18.083067092651756,
      "grad_norm": 4.680019855499268,
      "learning_rate": 1.2743589743589744e-05,
      "loss": 0.1849,
      "step": 1415
    },
    {
      "epoch": 18.095846645367413,
      "grad_norm": 4.512134075164795,
      "learning_rate": 1.273846153846154e-05,
      "loss": 0.2095,
      "step": 1416
    },
    {
      "epoch": 18.108626198083066,
      "grad_norm": 4.275150299072266,
      "learning_rate": 1.2733333333333336e-05,
      "loss": 0.1757,
      "step": 1417
    },
    {
      "epoch": 18.121405750798722,
      "grad_norm": 3.94869327545166,
      "learning_rate": 1.272820512820513e-05,
      "loss": 0.104,
      "step": 1418
    },
    {
      "epoch": 18.13418530351438,
      "grad_norm": 4.781008243560791,
      "learning_rate": 1.2723076923076925e-05,
      "loss": 0.1372,
      "step": 1419
    },
    {
      "epoch": 18.14696485623003,
      "grad_norm": 4.485940456390381,
      "learning_rate": 1.2717948717948718e-05,
      "loss": 0.1488,
      "step": 1420
    },
    {
      "epoch": 18.159744408945688,
      "grad_norm": 2.775332450866699,
      "learning_rate": 1.2712820512820514e-05,
      "loss": 0.1255,
      "step": 1421
    },
    {
      "epoch": 18.17252396166134,
      "grad_norm": 3.302737236022949,
      "learning_rate": 1.270769230769231e-05,
      "loss": 0.1398,
      "step": 1422
    },
    {
      "epoch": 18.185303514376997,
      "grad_norm": 5.000697612762451,
      "learning_rate": 1.2702564102564103e-05,
      "loss": 0.1503,
      "step": 1423
    },
    {
      "epoch": 18.19808306709265,
      "grad_norm": 4.334658622741699,
      "learning_rate": 1.2697435897435899e-05,
      "loss": 0.1535,
      "step": 1424
    },
    {
      "epoch": 18.210862619808307,
      "grad_norm": 4.618746280670166,
      "learning_rate": 1.2692307692307693e-05,
      "loss": 0.1605,
      "step": 1425
    },
    {
      "epoch": 18.223642172523963,
      "grad_norm": 4.05863094329834,
      "learning_rate": 1.2687179487179489e-05,
      "loss": 0.0956,
      "step": 1426
    },
    {
      "epoch": 18.236421725239616,
      "grad_norm": 4.230029106140137,
      "learning_rate": 1.2682051282051281e-05,
      "loss": 0.1435,
      "step": 1427
    },
    {
      "epoch": 18.249201277955272,
      "grad_norm": 2.6787288188934326,
      "learning_rate": 1.2676923076923077e-05,
      "loss": 0.0948,
      "step": 1428
    },
    {
      "epoch": 18.261980830670925,
      "grad_norm": 4.375114440917969,
      "learning_rate": 1.2671794871794873e-05,
      "loss": 0.1843,
      "step": 1429
    },
    {
      "epoch": 18.27476038338658,
      "grad_norm": 3.706968069076538,
      "learning_rate": 1.2666666666666667e-05,
      "loss": 0.1272,
      "step": 1430
    },
    {
      "epoch": 18.287539936102238,
      "grad_norm": 4.491248607635498,
      "learning_rate": 1.2661538461538463e-05,
      "loss": 0.3349,
      "step": 1431
    },
    {
      "epoch": 18.30031948881789,
      "grad_norm": 3.2398135662078857,
      "learning_rate": 1.2656410256410257e-05,
      "loss": 0.1115,
      "step": 1432
    },
    {
      "epoch": 18.313099041533548,
      "grad_norm": 4.5026702880859375,
      "learning_rate": 1.2651282051282052e-05,
      "loss": 0.1812,
      "step": 1433
    },
    {
      "epoch": 18.3258785942492,
      "grad_norm": 4.205916404724121,
      "learning_rate": 1.2646153846153848e-05,
      "loss": 0.1706,
      "step": 1434
    },
    {
      "epoch": 18.338658146964857,
      "grad_norm": 3.850865602493286,
      "learning_rate": 1.2641025641025642e-05,
      "loss": 0.1835,
      "step": 1435
    },
    {
      "epoch": 18.35143769968051,
      "grad_norm": 3.3740909099578857,
      "learning_rate": 1.2635897435897438e-05,
      "loss": 0.1409,
      "step": 1436
    },
    {
      "epoch": 18.364217252396166,
      "grad_norm": 3.58420467376709,
      "learning_rate": 1.263076923076923e-05,
      "loss": 0.1231,
      "step": 1437
    },
    {
      "epoch": 18.376996805111823,
      "grad_norm": 4.956149101257324,
      "learning_rate": 1.2625641025641026e-05,
      "loss": 0.1889,
      "step": 1438
    },
    {
      "epoch": 18.389776357827476,
      "grad_norm": 5.362755298614502,
      "learning_rate": 1.2620512820512822e-05,
      "loss": 0.1947,
      "step": 1439
    },
    {
      "epoch": 18.402555910543132,
      "grad_norm": 4.240985870361328,
      "learning_rate": 1.2615384615384616e-05,
      "loss": 0.1353,
      "step": 1440
    },
    {
      "epoch": 18.415335463258785,
      "grad_norm": 4.410498142242432,
      "learning_rate": 1.2610256410256412e-05,
      "loss": 0.1332,
      "step": 1441
    },
    {
      "epoch": 18.42811501597444,
      "grad_norm": 4.465317726135254,
      "learning_rate": 1.2605128205128206e-05,
      "loss": 0.1308,
      "step": 1442
    },
    {
      "epoch": 18.440894568690094,
      "grad_norm": 5.566956043243408,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.1684,
      "step": 1443
    },
    {
      "epoch": 18.45367412140575,
      "grad_norm": 4.643564701080322,
      "learning_rate": 1.2594871794871797e-05,
      "loss": 0.133,
      "step": 1444
    },
    {
      "epoch": 18.466453674121407,
      "grad_norm": 3.7559735774993896,
      "learning_rate": 1.2589743589743591e-05,
      "loss": 0.1165,
      "step": 1445
    },
    {
      "epoch": 18.47923322683706,
      "grad_norm": 4.310507297515869,
      "learning_rate": 1.2584615384615387e-05,
      "loss": 0.1284,
      "step": 1446
    },
    {
      "epoch": 18.492012779552716,
      "grad_norm": 4.490650177001953,
      "learning_rate": 1.257948717948718e-05,
      "loss": 0.1571,
      "step": 1447
    },
    {
      "epoch": 18.50479233226837,
      "grad_norm": 3.584522008895874,
      "learning_rate": 1.2574358974358975e-05,
      "loss": 0.0975,
      "step": 1448
    },
    {
      "epoch": 18.517571884984026,
      "grad_norm": 3.844524621963501,
      "learning_rate": 1.256923076923077e-05,
      "loss": 0.1871,
      "step": 1449
    },
    {
      "epoch": 18.53035143769968,
      "grad_norm": 4.957360744476318,
      "learning_rate": 1.2564102564102565e-05,
      "loss": 0.1563,
      "step": 1450
    },
    {
      "epoch": 18.543130990415335,
      "grad_norm": 3.925318717956543,
      "learning_rate": 1.255897435897436e-05,
      "loss": 0.1406,
      "step": 1451
    },
    {
      "epoch": 18.55591054313099,
      "grad_norm": 4.986582279205322,
      "learning_rate": 1.2553846153846155e-05,
      "loss": 0.1655,
      "step": 1452
    },
    {
      "epoch": 18.568690095846645,
      "grad_norm": 4.465045928955078,
      "learning_rate": 1.254871794871795e-05,
      "loss": 0.1403,
      "step": 1453
    },
    {
      "epoch": 18.5814696485623,
      "grad_norm": 5.537022113800049,
      "learning_rate": 1.2543589743589744e-05,
      "loss": 0.1297,
      "step": 1454
    },
    {
      "epoch": 18.594249201277954,
      "grad_norm": 4.612170696258545,
      "learning_rate": 1.253846153846154e-05,
      "loss": 0.1416,
      "step": 1455
    },
    {
      "epoch": 18.60702875399361,
      "grad_norm": 3.784298896789551,
      "learning_rate": 1.2533333333333336e-05,
      "loss": 0.1546,
      "step": 1456
    },
    {
      "epoch": 18.619808306709267,
      "grad_norm": 5.120687961578369,
      "learning_rate": 1.2528205128205128e-05,
      "loss": 0.2026,
      "step": 1457
    },
    {
      "epoch": 18.63258785942492,
      "grad_norm": 3.836606979370117,
      "learning_rate": 1.2523076923076924e-05,
      "loss": 0.1175,
      "step": 1458
    },
    {
      "epoch": 18.645367412140576,
      "grad_norm": 3.785749912261963,
      "learning_rate": 1.2517948717948718e-05,
      "loss": 0.1788,
      "step": 1459
    },
    {
      "epoch": 18.65814696485623,
      "grad_norm": 4.040433883666992,
      "learning_rate": 1.2512820512820514e-05,
      "loss": 0.1345,
      "step": 1460
    },
    {
      "epoch": 18.670926517571885,
      "grad_norm": 4.360098838806152,
      "learning_rate": 1.250769230769231e-05,
      "loss": 0.1565,
      "step": 1461
    },
    {
      "epoch": 18.68370607028754,
      "grad_norm": 5.078848838806152,
      "learning_rate": 1.2502564102564104e-05,
      "loss": 0.1558,
      "step": 1462
    },
    {
      "epoch": 18.696485623003195,
      "grad_norm": 6.132460117340088,
      "learning_rate": 1.24974358974359e-05,
      "loss": 0.1972,
      "step": 1463
    },
    {
      "epoch": 18.70926517571885,
      "grad_norm": 3.8772716522216797,
      "learning_rate": 1.2492307692307692e-05,
      "loss": 0.1498,
      "step": 1464
    },
    {
      "epoch": 18.722044728434504,
      "grad_norm": 4.878615856170654,
      "learning_rate": 1.2487179487179487e-05,
      "loss": 0.161,
      "step": 1465
    },
    {
      "epoch": 18.73482428115016,
      "grad_norm": 4.384540557861328,
      "learning_rate": 1.2482051282051285e-05,
      "loss": 0.1664,
      "step": 1466
    },
    {
      "epoch": 18.747603833865814,
      "grad_norm": 4.386280536651611,
      "learning_rate": 1.2476923076923077e-05,
      "loss": 0.1628,
      "step": 1467
    },
    {
      "epoch": 18.76038338658147,
      "grad_norm": 4.095793724060059,
      "learning_rate": 1.2471794871794873e-05,
      "loss": 0.1324,
      "step": 1468
    },
    {
      "epoch": 18.773162939297123,
      "grad_norm": 4.629232883453369,
      "learning_rate": 1.2466666666666667e-05,
      "loss": 0.1747,
      "step": 1469
    },
    {
      "epoch": 18.78594249201278,
      "grad_norm": 4.9055256843566895,
      "learning_rate": 1.2461538461538463e-05,
      "loss": 0.1577,
      "step": 1470
    },
    {
      "epoch": 18.798722044728436,
      "grad_norm": 4.488932132720947,
      "learning_rate": 1.2456410256410259e-05,
      "loss": 0.1546,
      "step": 1471
    },
    {
      "epoch": 18.81150159744409,
      "grad_norm": 5.880629539489746,
      "learning_rate": 1.2451282051282053e-05,
      "loss": 0.1649,
      "step": 1472
    },
    {
      "epoch": 18.824281150159745,
      "grad_norm": 5.382693767547607,
      "learning_rate": 1.2446153846153848e-05,
      "loss": 0.1926,
      "step": 1473
    },
    {
      "epoch": 18.837060702875398,
      "grad_norm": 4.9048848152160645,
      "learning_rate": 1.244102564102564e-05,
      "loss": 0.15,
      "step": 1474
    },
    {
      "epoch": 18.849840255591054,
      "grad_norm": 4.941222190856934,
      "learning_rate": 1.2435897435897436e-05,
      "loss": 0.1193,
      "step": 1475
    },
    {
      "epoch": 18.86261980830671,
      "grad_norm": 3.5953900814056396,
      "learning_rate": 1.243076923076923e-05,
      "loss": 0.1311,
      "step": 1476
    },
    {
      "epoch": 18.875399361022364,
      "grad_norm": 3.8183746337890625,
      "learning_rate": 1.2425641025641026e-05,
      "loss": 0.116,
      "step": 1477
    },
    {
      "epoch": 18.88817891373802,
      "grad_norm": 4.224637985229492,
      "learning_rate": 1.2420512820512822e-05,
      "loss": 0.1153,
      "step": 1478
    },
    {
      "epoch": 18.900958466453673,
      "grad_norm": 4.898967742919922,
      "learning_rate": 1.2415384615384616e-05,
      "loss": 0.1783,
      "step": 1479
    },
    {
      "epoch": 18.91373801916933,
      "grad_norm": 5.724062442779541,
      "learning_rate": 1.2410256410256412e-05,
      "loss": 0.1652,
      "step": 1480
    },
    {
      "epoch": 18.926517571884983,
      "grad_norm": 4.231504917144775,
      "learning_rate": 1.2405128205128206e-05,
      "loss": 0.1838,
      "step": 1481
    },
    {
      "epoch": 18.93929712460064,
      "grad_norm": 4.956578731536865,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.1708,
      "step": 1482
    },
    {
      "epoch": 18.952076677316295,
      "grad_norm": 4.780496120452881,
      "learning_rate": 1.2394871794871797e-05,
      "loss": 0.1335,
      "step": 1483
    },
    {
      "epoch": 18.96485623003195,
      "grad_norm": 4.9347028732299805,
      "learning_rate": 1.238974358974359e-05,
      "loss": 0.2332,
      "step": 1484
    },
    {
      "epoch": 18.977635782747605,
      "grad_norm": 5.3977837562561035,
      "learning_rate": 1.2384615384615385e-05,
      "loss": 0.2043,
      "step": 1485
    },
    {
      "epoch": 18.990415335463258,
      "grad_norm": 5.341475963592529,
      "learning_rate": 1.237948717948718e-05,
      "loss": 0.1632,
      "step": 1486
    },
    {
      "epoch": 19.003194888178914,
      "grad_norm": 4.257419586181641,
      "learning_rate": 1.2374358974358975e-05,
      "loss": 0.1374,
      "step": 1487
    },
    {
      "epoch": 19.015974440894567,
      "grad_norm": 4.08753776550293,
      "learning_rate": 1.2369230769230771e-05,
      "loss": 0.1346,
      "step": 1488
    },
    {
      "epoch": 19.028753993610223,
      "grad_norm": 3.8640146255493164,
      "learning_rate": 1.2364102564102565e-05,
      "loss": 0.107,
      "step": 1489
    },
    {
      "epoch": 19.04153354632588,
      "grad_norm": 4.158913612365723,
      "learning_rate": 1.235897435897436e-05,
      "loss": 0.1216,
      "step": 1490
    },
    {
      "epoch": 19.054313099041533,
      "grad_norm": 3.6827924251556396,
      "learning_rate": 1.2353846153846155e-05,
      "loss": 0.1169,
      "step": 1491
    },
    {
      "epoch": 19.06709265175719,
      "grad_norm": 3.9266042709350586,
      "learning_rate": 1.234871794871795e-05,
      "loss": 0.1164,
      "step": 1492
    },
    {
      "epoch": 19.079872204472842,
      "grad_norm": 4.07518196105957,
      "learning_rate": 1.2343589743589746e-05,
      "loss": 0.0927,
      "step": 1493
    },
    {
      "epoch": 19.0926517571885,
      "grad_norm": 4.363387107849121,
      "learning_rate": 1.2338461538461539e-05,
      "loss": 0.1345,
      "step": 1494
    },
    {
      "epoch": 19.105431309904155,
      "grad_norm": 4.984545707702637,
      "learning_rate": 1.2333333333333334e-05,
      "loss": 0.1205,
      "step": 1495
    },
    {
      "epoch": 19.118210862619808,
      "grad_norm": 2.710969924926758,
      "learning_rate": 1.2328205128205128e-05,
      "loss": 0.113,
      "step": 1496
    },
    {
      "epoch": 19.130990415335464,
      "grad_norm": 3.27824068069458,
      "learning_rate": 1.2323076923076924e-05,
      "loss": 0.0981,
      "step": 1497
    },
    {
      "epoch": 19.143769968051117,
      "grad_norm": 3.5957398414611816,
      "learning_rate": 1.231794871794872e-05,
      "loss": 0.1738,
      "step": 1498
    },
    {
      "epoch": 19.156549520766774,
      "grad_norm": 2.8213162422180176,
      "learning_rate": 1.2312820512820514e-05,
      "loss": 0.074,
      "step": 1499
    },
    {
      "epoch": 19.169329073482427,
      "grad_norm": 4.4678425788879395,
      "learning_rate": 1.230769230769231e-05,
      "loss": 0.1392,
      "step": 1500
    },
    {
      "epoch": 19.169329073482427,
      "eval_loss": 0.6049781441688538,
      "eval_runtime": 183.8411,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 1500
    },
    {
      "epoch": 19.182108626198083,
      "grad_norm": 4.500428676605225,
      "learning_rate": 1.2302564102564102e-05,
      "loss": 0.172,
      "step": 1501
    },
    {
      "epoch": 19.19488817891374,
      "grad_norm": 3.4990546703338623,
      "learning_rate": 1.22974358974359e-05,
      "loss": 0.1249,
      "step": 1502
    },
    {
      "epoch": 19.207667731629392,
      "grad_norm": 3.9662656784057617,
      "learning_rate": 1.2292307692307692e-05,
      "loss": 0.1258,
      "step": 1503
    },
    {
      "epoch": 19.22044728434505,
      "grad_norm": 5.195966720581055,
      "learning_rate": 1.2287179487179488e-05,
      "loss": 0.1837,
      "step": 1504
    },
    {
      "epoch": 19.233226837060702,
      "grad_norm": 3.6646034717559814,
      "learning_rate": 1.2282051282051283e-05,
      "loss": 0.0861,
      "step": 1505
    },
    {
      "epoch": 19.24600638977636,
      "grad_norm": 3.9439032077789307,
      "learning_rate": 1.2276923076923077e-05,
      "loss": 0.1212,
      "step": 1506
    },
    {
      "epoch": 19.25878594249201,
      "grad_norm": 4.427985191345215,
      "learning_rate": 1.2271794871794873e-05,
      "loss": 0.1297,
      "step": 1507
    },
    {
      "epoch": 19.271565495207668,
      "grad_norm": 3.02043080329895,
      "learning_rate": 1.2266666666666667e-05,
      "loss": 0.07,
      "step": 1508
    },
    {
      "epoch": 19.284345047923324,
      "grad_norm": 3.5911426544189453,
      "learning_rate": 1.2261538461538463e-05,
      "loss": 0.0998,
      "step": 1509
    },
    {
      "epoch": 19.297124600638977,
      "grad_norm": 3.7852537631988525,
      "learning_rate": 1.2256410256410259e-05,
      "loss": 0.0708,
      "step": 1510
    },
    {
      "epoch": 19.309904153354633,
      "grad_norm": 4.233508110046387,
      "learning_rate": 1.2251282051282051e-05,
      "loss": 0.1262,
      "step": 1511
    },
    {
      "epoch": 19.322683706070286,
      "grad_norm": 4.237662315368652,
      "learning_rate": 1.2246153846153847e-05,
      "loss": 0.1159,
      "step": 1512
    },
    {
      "epoch": 19.335463258785943,
      "grad_norm": 6.377729415893555,
      "learning_rate": 1.2241025641025641e-05,
      "loss": 0.2467,
      "step": 1513
    },
    {
      "epoch": 19.3482428115016,
      "grad_norm": 4.620576858520508,
      "learning_rate": 1.2235897435897437e-05,
      "loss": 0.1471,
      "step": 1514
    },
    {
      "epoch": 19.361022364217252,
      "grad_norm": 4.3901166915893555,
      "learning_rate": 1.2230769230769232e-05,
      "loss": 0.1519,
      "step": 1515
    },
    {
      "epoch": 19.37380191693291,
      "grad_norm": 6.668283462524414,
      "learning_rate": 1.2225641025641026e-05,
      "loss": 0.1385,
      "step": 1516
    },
    {
      "epoch": 19.38658146964856,
      "grad_norm": 3.3707025051116943,
      "learning_rate": 1.2220512820512822e-05,
      "loss": 0.1423,
      "step": 1517
    },
    {
      "epoch": 19.399361022364218,
      "grad_norm": 3.8963208198547363,
      "learning_rate": 1.2215384615384616e-05,
      "loss": 0.1763,
      "step": 1518
    },
    {
      "epoch": 19.41214057507987,
      "grad_norm": 3.0294902324676514,
      "learning_rate": 1.2210256410256412e-05,
      "loss": 0.0771,
      "step": 1519
    },
    {
      "epoch": 19.424920127795527,
      "grad_norm": 3.7675819396972656,
      "learning_rate": 1.2205128205128208e-05,
      "loss": 0.1174,
      "step": 1520
    },
    {
      "epoch": 19.437699680511184,
      "grad_norm": 3.797121047973633,
      "learning_rate": 1.22e-05,
      "loss": 0.1358,
      "step": 1521
    },
    {
      "epoch": 19.450479233226837,
      "grad_norm": 4.065491676330566,
      "learning_rate": 1.2194871794871796e-05,
      "loss": 0.108,
      "step": 1522
    },
    {
      "epoch": 19.463258785942493,
      "grad_norm": 3.6715574264526367,
      "learning_rate": 1.218974358974359e-05,
      "loss": 0.139,
      "step": 1523
    },
    {
      "epoch": 19.476038338658146,
      "grad_norm": 5.250418663024902,
      "learning_rate": 1.2184615384615386e-05,
      "loss": 0.1425,
      "step": 1524
    },
    {
      "epoch": 19.488817891373802,
      "grad_norm": 4.669528484344482,
      "learning_rate": 1.217948717948718e-05,
      "loss": 0.1321,
      "step": 1525
    },
    {
      "epoch": 19.501597444089455,
      "grad_norm": 4.585147380828857,
      "learning_rate": 1.2174358974358975e-05,
      "loss": 0.1232,
      "step": 1526
    },
    {
      "epoch": 19.51437699680511,
      "grad_norm": 3.8052077293395996,
      "learning_rate": 1.2169230769230771e-05,
      "loss": 0.1059,
      "step": 1527
    },
    {
      "epoch": 19.527156549520768,
      "grad_norm": 3.7040152549743652,
      "learning_rate": 1.2164102564102565e-05,
      "loss": 0.1017,
      "step": 1528
    },
    {
      "epoch": 19.53993610223642,
      "grad_norm": 3.7301838397979736,
      "learning_rate": 1.2158974358974361e-05,
      "loss": 0.0855,
      "step": 1529
    },
    {
      "epoch": 19.552715654952078,
      "grad_norm": 4.927285194396973,
      "learning_rate": 1.2153846153846153e-05,
      "loss": 0.1538,
      "step": 1530
    },
    {
      "epoch": 19.56549520766773,
      "grad_norm": 4.502018928527832,
      "learning_rate": 1.2148717948717949e-05,
      "loss": 0.097,
      "step": 1531
    },
    {
      "epoch": 19.578274760383387,
      "grad_norm": 4.707366943359375,
      "learning_rate": 1.2143589743589745e-05,
      "loss": 0.1858,
      "step": 1532
    },
    {
      "epoch": 19.591054313099043,
      "grad_norm": 5.845135688781738,
      "learning_rate": 1.2138461538461539e-05,
      "loss": 0.1434,
      "step": 1533
    },
    {
      "epoch": 19.603833865814696,
      "grad_norm": 4.260502338409424,
      "learning_rate": 1.2133333333333335e-05,
      "loss": 0.1188,
      "step": 1534
    },
    {
      "epoch": 19.616613418530353,
      "grad_norm": 4.49433708190918,
      "learning_rate": 1.2128205128205129e-05,
      "loss": 0.1205,
      "step": 1535
    },
    {
      "epoch": 19.629392971246006,
      "grad_norm": 4.624430179595947,
      "learning_rate": 1.2123076923076924e-05,
      "loss": 0.1194,
      "step": 1536
    },
    {
      "epoch": 19.642172523961662,
      "grad_norm": 4.401130676269531,
      "learning_rate": 1.211794871794872e-05,
      "loss": 0.1617,
      "step": 1537
    },
    {
      "epoch": 19.654952076677315,
      "grad_norm": 3.5289738178253174,
      "learning_rate": 1.2112820512820514e-05,
      "loss": 0.0927,
      "step": 1538
    },
    {
      "epoch": 19.66773162939297,
      "grad_norm": 6.962007522583008,
      "learning_rate": 1.210769230769231e-05,
      "loss": 0.1999,
      "step": 1539
    },
    {
      "epoch": 19.680511182108628,
      "grad_norm": 5.013849258422852,
      "learning_rate": 1.2102564102564102e-05,
      "loss": 0.1448,
      "step": 1540
    },
    {
      "epoch": 19.69329073482428,
      "grad_norm": 4.766885757446289,
      "learning_rate": 1.2097435897435898e-05,
      "loss": 0.1597,
      "step": 1541
    },
    {
      "epoch": 19.706070287539937,
      "grad_norm": 3.9394874572753906,
      "learning_rate": 1.2092307692307694e-05,
      "loss": 0.1112,
      "step": 1542
    },
    {
      "epoch": 19.71884984025559,
      "grad_norm": 5.361410140991211,
      "learning_rate": 1.2087179487179488e-05,
      "loss": 0.1157,
      "step": 1543
    },
    {
      "epoch": 19.731629392971247,
      "grad_norm": 3.999269962310791,
      "learning_rate": 1.2082051282051284e-05,
      "loss": 0.1391,
      "step": 1544
    },
    {
      "epoch": 19.7444089456869,
      "grad_norm": 5.273488998413086,
      "learning_rate": 1.2076923076923078e-05,
      "loss": 0.1272,
      "step": 1545
    },
    {
      "epoch": 19.757188498402556,
      "grad_norm": 4.8968329429626465,
      "learning_rate": 1.2071794871794873e-05,
      "loss": 0.1984,
      "step": 1546
    },
    {
      "epoch": 19.769968051118212,
      "grad_norm": 4.2163872718811035,
      "learning_rate": 1.206666666666667e-05,
      "loss": 0.1166,
      "step": 1547
    },
    {
      "epoch": 19.782747603833865,
      "grad_norm": 4.540549278259277,
      "learning_rate": 1.2061538461538462e-05,
      "loss": 0.1172,
      "step": 1548
    },
    {
      "epoch": 19.79552715654952,
      "grad_norm": 4.659392833709717,
      "learning_rate": 1.2056410256410257e-05,
      "loss": 0.1947,
      "step": 1549
    },
    {
      "epoch": 19.808306709265175,
      "grad_norm": 3.6807572841644287,
      "learning_rate": 1.2051282051282051e-05,
      "loss": 0.1054,
      "step": 1550
    },
    {
      "epoch": 19.82108626198083,
      "grad_norm": 4.467441558837891,
      "learning_rate": 1.2046153846153847e-05,
      "loss": 0.1532,
      "step": 1551
    },
    {
      "epoch": 19.833865814696484,
      "grad_norm": 3.2000980377197266,
      "learning_rate": 1.2041025641025641e-05,
      "loss": 0.0957,
      "step": 1552
    },
    {
      "epoch": 19.84664536741214,
      "grad_norm": 5.999240398406982,
      "learning_rate": 1.2035897435897437e-05,
      "loss": 0.1822,
      "step": 1553
    },
    {
      "epoch": 19.859424920127797,
      "grad_norm": 4.817787170410156,
      "learning_rate": 1.2030769230769233e-05,
      "loss": 0.1507,
      "step": 1554
    },
    {
      "epoch": 19.87220447284345,
      "grad_norm": 5.96992301940918,
      "learning_rate": 1.2025641025641027e-05,
      "loss": 0.1525,
      "step": 1555
    },
    {
      "epoch": 19.884984025559106,
      "grad_norm": 4.115090370178223,
      "learning_rate": 1.2020512820512822e-05,
      "loss": 0.1262,
      "step": 1556
    },
    {
      "epoch": 19.89776357827476,
      "grad_norm": 4.930627346038818,
      "learning_rate": 1.2015384615384615e-05,
      "loss": 0.2024,
      "step": 1557
    },
    {
      "epoch": 19.910543130990416,
      "grad_norm": 4.551126003265381,
      "learning_rate": 1.201025641025641e-05,
      "loss": 0.0929,
      "step": 1558
    },
    {
      "epoch": 19.923322683706072,
      "grad_norm": 4.8088836669921875,
      "learning_rate": 1.2005128205128206e-05,
      "loss": 0.1261,
      "step": 1559
    },
    {
      "epoch": 19.936102236421725,
      "grad_norm": 4.031205177307129,
      "learning_rate": 1.2e-05,
      "loss": 0.1601,
      "step": 1560
    },
    {
      "epoch": 19.94888178913738,
      "grad_norm": 4.078291893005371,
      "learning_rate": 1.1994871794871796e-05,
      "loss": 0.1488,
      "step": 1561
    },
    {
      "epoch": 19.961661341853034,
      "grad_norm": 3.802271842956543,
      "learning_rate": 1.198974358974359e-05,
      "loss": 0.1198,
      "step": 1562
    },
    {
      "epoch": 19.97444089456869,
      "grad_norm": 4.118734359741211,
      "learning_rate": 1.1984615384615386e-05,
      "loss": 0.1854,
      "step": 1563
    },
    {
      "epoch": 19.987220447284344,
      "grad_norm": 3.7523162364959717,
      "learning_rate": 1.1979487179487182e-05,
      "loss": 0.1037,
      "step": 1564
    },
    {
      "epoch": 20.0,
      "grad_norm": 4.222137451171875,
      "learning_rate": 1.1974358974358976e-05,
      "loss": 0.1086,
      "step": 1565
    },
    {
      "epoch": 20.012779552715656,
      "grad_norm": 4.975546836853027,
      "learning_rate": 1.1969230769230771e-05,
      "loss": 0.131,
      "step": 1566
    },
    {
      "epoch": 20.02555910543131,
      "grad_norm": 3.5944299697875977,
      "learning_rate": 1.1964102564102564e-05,
      "loss": 0.089,
      "step": 1567
    },
    {
      "epoch": 20.038338658146966,
      "grad_norm": 4.291761875152588,
      "learning_rate": 1.195897435897436e-05,
      "loss": 0.0949,
      "step": 1568
    },
    {
      "epoch": 20.05111821086262,
      "grad_norm": 4.3121337890625,
      "learning_rate": 1.1953846153846155e-05,
      "loss": 0.1194,
      "step": 1569
    },
    {
      "epoch": 20.063897763578275,
      "grad_norm": 5.17624044418335,
      "learning_rate": 1.194871794871795e-05,
      "loss": 0.1156,
      "step": 1570
    },
    {
      "epoch": 20.076677316293928,
      "grad_norm": 4.396629810333252,
      "learning_rate": 1.1943589743589745e-05,
      "loss": 0.1126,
      "step": 1571
    },
    {
      "epoch": 20.089456869009584,
      "grad_norm": 3.466047525405884,
      "learning_rate": 1.1938461538461539e-05,
      "loss": 0.1243,
      "step": 1572
    },
    {
      "epoch": 20.10223642172524,
      "grad_norm": 3.3175036907196045,
      "learning_rate": 1.1933333333333335e-05,
      "loss": 0.1041,
      "step": 1573
    },
    {
      "epoch": 20.115015974440894,
      "grad_norm": 3.193227767944336,
      "learning_rate": 1.1928205128205129e-05,
      "loss": 0.0859,
      "step": 1574
    },
    {
      "epoch": 20.12779552715655,
      "grad_norm": 3.787001132965088,
      "learning_rate": 1.1923076923076925e-05,
      "loss": 0.0922,
      "step": 1575
    },
    {
      "epoch": 20.140575079872203,
      "grad_norm": 4.286689758300781,
      "learning_rate": 1.191794871794872e-05,
      "loss": 0.0923,
      "step": 1576
    },
    {
      "epoch": 20.15335463258786,
      "grad_norm": 4.684820175170898,
      "learning_rate": 1.1912820512820513e-05,
      "loss": 0.1506,
      "step": 1577
    },
    {
      "epoch": 20.166134185303516,
      "grad_norm": 4.568631172180176,
      "learning_rate": 1.1907692307692308e-05,
      "loss": 0.0986,
      "step": 1578
    },
    {
      "epoch": 20.17891373801917,
      "grad_norm": 4.025661945343018,
      "learning_rate": 1.1902564102564103e-05,
      "loss": 0.0853,
      "step": 1579
    },
    {
      "epoch": 20.191693290734825,
      "grad_norm": 6.226049900054932,
      "learning_rate": 1.1897435897435898e-05,
      "loss": 0.1414,
      "step": 1580
    },
    {
      "epoch": 20.20447284345048,
      "grad_norm": 4.703970909118652,
      "learning_rate": 1.1892307692307694e-05,
      "loss": 0.1196,
      "step": 1581
    },
    {
      "epoch": 20.217252396166135,
      "grad_norm": 3.3595998287200928,
      "learning_rate": 1.1887179487179488e-05,
      "loss": 0.085,
      "step": 1582
    },
    {
      "epoch": 20.230031948881788,
      "grad_norm": 6.735140323638916,
      "learning_rate": 1.1882051282051284e-05,
      "loss": 0.1397,
      "step": 1583
    },
    {
      "epoch": 20.242811501597444,
      "grad_norm": 4.5568342208862305,
      "learning_rate": 1.1876923076923076e-05,
      "loss": 0.0925,
      "step": 1584
    },
    {
      "epoch": 20.2555910543131,
      "grad_norm": 4.656588554382324,
      "learning_rate": 1.1871794871794872e-05,
      "loss": 0.1483,
      "step": 1585
    },
    {
      "epoch": 20.268370607028753,
      "grad_norm": 3.1620121002197266,
      "learning_rate": 1.186666666666667e-05,
      "loss": 0.0802,
      "step": 1586
    },
    {
      "epoch": 20.28115015974441,
      "grad_norm": 3.6990063190460205,
      "learning_rate": 1.1861538461538462e-05,
      "loss": 0.1172,
      "step": 1587
    },
    {
      "epoch": 20.293929712460063,
      "grad_norm": 3.8228986263275146,
      "learning_rate": 1.1856410256410257e-05,
      "loss": 0.1211,
      "step": 1588
    },
    {
      "epoch": 20.30670926517572,
      "grad_norm": 4.515007972717285,
      "learning_rate": 1.1851282051282052e-05,
      "loss": 0.1159,
      "step": 1589
    },
    {
      "epoch": 20.319488817891372,
      "grad_norm": 4.058969974517822,
      "learning_rate": 1.1846153846153847e-05,
      "loss": 0.0675,
      "step": 1590
    },
    {
      "epoch": 20.33226837060703,
      "grad_norm": 5.006679058074951,
      "learning_rate": 1.1841025641025643e-05,
      "loss": 0.1014,
      "step": 1591
    },
    {
      "epoch": 20.345047923322685,
      "grad_norm": 4.770201206207275,
      "learning_rate": 1.1835897435897437e-05,
      "loss": 0.1304,
      "step": 1592
    },
    {
      "epoch": 20.357827476038338,
      "grad_norm": 4.045782566070557,
      "learning_rate": 1.1830769230769233e-05,
      "loss": 0.1109,
      "step": 1593
    },
    {
      "epoch": 20.370607028753994,
      "grad_norm": 4.972194194793701,
      "learning_rate": 1.1825641025641025e-05,
      "loss": 0.1055,
      "step": 1594
    },
    {
      "epoch": 20.383386581469647,
      "grad_norm": 3.656635284423828,
      "learning_rate": 1.1820512820512821e-05,
      "loss": 0.0696,
      "step": 1595
    },
    {
      "epoch": 20.396166134185304,
      "grad_norm": 4.930873394012451,
      "learning_rate": 1.1815384615384617e-05,
      "loss": 0.1229,
      "step": 1596
    },
    {
      "epoch": 20.408945686900957,
      "grad_norm": 3.699876070022583,
      "learning_rate": 1.181025641025641e-05,
      "loss": 0.0886,
      "step": 1597
    },
    {
      "epoch": 20.421725239616613,
      "grad_norm": 5.458634853363037,
      "learning_rate": 1.1805128205128206e-05,
      "loss": 0.0778,
      "step": 1598
    },
    {
      "epoch": 20.43450479233227,
      "grad_norm": 4.316697597503662,
      "learning_rate": 1.18e-05,
      "loss": 0.098,
      "step": 1599
    },
    {
      "epoch": 20.447284345047922,
      "grad_norm": 3.8660194873809814,
      "learning_rate": 1.1794871794871796e-05,
      "loss": 0.0971,
      "step": 1600
    },
    {
      "epoch": 20.447284345047922,
      "eval_loss": 0.6326173543930054,
      "eval_runtime": 183.9465,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 1600
    },
    {
      "epoch": 20.46006389776358,
      "grad_norm": 5.642141342163086,
      "learning_rate": 1.178974358974359e-05,
      "loss": 0.143,
      "step": 1601
    },
    {
      "epoch": 20.472843450479232,
      "grad_norm": 6.414536952972412,
      "learning_rate": 1.1784615384615386e-05,
      "loss": 0.1125,
      "step": 1602
    },
    {
      "epoch": 20.48562300319489,
      "grad_norm": 4.429561138153076,
      "learning_rate": 1.1779487179487182e-05,
      "loss": 0.1181,
      "step": 1603
    },
    {
      "epoch": 20.498402555910545,
      "grad_norm": 3.8313069343566895,
      "learning_rate": 1.1774358974358974e-05,
      "loss": 0.1016,
      "step": 1604
    },
    {
      "epoch": 20.511182108626198,
      "grad_norm": 3.912367105484009,
      "learning_rate": 1.176923076923077e-05,
      "loss": 0.1038,
      "step": 1605
    },
    {
      "epoch": 20.523961661341854,
      "grad_norm": 5.56271505355835,
      "learning_rate": 1.1764102564102564e-05,
      "loss": 0.1639,
      "step": 1606
    },
    {
      "epoch": 20.536741214057507,
      "grad_norm": 3.5914433002471924,
      "learning_rate": 1.175897435897436e-05,
      "loss": 0.0811,
      "step": 1607
    },
    {
      "epoch": 20.549520766773163,
      "grad_norm": 5.16848087310791,
      "learning_rate": 1.1753846153846155e-05,
      "loss": 0.1301,
      "step": 1608
    },
    {
      "epoch": 20.562300319488816,
      "grad_norm": 3.407752513885498,
      "learning_rate": 1.174871794871795e-05,
      "loss": 0.0736,
      "step": 1609
    },
    {
      "epoch": 20.575079872204473,
      "grad_norm": 3.251795530319214,
      "learning_rate": 1.1743589743589745e-05,
      "loss": 0.0633,
      "step": 1610
    },
    {
      "epoch": 20.58785942492013,
      "grad_norm": 5.2076029777526855,
      "learning_rate": 1.173846153846154e-05,
      "loss": 0.1263,
      "step": 1611
    },
    {
      "epoch": 20.600638977635782,
      "grad_norm": 5.330387115478516,
      "learning_rate": 1.1733333333333335e-05,
      "loss": 0.1103,
      "step": 1612
    },
    {
      "epoch": 20.61341853035144,
      "grad_norm": 4.368866920471191,
      "learning_rate": 1.172820512820513e-05,
      "loss": 0.1167,
      "step": 1613
    },
    {
      "epoch": 20.62619808306709,
      "grad_norm": 3.9925103187561035,
      "learning_rate": 1.1723076923076923e-05,
      "loss": 0.0782,
      "step": 1614
    },
    {
      "epoch": 20.638977635782748,
      "grad_norm": 3.283320426940918,
      "learning_rate": 1.1717948717948719e-05,
      "loss": 0.0994,
      "step": 1615
    },
    {
      "epoch": 20.6517571884984,
      "grad_norm": 4.990774154663086,
      "learning_rate": 1.1712820512820513e-05,
      "loss": 0.1341,
      "step": 1616
    },
    {
      "epoch": 20.664536741214057,
      "grad_norm": 3.5431063175201416,
      "learning_rate": 1.1707692307692309e-05,
      "loss": 0.0923,
      "step": 1617
    },
    {
      "epoch": 20.677316293929714,
      "grad_norm": 4.684637546539307,
      "learning_rate": 1.1702564102564104e-05,
      "loss": 0.084,
      "step": 1618
    },
    {
      "epoch": 20.690095846645367,
      "grad_norm": 4.183215618133545,
      "learning_rate": 1.1697435897435899e-05,
      "loss": 0.101,
      "step": 1619
    },
    {
      "epoch": 20.702875399361023,
      "grad_norm": 3.7057716846466064,
      "learning_rate": 1.1692307692307694e-05,
      "loss": 0.1015,
      "step": 1620
    },
    {
      "epoch": 20.715654952076676,
      "grad_norm": 5.4087324142456055,
      "learning_rate": 1.1687179487179488e-05,
      "loss": 0.1613,
      "step": 1621
    },
    {
      "epoch": 20.728434504792332,
      "grad_norm": 4.276289463043213,
      "learning_rate": 1.1682051282051284e-05,
      "loss": 0.1231,
      "step": 1622
    },
    {
      "epoch": 20.74121405750799,
      "grad_norm": 6.077336311340332,
      "learning_rate": 1.1676923076923076e-05,
      "loss": 0.1098,
      "step": 1623
    },
    {
      "epoch": 20.75399361022364,
      "grad_norm": 5.829557418823242,
      "learning_rate": 1.1671794871794872e-05,
      "loss": 0.193,
      "step": 1624
    },
    {
      "epoch": 20.766773162939298,
      "grad_norm": 4.161832809448242,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.0767,
      "step": 1625
    },
    {
      "epoch": 20.77955271565495,
      "grad_norm": 5.5048441886901855,
      "learning_rate": 1.1661538461538462e-05,
      "loss": 0.1275,
      "step": 1626
    },
    {
      "epoch": 20.792332268370608,
      "grad_norm": 4.068666458129883,
      "learning_rate": 1.1656410256410258e-05,
      "loss": 0.1387,
      "step": 1627
    },
    {
      "epoch": 20.80511182108626,
      "grad_norm": 4.471652507781982,
      "learning_rate": 1.1651282051282052e-05,
      "loss": 0.1038,
      "step": 1628
    },
    {
      "epoch": 20.817891373801917,
      "grad_norm": 4.135310173034668,
      "learning_rate": 1.1646153846153848e-05,
      "loss": 0.0855,
      "step": 1629
    },
    {
      "epoch": 20.830670926517573,
      "grad_norm": 4.095508098602295,
      "learning_rate": 1.1641025641025643e-05,
      "loss": 0.117,
      "step": 1630
    },
    {
      "epoch": 20.843450479233226,
      "grad_norm": 3.997915029525757,
      "learning_rate": 1.1635897435897436e-05,
      "loss": 0.1411,
      "step": 1631
    },
    {
      "epoch": 20.856230031948883,
      "grad_norm": 4.30368709564209,
      "learning_rate": 1.1630769230769231e-05,
      "loss": 0.1085,
      "step": 1632
    },
    {
      "epoch": 20.869009584664536,
      "grad_norm": 4.3267107009887695,
      "learning_rate": 1.1625641025641025e-05,
      "loss": 0.1119,
      "step": 1633
    },
    {
      "epoch": 20.881789137380192,
      "grad_norm": 4.337647438049316,
      "learning_rate": 1.1620512820512821e-05,
      "loss": 0.1402,
      "step": 1634
    },
    {
      "epoch": 20.894568690095845,
      "grad_norm": 4.954765319824219,
      "learning_rate": 1.1615384615384617e-05,
      "loss": 0.1718,
      "step": 1635
    },
    {
      "epoch": 20.9073482428115,
      "grad_norm": 3.596308946609497,
      "learning_rate": 1.1610256410256411e-05,
      "loss": 0.0814,
      "step": 1636
    },
    {
      "epoch": 20.920127795527158,
      "grad_norm": 3.3369619846343994,
      "learning_rate": 1.1605128205128207e-05,
      "loss": 0.0947,
      "step": 1637
    },
    {
      "epoch": 20.93290734824281,
      "grad_norm": 4.8418755531311035,
      "learning_rate": 1.16e-05,
      "loss": 0.1547,
      "step": 1638
    },
    {
      "epoch": 20.945686900958467,
      "grad_norm": 6.589743614196777,
      "learning_rate": 1.1594871794871796e-05,
      "loss": 0.1832,
      "step": 1639
    },
    {
      "epoch": 20.95846645367412,
      "grad_norm": 3.8558123111724854,
      "learning_rate": 1.1589743589743592e-05,
      "loss": 0.0987,
      "step": 1640
    },
    {
      "epoch": 20.971246006389777,
      "grad_norm": 4.8623127937316895,
      "learning_rate": 1.1584615384615385e-05,
      "loss": 0.1301,
      "step": 1641
    },
    {
      "epoch": 20.984025559105433,
      "grad_norm": 4.397773265838623,
      "learning_rate": 1.157948717948718e-05,
      "loss": 0.0928,
      "step": 1642
    },
    {
      "epoch": 20.996805111821086,
      "grad_norm": 4.168147563934326,
      "learning_rate": 1.1574358974358974e-05,
      "loss": 0.1086,
      "step": 1643
    },
    {
      "epoch": 21.009584664536742,
      "grad_norm": 4.462241172790527,
      "learning_rate": 1.156923076923077e-05,
      "loss": 0.0797,
      "step": 1644
    },
    {
      "epoch": 21.022364217252395,
      "grad_norm": 2.952622413635254,
      "learning_rate": 1.1564102564102566e-05,
      "loss": 0.0535,
      "step": 1645
    },
    {
      "epoch": 21.03514376996805,
      "grad_norm": 2.7388765811920166,
      "learning_rate": 1.155897435897436e-05,
      "loss": 0.0567,
      "step": 1646
    },
    {
      "epoch": 21.047923322683705,
      "grad_norm": 4.662555694580078,
      "learning_rate": 1.1553846153846156e-05,
      "loss": 0.0913,
      "step": 1647
    },
    {
      "epoch": 21.06070287539936,
      "grad_norm": 4.469814777374268,
      "learning_rate": 1.154871794871795e-05,
      "loss": 0.0873,
      "step": 1648
    },
    {
      "epoch": 21.073482428115017,
      "grad_norm": 4.687971591949463,
      "learning_rate": 1.1543589743589745e-05,
      "loss": 0.1021,
      "step": 1649
    },
    {
      "epoch": 21.08626198083067,
      "grad_norm": 3.111229181289673,
      "learning_rate": 1.1538461538461538e-05,
      "loss": 0.0719,
      "step": 1650
    },
    {
      "epoch": 21.099041533546327,
      "grad_norm": 4.174816131591797,
      "learning_rate": 1.1533333333333334e-05,
      "loss": 0.0751,
      "step": 1651
    },
    {
      "epoch": 21.11182108626198,
      "grad_norm": 2.788222312927246,
      "learning_rate": 1.152820512820513e-05,
      "loss": 0.0743,
      "step": 1652
    },
    {
      "epoch": 21.124600638977636,
      "grad_norm": 3.829700231552124,
      "learning_rate": 1.1523076923076923e-05,
      "loss": 0.0927,
      "step": 1653
    },
    {
      "epoch": 21.13738019169329,
      "grad_norm": 3.773012399673462,
      "learning_rate": 1.1517948717948719e-05,
      "loss": 0.0859,
      "step": 1654
    },
    {
      "epoch": 21.150159744408946,
      "grad_norm": 3.516838788986206,
      "learning_rate": 1.1512820512820513e-05,
      "loss": 0.0993,
      "step": 1655
    },
    {
      "epoch": 21.162939297124602,
      "grad_norm": 5.086302280426025,
      "learning_rate": 1.1507692307692309e-05,
      "loss": 0.0787,
      "step": 1656
    },
    {
      "epoch": 21.175718849840255,
      "grad_norm": 3.660397529602051,
      "learning_rate": 1.1502564102564105e-05,
      "loss": 0.0716,
      "step": 1657
    },
    {
      "epoch": 21.18849840255591,
      "grad_norm": 3.944607734680176,
      "learning_rate": 1.1497435897435899e-05,
      "loss": 0.1091,
      "step": 1658
    },
    {
      "epoch": 21.201277955271564,
      "grad_norm": 4.121984004974365,
      "learning_rate": 1.1492307692307694e-05,
      "loss": 0.0729,
      "step": 1659
    },
    {
      "epoch": 21.21405750798722,
      "grad_norm": 3.834733009338379,
      "learning_rate": 1.1487179487179487e-05,
      "loss": 0.0966,
      "step": 1660
    },
    {
      "epoch": 21.226837060702877,
      "grad_norm": 4.8407883644104,
      "learning_rate": 1.1482051282051283e-05,
      "loss": 0.1234,
      "step": 1661
    },
    {
      "epoch": 21.23961661341853,
      "grad_norm": 3.637965679168701,
      "learning_rate": 1.1476923076923078e-05,
      "loss": 0.1126,
      "step": 1662
    },
    {
      "epoch": 21.252396166134186,
      "grad_norm": 3.130990743637085,
      "learning_rate": 1.1471794871794872e-05,
      "loss": 0.0941,
      "step": 1663
    },
    {
      "epoch": 21.26517571884984,
      "grad_norm": 5.102894306182861,
      "learning_rate": 1.1466666666666668e-05,
      "loss": 0.1326,
      "step": 1664
    },
    {
      "epoch": 21.277955271565496,
      "grad_norm": 3.611659049987793,
      "learning_rate": 1.1461538461538462e-05,
      "loss": 0.0885,
      "step": 1665
    },
    {
      "epoch": 21.29073482428115,
      "grad_norm": 3.9154715538024902,
      "learning_rate": 1.1456410256410258e-05,
      "loss": 0.0991,
      "step": 1666
    },
    {
      "epoch": 21.303514376996805,
      "grad_norm": 4.346821308135986,
      "learning_rate": 1.1451282051282054e-05,
      "loss": 0.1114,
      "step": 1667
    },
    {
      "epoch": 21.31629392971246,
      "grad_norm": 3.284734010696411,
      "learning_rate": 1.1446153846153846e-05,
      "loss": 0.0624,
      "step": 1668
    },
    {
      "epoch": 21.329073482428115,
      "grad_norm": 3.343594789505005,
      "learning_rate": 1.1441025641025642e-05,
      "loss": 0.1077,
      "step": 1669
    },
    {
      "epoch": 21.34185303514377,
      "grad_norm": 3.939784049987793,
      "learning_rate": 1.1435897435897436e-05,
      "loss": 0.1046,
      "step": 1670
    },
    {
      "epoch": 21.354632587859424,
      "grad_norm": 5.225682258605957,
      "learning_rate": 1.1430769230769232e-05,
      "loss": 0.1022,
      "step": 1671
    },
    {
      "epoch": 21.36741214057508,
      "grad_norm": 3.0708720684051514,
      "learning_rate": 1.1425641025641026e-05,
      "loss": 0.065,
      "step": 1672
    },
    {
      "epoch": 21.380191693290733,
      "grad_norm": 3.394704580307007,
      "learning_rate": 1.1420512820512821e-05,
      "loss": 0.0904,
      "step": 1673
    },
    {
      "epoch": 21.39297124600639,
      "grad_norm": 3.961637496948242,
      "learning_rate": 1.1415384615384617e-05,
      "loss": 0.1179,
      "step": 1674
    },
    {
      "epoch": 21.405750798722046,
      "grad_norm": 4.387579917907715,
      "learning_rate": 1.1410256410256411e-05,
      "loss": 0.0879,
      "step": 1675
    },
    {
      "epoch": 21.4185303514377,
      "grad_norm": 2.8807778358459473,
      "learning_rate": 1.1405128205128207e-05,
      "loss": 0.0582,
      "step": 1676
    },
    {
      "epoch": 21.431309904153355,
      "grad_norm": 3.1875922679901123,
      "learning_rate": 1.14e-05,
      "loss": 0.0685,
      "step": 1677
    },
    {
      "epoch": 21.44408945686901,
      "grad_norm": 3.213134527206421,
      "learning_rate": 1.1394871794871795e-05,
      "loss": 0.0933,
      "step": 1678
    },
    {
      "epoch": 21.456869009584665,
      "grad_norm": 4.480829238891602,
      "learning_rate": 1.138974358974359e-05,
      "loss": 0.0779,
      "step": 1679
    },
    {
      "epoch": 21.46964856230032,
      "grad_norm": 4.427398204803467,
      "learning_rate": 1.1384615384615385e-05,
      "loss": 0.1158,
      "step": 1680
    },
    {
      "epoch": 21.482428115015974,
      "grad_norm": 3.7674429416656494,
      "learning_rate": 1.137948717948718e-05,
      "loss": 0.0694,
      "step": 1681
    },
    {
      "epoch": 21.49520766773163,
      "grad_norm": 3.895766258239746,
      "learning_rate": 1.1374358974358975e-05,
      "loss": 0.0984,
      "step": 1682
    },
    {
      "epoch": 21.507987220447284,
      "grad_norm": 5.089962959289551,
      "learning_rate": 1.136923076923077e-05,
      "loss": 0.126,
      "step": 1683
    },
    {
      "epoch": 21.52076677316294,
      "grad_norm": 3.1773629188537598,
      "learning_rate": 1.1364102564102566e-05,
      "loss": 0.1193,
      "step": 1684
    },
    {
      "epoch": 21.533546325878593,
      "grad_norm": 4.2216949462890625,
      "learning_rate": 1.135897435897436e-05,
      "loss": 0.0929,
      "step": 1685
    },
    {
      "epoch": 21.54632587859425,
      "grad_norm": 3.353468179702759,
      "learning_rate": 1.1353846153846156e-05,
      "loss": 0.0954,
      "step": 1686
    },
    {
      "epoch": 21.559105431309906,
      "grad_norm": 4.303814888000488,
      "learning_rate": 1.1348717948717948e-05,
      "loss": 0.0674,
      "step": 1687
    },
    {
      "epoch": 21.57188498402556,
      "grad_norm": 4.027798175811768,
      "learning_rate": 1.1343589743589744e-05,
      "loss": 0.0724,
      "step": 1688
    },
    {
      "epoch": 21.584664536741215,
      "grad_norm": 2.9565556049346924,
      "learning_rate": 1.133846153846154e-05,
      "loss": 0.0624,
      "step": 1689
    },
    {
      "epoch": 21.597444089456868,
      "grad_norm": 4.148669242858887,
      "learning_rate": 1.1333333333333334e-05,
      "loss": 0.1211,
      "step": 1690
    },
    {
      "epoch": 21.610223642172524,
      "grad_norm": 5.265037536621094,
      "learning_rate": 1.132820512820513e-05,
      "loss": 0.1015,
      "step": 1691
    },
    {
      "epoch": 21.623003194888177,
      "grad_norm": 4.325460910797119,
      "learning_rate": 1.1323076923076924e-05,
      "loss": 0.1633,
      "step": 1692
    },
    {
      "epoch": 21.635782747603834,
      "grad_norm": 5.78082275390625,
      "learning_rate": 1.131794871794872e-05,
      "loss": 0.1113,
      "step": 1693
    },
    {
      "epoch": 21.64856230031949,
      "grad_norm": 4.066515922546387,
      "learning_rate": 1.1312820512820515e-05,
      "loss": 0.0874,
      "step": 1694
    },
    {
      "epoch": 21.661341853035143,
      "grad_norm": 4.018939018249512,
      "learning_rate": 1.1307692307692309e-05,
      "loss": 0.0933,
      "step": 1695
    },
    {
      "epoch": 21.6741214057508,
      "grad_norm": 2.7681217193603516,
      "learning_rate": 1.1302564102564105e-05,
      "loss": 0.0521,
      "step": 1696
    },
    {
      "epoch": 21.686900958466452,
      "grad_norm": 4.680244445800781,
      "learning_rate": 1.1297435897435897e-05,
      "loss": 0.1296,
      "step": 1697
    },
    {
      "epoch": 21.69968051118211,
      "grad_norm": 3.169238805770874,
      "learning_rate": 1.1292307692307693e-05,
      "loss": 0.0533,
      "step": 1698
    },
    {
      "epoch": 21.712460063897765,
      "grad_norm": 4.816842555999756,
      "learning_rate": 1.1287179487179487e-05,
      "loss": 0.1525,
      "step": 1699
    },
    {
      "epoch": 21.72523961661342,
      "grad_norm": 3.364724636077881,
      "learning_rate": 1.1282051282051283e-05,
      "loss": 0.0825,
      "step": 1700
    },
    {
      "epoch": 21.72523961661342,
      "eval_loss": 0.680626630783081,
      "eval_runtime": 184.1048,
      "eval_samples_per_second": 0.853,
      "eval_steps_per_second": 0.109,
      "step": 1700
    },
    {
      "epoch": 21.738019169329075,
      "grad_norm": 6.44060754776001,
      "learning_rate": 1.1276923076923079e-05,
      "loss": 0.1274,
      "step": 1701
    },
    {
      "epoch": 21.750798722044728,
      "grad_norm": 7.465785026550293,
      "learning_rate": 1.1271794871794873e-05,
      "loss": 0.0803,
      "step": 1702
    },
    {
      "epoch": 21.763578274760384,
      "grad_norm": 4.474442481994629,
      "learning_rate": 1.1266666666666668e-05,
      "loss": 0.0981,
      "step": 1703
    },
    {
      "epoch": 21.776357827476037,
      "grad_norm": 3.418708324432373,
      "learning_rate": 1.126153846153846e-05,
      "loss": 0.084,
      "step": 1704
    },
    {
      "epoch": 21.789137380191693,
      "grad_norm": 3.9616029262542725,
      "learning_rate": 1.1256410256410258e-05,
      "loss": 0.0726,
      "step": 1705
    },
    {
      "epoch": 21.80191693290735,
      "grad_norm": 6.344253063201904,
      "learning_rate": 1.1251282051282054e-05,
      "loss": 0.1246,
      "step": 1706
    },
    {
      "epoch": 21.814696485623003,
      "grad_norm": 3.4718990325927734,
      "learning_rate": 1.1246153846153846e-05,
      "loss": 0.0749,
      "step": 1707
    },
    {
      "epoch": 21.82747603833866,
      "grad_norm": 2.6628403663635254,
      "learning_rate": 1.1241025641025642e-05,
      "loss": 0.067,
      "step": 1708
    },
    {
      "epoch": 21.840255591054312,
      "grad_norm": 5.664834976196289,
      "learning_rate": 1.1235897435897436e-05,
      "loss": 0.1274,
      "step": 1709
    },
    {
      "epoch": 21.85303514376997,
      "grad_norm": 4.317172527313232,
      "learning_rate": 1.1230769230769232e-05,
      "loss": 0.0867,
      "step": 1710
    },
    {
      "epoch": 21.86581469648562,
      "grad_norm": 4.82378625869751,
      "learning_rate": 1.1225641025641028e-05,
      "loss": 0.0951,
      "step": 1711
    },
    {
      "epoch": 21.878594249201278,
      "grad_norm": 4.036247253417969,
      "learning_rate": 1.1220512820512822e-05,
      "loss": 0.1049,
      "step": 1712
    },
    {
      "epoch": 21.891373801916934,
      "grad_norm": 4.39935302734375,
      "learning_rate": 1.1215384615384617e-05,
      "loss": 0.1128,
      "step": 1713
    },
    {
      "epoch": 21.904153354632587,
      "grad_norm": 4.117251396179199,
      "learning_rate": 1.121025641025641e-05,
      "loss": 0.0852,
      "step": 1714
    },
    {
      "epoch": 21.916932907348244,
      "grad_norm": 4.607010841369629,
      "learning_rate": 1.1205128205128205e-05,
      "loss": 0.1085,
      "step": 1715
    },
    {
      "epoch": 21.929712460063897,
      "grad_norm": 3.569995641708374,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.0757,
      "step": 1716
    },
    {
      "epoch": 21.942492012779553,
      "grad_norm": 3.508486747741699,
      "learning_rate": 1.1194871794871795e-05,
      "loss": 0.0843,
      "step": 1717
    },
    {
      "epoch": 21.955271565495206,
      "grad_norm": 3.5567374229431152,
      "learning_rate": 1.1189743589743591e-05,
      "loss": 0.0935,
      "step": 1718
    },
    {
      "epoch": 21.968051118210862,
      "grad_norm": 4.282688617706299,
      "learning_rate": 1.1184615384615385e-05,
      "loss": 0.1008,
      "step": 1719
    },
    {
      "epoch": 21.98083067092652,
      "grad_norm": 3.526252508163452,
      "learning_rate": 1.117948717948718e-05,
      "loss": 0.0816,
      "step": 1720
    },
    {
      "epoch": 21.99361022364217,
      "grad_norm": 3.7250707149505615,
      "learning_rate": 1.1174358974358975e-05,
      "loss": 0.0985,
      "step": 1721
    },
    {
      "epoch": 22.00638977635783,
      "grad_norm": 3.0080602169036865,
      "learning_rate": 1.116923076923077e-05,
      "loss": 0.055,
      "step": 1722
    },
    {
      "epoch": 22.01916932907348,
      "grad_norm": 3.0814597606658936,
      "learning_rate": 1.1164102564102566e-05,
      "loss": 0.0546,
      "step": 1723
    },
    {
      "epoch": 22.031948881789138,
      "grad_norm": 2.9059879779815674,
      "learning_rate": 1.1158974358974359e-05,
      "loss": 0.055,
      "step": 1724
    },
    {
      "epoch": 22.044728434504794,
      "grad_norm": 2.3464226722717285,
      "learning_rate": 1.1153846153846154e-05,
      "loss": 0.0488,
      "step": 1725
    },
    {
      "epoch": 22.057507987220447,
      "grad_norm": 3.984386444091797,
      "learning_rate": 1.1148717948717948e-05,
      "loss": 0.0724,
      "step": 1726
    },
    {
      "epoch": 22.070287539936103,
      "grad_norm": 3.7431716918945312,
      "learning_rate": 1.1143589743589744e-05,
      "loss": 0.0606,
      "step": 1727
    },
    {
      "epoch": 22.083067092651756,
      "grad_norm": 2.889275074005127,
      "learning_rate": 1.113846153846154e-05,
      "loss": 0.0478,
      "step": 1728
    },
    {
      "epoch": 22.095846645367413,
      "grad_norm": 4.641188621520996,
      "learning_rate": 1.1133333333333334e-05,
      "loss": 0.1025,
      "step": 1729
    },
    {
      "epoch": 22.108626198083066,
      "grad_norm": 3.6719138622283936,
      "learning_rate": 1.112820512820513e-05,
      "loss": 0.0824,
      "step": 1730
    },
    {
      "epoch": 22.121405750798722,
      "grad_norm": 4.5112810134887695,
      "learning_rate": 1.1123076923076924e-05,
      "loss": 0.0888,
      "step": 1731
    },
    {
      "epoch": 22.13418530351438,
      "grad_norm": 3.478308916091919,
      "learning_rate": 1.111794871794872e-05,
      "loss": 0.0689,
      "step": 1732
    },
    {
      "epoch": 22.14696485623003,
      "grad_norm": 2.2676029205322266,
      "learning_rate": 1.1112820512820515e-05,
      "loss": 0.0407,
      "step": 1733
    },
    {
      "epoch": 22.159744408945688,
      "grad_norm": 3.8547441959381104,
      "learning_rate": 1.1107692307692308e-05,
      "loss": 0.0598,
      "step": 1734
    },
    {
      "epoch": 22.17252396166134,
      "grad_norm": 3.650141716003418,
      "learning_rate": 1.1102564102564103e-05,
      "loss": 0.08,
      "step": 1735
    },
    {
      "epoch": 22.185303514376997,
      "grad_norm": 3.7249064445495605,
      "learning_rate": 1.1097435897435897e-05,
      "loss": 0.0573,
      "step": 1736
    },
    {
      "epoch": 22.19808306709265,
      "grad_norm": 2.4170174598693848,
      "learning_rate": 1.1092307692307693e-05,
      "loss": 0.0462,
      "step": 1737
    },
    {
      "epoch": 22.210862619808307,
      "grad_norm": 3.232990026473999,
      "learning_rate": 1.1087179487179489e-05,
      "loss": 0.0778,
      "step": 1738
    },
    {
      "epoch": 22.223642172523963,
      "grad_norm": 3.2713253498077393,
      "learning_rate": 1.1082051282051283e-05,
      "loss": 0.0375,
      "step": 1739
    },
    {
      "epoch": 22.236421725239616,
      "grad_norm": 2.955406427383423,
      "learning_rate": 1.1076923076923079e-05,
      "loss": 0.0592,
      "step": 1740
    },
    {
      "epoch": 22.249201277955272,
      "grad_norm": 3.5802533626556396,
      "learning_rate": 1.1071794871794873e-05,
      "loss": 0.072,
      "step": 1741
    },
    {
      "epoch": 22.261980830670925,
      "grad_norm": 4.030676364898682,
      "learning_rate": 1.1066666666666669e-05,
      "loss": 0.0934,
      "step": 1742
    },
    {
      "epoch": 22.27476038338658,
      "grad_norm": 4.585970401763916,
      "learning_rate": 1.1061538461538461e-05,
      "loss": 0.0751,
      "step": 1743
    },
    {
      "epoch": 22.287539936102238,
      "grad_norm": 3.766411781311035,
      "learning_rate": 1.1056410256410257e-05,
      "loss": 0.0828,
      "step": 1744
    },
    {
      "epoch": 22.30031948881789,
      "grad_norm": 2.6844403743743896,
      "learning_rate": 1.1051282051282052e-05,
      "loss": 0.0778,
      "step": 1745
    },
    {
      "epoch": 22.313099041533548,
      "grad_norm": 2.612366199493408,
      "learning_rate": 1.1046153846153846e-05,
      "loss": 0.0561,
      "step": 1746
    },
    {
      "epoch": 22.3258785942492,
      "grad_norm": 3.555194854736328,
      "learning_rate": 1.1041025641025642e-05,
      "loss": 0.0822,
      "step": 1747
    },
    {
      "epoch": 22.338658146964857,
      "grad_norm": 3.7457146644592285,
      "learning_rate": 1.1035897435897436e-05,
      "loss": 0.1377,
      "step": 1748
    },
    {
      "epoch": 22.35143769968051,
      "grad_norm": 3.957167625427246,
      "learning_rate": 1.1030769230769232e-05,
      "loss": 0.0838,
      "step": 1749
    },
    {
      "epoch": 22.364217252396166,
      "grad_norm": 3.65297794342041,
      "learning_rate": 1.1025641025641028e-05,
      "loss": 0.0689,
      "step": 1750
    },
    {
      "epoch": 22.376996805111823,
      "grad_norm": 7.409717082977295,
      "learning_rate": 1.102051282051282e-05,
      "loss": 0.1203,
      "step": 1751
    },
    {
      "epoch": 22.389776357827476,
      "grad_norm": 5.040288925170898,
      "learning_rate": 1.1015384615384616e-05,
      "loss": 0.0896,
      "step": 1752
    },
    {
      "epoch": 22.402555910543132,
      "grad_norm": 3.3070826530456543,
      "learning_rate": 1.101025641025641e-05,
      "loss": 0.0815,
      "step": 1753
    },
    {
      "epoch": 22.415335463258785,
      "grad_norm": 4.675379276275635,
      "learning_rate": 1.1005128205128206e-05,
      "loss": 0.0757,
      "step": 1754
    },
    {
      "epoch": 22.42811501597444,
      "grad_norm": 3.5492966175079346,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.0759,
      "step": 1755
    },
    {
      "epoch": 22.440894568690094,
      "grad_norm": 2.287569999694824,
      "learning_rate": 1.0994871794871795e-05,
      "loss": 0.0337,
      "step": 1756
    },
    {
      "epoch": 22.45367412140575,
      "grad_norm": 4.1564764976501465,
      "learning_rate": 1.0989743589743591e-05,
      "loss": 0.0822,
      "step": 1757
    },
    {
      "epoch": 22.466453674121407,
      "grad_norm": 3.486571788787842,
      "learning_rate": 1.0984615384615385e-05,
      "loss": 0.0822,
      "step": 1758
    },
    {
      "epoch": 22.47923322683706,
      "grad_norm": 7.400767803192139,
      "learning_rate": 1.0979487179487181e-05,
      "loss": 0.1447,
      "step": 1759
    },
    {
      "epoch": 22.492012779552716,
      "grad_norm": 3.0589351654052734,
      "learning_rate": 1.0974358974358977e-05,
      "loss": 0.0506,
      "step": 1760
    },
    {
      "epoch": 22.50479233226837,
      "grad_norm": 2.8320963382720947,
      "learning_rate": 1.0969230769230769e-05,
      "loss": 0.0386,
      "step": 1761
    },
    {
      "epoch": 22.517571884984026,
      "grad_norm": 2.8496134281158447,
      "learning_rate": 1.0964102564102565e-05,
      "loss": 0.0607,
      "step": 1762
    },
    {
      "epoch": 22.53035143769968,
      "grad_norm": 3.732996702194214,
      "learning_rate": 1.0958974358974359e-05,
      "loss": 0.0981,
      "step": 1763
    },
    {
      "epoch": 22.543130990415335,
      "grad_norm": 4.311052322387695,
      "learning_rate": 1.0953846153846155e-05,
      "loss": 0.1022,
      "step": 1764
    },
    {
      "epoch": 22.55591054313099,
      "grad_norm": 3.167752504348755,
      "learning_rate": 1.094871794871795e-05,
      "loss": 0.0496,
      "step": 1765
    },
    {
      "epoch": 22.568690095846645,
      "grad_norm": 3.640716314315796,
      "learning_rate": 1.0943589743589744e-05,
      "loss": 0.0579,
      "step": 1766
    },
    {
      "epoch": 22.5814696485623,
      "grad_norm": 3.016212224960327,
      "learning_rate": 1.093846153846154e-05,
      "loss": 0.0725,
      "step": 1767
    },
    {
      "epoch": 22.594249201277954,
      "grad_norm": 3.809197187423706,
      "learning_rate": 1.0933333333333334e-05,
      "loss": 0.0548,
      "step": 1768
    },
    {
      "epoch": 22.60702875399361,
      "grad_norm": 3.489170789718628,
      "learning_rate": 1.092820512820513e-05,
      "loss": 0.0876,
      "step": 1769
    },
    {
      "epoch": 22.619808306709267,
      "grad_norm": 4.824202537536621,
      "learning_rate": 1.0923076923076922e-05,
      "loss": 0.0878,
      "step": 1770
    },
    {
      "epoch": 22.63258785942492,
      "grad_norm": 2.8183517456054688,
      "learning_rate": 1.0917948717948718e-05,
      "loss": 0.0377,
      "step": 1771
    },
    {
      "epoch": 22.645367412140576,
      "grad_norm": 3.3342647552490234,
      "learning_rate": 1.0912820512820514e-05,
      "loss": 0.0507,
      "step": 1772
    },
    {
      "epoch": 22.65814696485623,
      "grad_norm": 11.420624732971191,
      "learning_rate": 1.0907692307692308e-05,
      "loss": 0.117,
      "step": 1773
    },
    {
      "epoch": 22.670926517571885,
      "grad_norm": 5.182732582092285,
      "learning_rate": 1.0902564102564104e-05,
      "loss": 0.0992,
      "step": 1774
    },
    {
      "epoch": 22.68370607028754,
      "grad_norm": 3.9871864318847656,
      "learning_rate": 1.0897435897435898e-05,
      "loss": 0.063,
      "step": 1775
    },
    {
      "epoch": 22.696485623003195,
      "grad_norm": 3.9142467975616455,
      "learning_rate": 1.0892307692307693e-05,
      "loss": 0.057,
      "step": 1776
    },
    {
      "epoch": 22.70926517571885,
      "grad_norm": 2.7767791748046875,
      "learning_rate": 1.088717948717949e-05,
      "loss": 0.0532,
      "step": 1777
    },
    {
      "epoch": 22.722044728434504,
      "grad_norm": 9.83389949798584,
      "learning_rate": 1.0882051282051283e-05,
      "loss": 0.1355,
      "step": 1778
    },
    {
      "epoch": 22.73482428115016,
      "grad_norm": 3.4298386573791504,
      "learning_rate": 1.0876923076923079e-05,
      "loss": 0.0562,
      "step": 1779
    },
    {
      "epoch": 22.747603833865814,
      "grad_norm": 4.0902485847473145,
      "learning_rate": 1.0871794871794871e-05,
      "loss": 0.1059,
      "step": 1780
    },
    {
      "epoch": 22.76038338658147,
      "grad_norm": 2.769785165786743,
      "learning_rate": 1.0866666666666667e-05,
      "loss": 0.063,
      "step": 1781
    },
    {
      "epoch": 22.773162939297123,
      "grad_norm": 4.616344928741455,
      "learning_rate": 1.0861538461538463e-05,
      "loss": 0.0676,
      "step": 1782
    },
    {
      "epoch": 22.78594249201278,
      "grad_norm": 5.222856521606445,
      "learning_rate": 1.0856410256410257e-05,
      "loss": 0.0986,
      "step": 1783
    },
    {
      "epoch": 22.798722044728436,
      "grad_norm": 3.8730037212371826,
      "learning_rate": 1.0851282051282053e-05,
      "loss": 0.0627,
      "step": 1784
    },
    {
      "epoch": 22.81150159744409,
      "grad_norm": 4.0193681716918945,
      "learning_rate": 1.0846153846153847e-05,
      "loss": 0.0955,
      "step": 1785
    },
    {
      "epoch": 22.824281150159745,
      "grad_norm": 4.4240007400512695,
      "learning_rate": 1.0841025641025642e-05,
      "loss": 0.0901,
      "step": 1786
    },
    {
      "epoch": 22.837060702875398,
      "grad_norm": 3.280074119567871,
      "learning_rate": 1.0835897435897438e-05,
      "loss": 0.0732,
      "step": 1787
    },
    {
      "epoch": 22.849840255591054,
      "grad_norm": 3.105015993118286,
      "learning_rate": 1.083076923076923e-05,
      "loss": 0.0615,
      "step": 1788
    },
    {
      "epoch": 22.86261980830671,
      "grad_norm": 5.285585403442383,
      "learning_rate": 1.0825641025641028e-05,
      "loss": 0.0824,
      "step": 1789
    },
    {
      "epoch": 22.875399361022364,
      "grad_norm": 6.821863174438477,
      "learning_rate": 1.082051282051282e-05,
      "loss": 0.0929,
      "step": 1790
    },
    {
      "epoch": 22.88817891373802,
      "grad_norm": 7.419590473175049,
      "learning_rate": 1.0815384615384616e-05,
      "loss": 0.0764,
      "step": 1791
    },
    {
      "epoch": 22.900958466453673,
      "grad_norm": 6.400094032287598,
      "learning_rate": 1.081025641025641e-05,
      "loss": 0.1308,
      "step": 1792
    },
    {
      "epoch": 22.91373801916933,
      "grad_norm": 4.24391508102417,
      "learning_rate": 1.0805128205128206e-05,
      "loss": 0.0671,
      "step": 1793
    },
    {
      "epoch": 22.926517571884983,
      "grad_norm": 4.861712455749512,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.0779,
      "step": 1794
    },
    {
      "epoch": 22.93929712460064,
      "grad_norm": 10.362529754638672,
      "learning_rate": 1.0794871794871796e-05,
      "loss": 0.0948,
      "step": 1795
    },
    {
      "epoch": 22.952076677316295,
      "grad_norm": 3.3236842155456543,
      "learning_rate": 1.0789743589743591e-05,
      "loss": 0.0666,
      "step": 1796
    },
    {
      "epoch": 22.96485623003195,
      "grad_norm": 4.896422863006592,
      "learning_rate": 1.0784615384615384e-05,
      "loss": 0.0983,
      "step": 1797
    },
    {
      "epoch": 22.977635782747605,
      "grad_norm": 3.114579916000366,
      "learning_rate": 1.077948717948718e-05,
      "loss": 0.0882,
      "step": 1798
    },
    {
      "epoch": 22.990415335463258,
      "grad_norm": 3.008934497833252,
      "learning_rate": 1.0774358974358975e-05,
      "loss": 0.0734,
      "step": 1799
    },
    {
      "epoch": 23.003194888178914,
      "grad_norm": 4.6161208152771,
      "learning_rate": 1.076923076923077e-05,
      "loss": 0.087,
      "step": 1800
    },
    {
      "epoch": 23.003194888178914,
      "eval_loss": 0.7027307748794556,
      "eval_runtime": 183.6326,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 1800
    },
    {
      "epoch": 23.015974440894567,
      "grad_norm": 2.4176926612854004,
      "learning_rate": 1.0764102564102565e-05,
      "loss": 0.0332,
      "step": 1801
    },
    {
      "epoch": 23.028753993610223,
      "grad_norm": 5.2224321365356445,
      "learning_rate": 1.0758974358974359e-05,
      "loss": 0.0607,
      "step": 1802
    },
    {
      "epoch": 23.04153354632588,
      "grad_norm": 3.5653960704803467,
      "learning_rate": 1.0753846153846155e-05,
      "loss": 0.0661,
      "step": 1803
    },
    {
      "epoch": 23.054313099041533,
      "grad_norm": 5.087371349334717,
      "learning_rate": 1.074871794871795e-05,
      "loss": 0.0714,
      "step": 1804
    },
    {
      "epoch": 23.06709265175719,
      "grad_norm": 2.5737032890319824,
      "learning_rate": 1.0743589743589745e-05,
      "loss": 0.0641,
      "step": 1805
    },
    {
      "epoch": 23.079872204472842,
      "grad_norm": 5.5952301025390625,
      "learning_rate": 1.073846153846154e-05,
      "loss": 0.0606,
      "step": 1806
    },
    {
      "epoch": 23.0926517571885,
      "grad_norm": 2.490704298019409,
      "learning_rate": 1.0733333333333333e-05,
      "loss": 0.0488,
      "step": 1807
    },
    {
      "epoch": 23.105431309904155,
      "grad_norm": 2.7343122959136963,
      "learning_rate": 1.0728205128205129e-05,
      "loss": 0.0427,
      "step": 1808
    },
    {
      "epoch": 23.118210862619808,
      "grad_norm": 2.6564457416534424,
      "learning_rate": 1.0723076923076924e-05,
      "loss": 0.0425,
      "step": 1809
    },
    {
      "epoch": 23.130990415335464,
      "grad_norm": 2.8839149475097656,
      "learning_rate": 1.0717948717948718e-05,
      "loss": 0.0586,
      "step": 1810
    },
    {
      "epoch": 23.143769968051117,
      "grad_norm": 2.411672592163086,
      "learning_rate": 1.0712820512820514e-05,
      "loss": 0.055,
      "step": 1811
    },
    {
      "epoch": 23.156549520766774,
      "grad_norm": 4.945863723754883,
      "learning_rate": 1.0707692307692308e-05,
      "loss": 0.0763,
      "step": 1812
    },
    {
      "epoch": 23.169329073482427,
      "grad_norm": 4.193752288818359,
      "learning_rate": 1.0702564102564104e-05,
      "loss": 0.0667,
      "step": 1813
    },
    {
      "epoch": 23.182108626198083,
      "grad_norm": 3.646803140640259,
      "learning_rate": 1.06974358974359e-05,
      "loss": 0.0677,
      "step": 1814
    },
    {
      "epoch": 23.19488817891374,
      "grad_norm": 2.4909188747406006,
      "learning_rate": 1.0692307692307694e-05,
      "loss": 0.0541,
      "step": 1815
    },
    {
      "epoch": 23.207667731629392,
      "grad_norm": 4.249711513519287,
      "learning_rate": 1.068717948717949e-05,
      "loss": 0.0685,
      "step": 1816
    },
    {
      "epoch": 23.22044728434505,
      "grad_norm": 3.983074188232422,
      "learning_rate": 1.0682051282051282e-05,
      "loss": 0.0587,
      "step": 1817
    },
    {
      "epoch": 23.233226837060702,
      "grad_norm": 3.1439638137817383,
      "learning_rate": 1.0676923076923078e-05,
      "loss": 0.0511,
      "step": 1818
    },
    {
      "epoch": 23.24600638977636,
      "grad_norm": 2.408771514892578,
      "learning_rate": 1.0671794871794872e-05,
      "loss": 0.0319,
      "step": 1819
    },
    {
      "epoch": 23.25878594249201,
      "grad_norm": 3.978127956390381,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.0511,
      "step": 1820
    },
    {
      "epoch": 23.271565495207668,
      "grad_norm": 3.7737467288970947,
      "learning_rate": 1.0661538461538463e-05,
      "loss": 0.0651,
      "step": 1821
    },
    {
      "epoch": 23.284345047923324,
      "grad_norm": 2.5543150901794434,
      "learning_rate": 1.0656410256410257e-05,
      "loss": 0.0324,
      "step": 1822
    },
    {
      "epoch": 23.297124600638977,
      "grad_norm": 2.3911960124969482,
      "learning_rate": 1.0651282051282053e-05,
      "loss": 0.0446,
      "step": 1823
    },
    {
      "epoch": 23.309904153354633,
      "grad_norm": 3.72542405128479,
      "learning_rate": 1.0646153846153845e-05,
      "loss": 0.069,
      "step": 1824
    },
    {
      "epoch": 23.322683706070286,
      "grad_norm": 6.744537830352783,
      "learning_rate": 1.0641025641025643e-05,
      "loss": 0.0569,
      "step": 1825
    },
    {
      "epoch": 23.335463258785943,
      "grad_norm": 2.999110460281372,
      "learning_rate": 1.0635897435897438e-05,
      "loss": 0.0586,
      "step": 1826
    },
    {
      "epoch": 23.3482428115016,
      "grad_norm": 2.447598695755005,
      "learning_rate": 1.063076923076923e-05,
      "loss": 0.043,
      "step": 1827
    },
    {
      "epoch": 23.361022364217252,
      "grad_norm": 3.7535173892974854,
      "learning_rate": 1.0625641025641027e-05,
      "loss": 0.1043,
      "step": 1828
    },
    {
      "epoch": 23.37380191693291,
      "grad_norm": 4.816366672515869,
      "learning_rate": 1.062051282051282e-05,
      "loss": 0.0893,
      "step": 1829
    },
    {
      "epoch": 23.38658146964856,
      "grad_norm": 4.106537342071533,
      "learning_rate": 1.0615384615384616e-05,
      "loss": 0.0899,
      "step": 1830
    },
    {
      "epoch": 23.399361022364218,
      "grad_norm": 4.535812854766846,
      "learning_rate": 1.0610256410256412e-05,
      "loss": 0.0678,
      "step": 1831
    },
    {
      "epoch": 23.41214057507987,
      "grad_norm": 9.053688049316406,
      "learning_rate": 1.0605128205128206e-05,
      "loss": 0.0905,
      "step": 1832
    },
    {
      "epoch": 23.424920127795527,
      "grad_norm": 8.244091033935547,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 0.0627,
      "step": 1833
    },
    {
      "epoch": 23.437699680511184,
      "grad_norm": 5.316174030303955,
      "learning_rate": 1.0594871794871794e-05,
      "loss": 0.0652,
      "step": 1834
    },
    {
      "epoch": 23.450479233226837,
      "grad_norm": 4.909565448760986,
      "learning_rate": 1.058974358974359e-05,
      "loss": 0.0836,
      "step": 1835
    },
    {
      "epoch": 23.463258785942493,
      "grad_norm": 4.151371955871582,
      "learning_rate": 1.0584615384615386e-05,
      "loss": 0.0482,
      "step": 1836
    },
    {
      "epoch": 23.476038338658146,
      "grad_norm": 3.2326231002807617,
      "learning_rate": 1.057948717948718e-05,
      "loss": 0.0522,
      "step": 1837
    },
    {
      "epoch": 23.488817891373802,
      "grad_norm": 2.452183723449707,
      "learning_rate": 1.0574358974358975e-05,
      "loss": 0.042,
      "step": 1838
    },
    {
      "epoch": 23.501597444089455,
      "grad_norm": 2.3406856060028076,
      "learning_rate": 1.056923076923077e-05,
      "loss": 0.0421,
      "step": 1839
    },
    {
      "epoch": 23.51437699680511,
      "grad_norm": 4.51849889755249,
      "learning_rate": 1.0564102564102565e-05,
      "loss": 0.0851,
      "step": 1840
    },
    {
      "epoch": 23.527156549520768,
      "grad_norm": 4.629650115966797,
      "learning_rate": 1.055897435897436e-05,
      "loss": 0.0669,
      "step": 1841
    },
    {
      "epoch": 23.53993610223642,
      "grad_norm": 7.944886207580566,
      "learning_rate": 1.0553846153846155e-05,
      "loss": 0.1098,
      "step": 1842
    },
    {
      "epoch": 23.552715654952078,
      "grad_norm": 5.454171180725098,
      "learning_rate": 1.054871794871795e-05,
      "loss": 0.1067,
      "step": 1843
    },
    {
      "epoch": 23.56549520766773,
      "grad_norm": 4.480544090270996,
      "learning_rate": 1.0543589743589743e-05,
      "loss": 0.0768,
      "step": 1844
    },
    {
      "epoch": 23.578274760383387,
      "grad_norm": 4.150940895080566,
      "learning_rate": 1.0538461538461539e-05,
      "loss": 0.0517,
      "step": 1845
    },
    {
      "epoch": 23.591054313099043,
      "grad_norm": 3.9581117630004883,
      "learning_rate": 1.0533333333333333e-05,
      "loss": 0.0553,
      "step": 1846
    },
    {
      "epoch": 23.603833865814696,
      "grad_norm": 2.3760437965393066,
      "learning_rate": 1.0528205128205129e-05,
      "loss": 0.0593,
      "step": 1847
    },
    {
      "epoch": 23.616613418530353,
      "grad_norm": 3.2939131259918213,
      "learning_rate": 1.0523076923076924e-05,
      "loss": 0.0452,
      "step": 1848
    },
    {
      "epoch": 23.629392971246006,
      "grad_norm": 3.511467218399048,
      "learning_rate": 1.0517948717948719e-05,
      "loss": 0.0528,
      "step": 1849
    },
    {
      "epoch": 23.642172523961662,
      "grad_norm": 4.303265571594238,
      "learning_rate": 1.0512820512820514e-05,
      "loss": 0.0739,
      "step": 1850
    },
    {
      "epoch": 23.654952076677315,
      "grad_norm": 4.462754726409912,
      "learning_rate": 1.0507692307692308e-05,
      "loss": 0.0698,
      "step": 1851
    },
    {
      "epoch": 23.66773162939297,
      "grad_norm": 2.4054338932037354,
      "learning_rate": 1.0502564102564104e-05,
      "loss": 0.0327,
      "step": 1852
    },
    {
      "epoch": 23.680511182108628,
      "grad_norm": 3.435103178024292,
      "learning_rate": 1.04974358974359e-05,
      "loss": 0.0523,
      "step": 1853
    },
    {
      "epoch": 23.69329073482428,
      "grad_norm": 3.345883846282959,
      "learning_rate": 1.0492307692307692e-05,
      "loss": 0.0894,
      "step": 1854
    },
    {
      "epoch": 23.706070287539937,
      "grad_norm": 3.384603500366211,
      "learning_rate": 1.0487179487179488e-05,
      "loss": 0.0569,
      "step": 1855
    },
    {
      "epoch": 23.71884984025559,
      "grad_norm": 3.4234015941619873,
      "learning_rate": 1.0482051282051282e-05,
      "loss": 0.039,
      "step": 1856
    },
    {
      "epoch": 23.731629392971247,
      "grad_norm": 3.0080175399780273,
      "learning_rate": 1.0476923076923078e-05,
      "loss": 0.0455,
      "step": 1857
    },
    {
      "epoch": 23.7444089456869,
      "grad_norm": 3.3850762844085693,
      "learning_rate": 1.0471794871794873e-05,
      "loss": 0.0846,
      "step": 1858
    },
    {
      "epoch": 23.757188498402556,
      "grad_norm": 5.632648468017578,
      "learning_rate": 1.0466666666666668e-05,
      "loss": 0.0632,
      "step": 1859
    },
    {
      "epoch": 23.769968051118212,
      "grad_norm": 4.267935276031494,
      "learning_rate": 1.0461538461538463e-05,
      "loss": 0.0674,
      "step": 1860
    },
    {
      "epoch": 23.782747603833865,
      "grad_norm": 4.575784206390381,
      "learning_rate": 1.0456410256410257e-05,
      "loss": 0.0822,
      "step": 1861
    },
    {
      "epoch": 23.79552715654952,
      "grad_norm": 7.271210193634033,
      "learning_rate": 1.0451282051282053e-05,
      "loss": 0.1074,
      "step": 1862
    },
    {
      "epoch": 23.808306709265175,
      "grad_norm": 4.030458927154541,
      "learning_rate": 1.0446153846153849e-05,
      "loss": 0.0714,
      "step": 1863
    },
    {
      "epoch": 23.82108626198083,
      "grad_norm": 5.308885097503662,
      "learning_rate": 1.0441025641025641e-05,
      "loss": 0.0888,
      "step": 1864
    },
    {
      "epoch": 23.833865814696484,
      "grad_norm": 5.200422763824463,
      "learning_rate": 1.0435897435897437e-05,
      "loss": 0.0694,
      "step": 1865
    },
    {
      "epoch": 23.84664536741214,
      "grad_norm": 2.9071249961853027,
      "learning_rate": 1.0430769230769231e-05,
      "loss": 0.0497,
      "step": 1866
    },
    {
      "epoch": 23.859424920127797,
      "grad_norm": 3.448936939239502,
      "learning_rate": 1.0425641025641027e-05,
      "loss": 0.0596,
      "step": 1867
    },
    {
      "epoch": 23.87220447284345,
      "grad_norm": 5.066628456115723,
      "learning_rate": 1.042051282051282e-05,
      "loss": 0.1144,
      "step": 1868
    },
    {
      "epoch": 23.884984025559106,
      "grad_norm": 3.3270418643951416,
      "learning_rate": 1.0415384615384617e-05,
      "loss": 0.0448,
      "step": 1869
    },
    {
      "epoch": 23.89776357827476,
      "grad_norm": 4.035781383514404,
      "learning_rate": 1.0410256410256412e-05,
      "loss": 0.036,
      "step": 1870
    },
    {
      "epoch": 23.910543130990416,
      "grad_norm": 4.189550399780273,
      "learning_rate": 1.0405128205128205e-05,
      "loss": 0.0761,
      "step": 1871
    },
    {
      "epoch": 23.923322683706072,
      "grad_norm": 4.90827751159668,
      "learning_rate": 1.04e-05,
      "loss": 0.057,
      "step": 1872
    },
    {
      "epoch": 23.936102236421725,
      "grad_norm": 6.179035186767578,
      "learning_rate": 1.0394871794871794e-05,
      "loss": 0.071,
      "step": 1873
    },
    {
      "epoch": 23.94888178913738,
      "grad_norm": 4.103394508361816,
      "learning_rate": 1.038974358974359e-05,
      "loss": 0.0509,
      "step": 1874
    },
    {
      "epoch": 23.961661341853034,
      "grad_norm": 3.149423599243164,
      "learning_rate": 1.0384615384615386e-05,
      "loss": 0.0585,
      "step": 1875
    },
    {
      "epoch": 23.97444089456869,
      "grad_norm": 7.171144485473633,
      "learning_rate": 1.037948717948718e-05,
      "loss": 0.1216,
      "step": 1876
    },
    {
      "epoch": 23.987220447284344,
      "grad_norm": 4.05762243270874,
      "learning_rate": 1.0374358974358976e-05,
      "loss": 0.0716,
      "step": 1877
    },
    {
      "epoch": 24.0,
      "grad_norm": 2.2400803565979004,
      "learning_rate": 1.036923076923077e-05,
      "loss": 0.0421,
      "step": 1878
    },
    {
      "epoch": 24.012779552715656,
      "grad_norm": 4.6594390869140625,
      "learning_rate": 1.0364102564102566e-05,
      "loss": 0.0667,
      "step": 1879
    },
    {
      "epoch": 24.02555910543131,
      "grad_norm": 2.486776351928711,
      "learning_rate": 1.0358974358974361e-05,
      "loss": 0.0367,
      "step": 1880
    },
    {
      "epoch": 24.038338658146966,
      "grad_norm": 2.264676332473755,
      "learning_rate": 1.0353846153846154e-05,
      "loss": 0.0314,
      "step": 1881
    },
    {
      "epoch": 24.05111821086262,
      "grad_norm": 3.9006893634796143,
      "learning_rate": 1.034871794871795e-05,
      "loss": 0.0685,
      "step": 1882
    },
    {
      "epoch": 24.063897763578275,
      "grad_norm": 7.912015438079834,
      "learning_rate": 1.0343589743589743e-05,
      "loss": 0.0558,
      "step": 1883
    },
    {
      "epoch": 24.076677316293928,
      "grad_norm": 3.379222869873047,
      "learning_rate": 1.033846153846154e-05,
      "loss": 0.0548,
      "step": 1884
    },
    {
      "epoch": 24.089456869009584,
      "grad_norm": 3.533841848373413,
      "learning_rate": 1.0333333333333335e-05,
      "loss": 0.0473,
      "step": 1885
    },
    {
      "epoch": 24.10223642172524,
      "grad_norm": 6.102855205535889,
      "learning_rate": 1.0328205128205129e-05,
      "loss": 0.0969,
      "step": 1886
    },
    {
      "epoch": 24.115015974440894,
      "grad_norm": 2.1626064777374268,
      "learning_rate": 1.0323076923076925e-05,
      "loss": 0.0393,
      "step": 1887
    },
    {
      "epoch": 24.12779552715655,
      "grad_norm": 2.1872377395629883,
      "learning_rate": 1.0317948717948719e-05,
      "loss": 0.0258,
      "step": 1888
    },
    {
      "epoch": 24.140575079872203,
      "grad_norm": 3.058176279067993,
      "learning_rate": 1.0312820512820515e-05,
      "loss": 0.0401,
      "step": 1889
    },
    {
      "epoch": 24.15335463258786,
      "grad_norm": 1.8977614641189575,
      "learning_rate": 1.0307692307692307e-05,
      "loss": 0.0352,
      "step": 1890
    },
    {
      "epoch": 24.166134185303516,
      "grad_norm": 3.599254846572876,
      "learning_rate": 1.0302564102564103e-05,
      "loss": 0.0491,
      "step": 1891
    },
    {
      "epoch": 24.17891373801917,
      "grad_norm": 4.911500453948975,
      "learning_rate": 1.0297435897435898e-05,
      "loss": 0.0512,
      "step": 1892
    },
    {
      "epoch": 24.191693290734825,
      "grad_norm": 3.6773841381073,
      "learning_rate": 1.0292307692307692e-05,
      "loss": 0.0503,
      "step": 1893
    },
    {
      "epoch": 24.20447284345048,
      "grad_norm": 2.7655606269836426,
      "learning_rate": 1.0287179487179488e-05,
      "loss": 0.0337,
      "step": 1894
    },
    {
      "epoch": 24.217252396166135,
      "grad_norm": 3.185929298400879,
      "learning_rate": 1.0282051282051282e-05,
      "loss": 0.0486,
      "step": 1895
    },
    {
      "epoch": 24.230031948881788,
      "grad_norm": 2.5345399379730225,
      "learning_rate": 1.0276923076923078e-05,
      "loss": 0.0418,
      "step": 1896
    },
    {
      "epoch": 24.242811501597444,
      "grad_norm": 3.7780027389526367,
      "learning_rate": 1.0271794871794874e-05,
      "loss": 0.0473,
      "step": 1897
    },
    {
      "epoch": 24.2555910543131,
      "grad_norm": 4.549624919891357,
      "learning_rate": 1.0266666666666668e-05,
      "loss": 0.0657,
      "step": 1898
    },
    {
      "epoch": 24.268370607028753,
      "grad_norm": 2.380779266357422,
      "learning_rate": 1.0261538461538464e-05,
      "loss": 0.0336,
      "step": 1899
    },
    {
      "epoch": 24.28115015974441,
      "grad_norm": 1.827032446861267,
      "learning_rate": 1.0256410256410256e-05,
      "loss": 0.0264,
      "step": 1900
    },
    {
      "epoch": 24.28115015974441,
      "eval_loss": 0.7386012077331543,
      "eval_runtime": 183.8505,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 1900
    },
    {
      "epoch": 24.293929712460063,
      "grad_norm": 2.8570687770843506,
      "learning_rate": 1.0251282051282052e-05,
      "loss": 0.0482,
      "step": 1901
    },
    {
      "epoch": 24.30670926517572,
      "grad_norm": 7.331059455871582,
      "learning_rate": 1.0246153846153847e-05,
      "loss": 0.0822,
      "step": 1902
    },
    {
      "epoch": 24.319488817891372,
      "grad_norm": 2.7676453590393066,
      "learning_rate": 1.0241025641025641e-05,
      "loss": 0.0311,
      "step": 1903
    },
    {
      "epoch": 24.33226837060703,
      "grad_norm": 1.8847846984863281,
      "learning_rate": 1.0235897435897437e-05,
      "loss": 0.0342,
      "step": 1904
    },
    {
      "epoch": 24.345047923322685,
      "grad_norm": 1.8493037223815918,
      "learning_rate": 1.0230769230769231e-05,
      "loss": 0.0267,
      "step": 1905
    },
    {
      "epoch": 24.357827476038338,
      "grad_norm": 1.7126322984695435,
      "learning_rate": 1.0225641025641027e-05,
      "loss": 0.0277,
      "step": 1906
    },
    {
      "epoch": 24.370607028753994,
      "grad_norm": 2.872636318206787,
      "learning_rate": 1.0220512820512823e-05,
      "loss": 0.0407,
      "step": 1907
    },
    {
      "epoch": 24.383386581469647,
      "grad_norm": 3.7814128398895264,
      "learning_rate": 1.0215384615384615e-05,
      "loss": 0.0338,
      "step": 1908
    },
    {
      "epoch": 24.396166134185304,
      "grad_norm": 8.474327087402344,
      "learning_rate": 1.0210256410256412e-05,
      "loss": 0.0839,
      "step": 1909
    },
    {
      "epoch": 24.408945686900957,
      "grad_norm": 4.64918327331543,
      "learning_rate": 1.0205128205128205e-05,
      "loss": 0.0722,
      "step": 1910
    },
    {
      "epoch": 24.421725239616613,
      "grad_norm": 3.299110174179077,
      "learning_rate": 1.02e-05,
      "loss": 0.0652,
      "step": 1911
    },
    {
      "epoch": 24.43450479233227,
      "grad_norm": 2.726750612258911,
      "learning_rate": 1.0194871794871796e-05,
      "loss": 0.0442,
      "step": 1912
    },
    {
      "epoch": 24.447284345047922,
      "grad_norm": 2.1698837280273438,
      "learning_rate": 1.018974358974359e-05,
      "loss": 0.0427,
      "step": 1913
    },
    {
      "epoch": 24.46006389776358,
      "grad_norm": 4.142160892486572,
      "learning_rate": 1.0184615384615386e-05,
      "loss": 0.0605,
      "step": 1914
    },
    {
      "epoch": 24.472843450479232,
      "grad_norm": 2.4861974716186523,
      "learning_rate": 1.017948717948718e-05,
      "loss": 0.0372,
      "step": 1915
    },
    {
      "epoch": 24.48562300319489,
      "grad_norm": 2.4246034622192383,
      "learning_rate": 1.0174358974358976e-05,
      "loss": 0.0332,
      "step": 1916
    },
    {
      "epoch": 24.498402555910545,
      "grad_norm": 4.91164493560791,
      "learning_rate": 1.0169230769230768e-05,
      "loss": 0.0329,
      "step": 1917
    },
    {
      "epoch": 24.511182108626198,
      "grad_norm": 5.635557174682617,
      "learning_rate": 1.0164102564102564e-05,
      "loss": 0.0745,
      "step": 1918
    },
    {
      "epoch": 24.523961661341854,
      "grad_norm": 6.509224891662598,
      "learning_rate": 1.015897435897436e-05,
      "loss": 0.0853,
      "step": 1919
    },
    {
      "epoch": 24.536741214057507,
      "grad_norm": 3.0304408073425293,
      "learning_rate": 1.0153846153846154e-05,
      "loss": 0.0568,
      "step": 1920
    },
    {
      "epoch": 24.549520766773163,
      "grad_norm": 4.4625959396362305,
      "learning_rate": 1.014871794871795e-05,
      "loss": 0.0489,
      "step": 1921
    },
    {
      "epoch": 24.562300319488816,
      "grad_norm": 3.8643152713775635,
      "learning_rate": 1.0143589743589744e-05,
      "loss": 0.0516,
      "step": 1922
    },
    {
      "epoch": 24.575079872204473,
      "grad_norm": 3.6063058376312256,
      "learning_rate": 1.013846153846154e-05,
      "loss": 0.0465,
      "step": 1923
    },
    {
      "epoch": 24.58785942492013,
      "grad_norm": 3.255998134613037,
      "learning_rate": 1.0133333333333335e-05,
      "loss": 0.0469,
      "step": 1924
    },
    {
      "epoch": 24.600638977635782,
      "grad_norm": 3.0367839336395264,
      "learning_rate": 1.012820512820513e-05,
      "loss": 0.0443,
      "step": 1925
    },
    {
      "epoch": 24.61341853035144,
      "grad_norm": 3.610450267791748,
      "learning_rate": 1.0123076923076925e-05,
      "loss": 0.0517,
      "step": 1926
    },
    {
      "epoch": 24.62619808306709,
      "grad_norm": 5.996733665466309,
      "learning_rate": 1.0117948717948717e-05,
      "loss": 0.0512,
      "step": 1927
    },
    {
      "epoch": 24.638977635782748,
      "grad_norm": 8.607100486755371,
      "learning_rate": 1.0112820512820513e-05,
      "loss": 0.0911,
      "step": 1928
    },
    {
      "epoch": 24.6517571884984,
      "grad_norm": 2.6721339225769043,
      "learning_rate": 1.0107692307692309e-05,
      "loss": 0.0366,
      "step": 1929
    },
    {
      "epoch": 24.664536741214057,
      "grad_norm": 2.7931532859802246,
      "learning_rate": 1.0102564102564103e-05,
      "loss": 0.0236,
      "step": 1930
    },
    {
      "epoch": 24.677316293929714,
      "grad_norm": 4.758185863494873,
      "learning_rate": 1.0097435897435899e-05,
      "loss": 0.0812,
      "step": 1931
    },
    {
      "epoch": 24.690095846645367,
      "grad_norm": 2.5296826362609863,
      "learning_rate": 1.0092307692307693e-05,
      "loss": 0.0384,
      "step": 1932
    },
    {
      "epoch": 24.702875399361023,
      "grad_norm": 4.656088829040527,
      "learning_rate": 1.0087179487179488e-05,
      "loss": 0.072,
      "step": 1933
    },
    {
      "epoch": 24.715654952076676,
      "grad_norm": 4.275123596191406,
      "learning_rate": 1.0082051282051284e-05,
      "loss": 0.0486,
      "step": 1934
    },
    {
      "epoch": 24.728434504792332,
      "grad_norm": 2.7582848072052,
      "learning_rate": 1.0076923076923078e-05,
      "loss": 0.0533,
      "step": 1935
    },
    {
      "epoch": 24.74121405750799,
      "grad_norm": 4.323748588562012,
      "learning_rate": 1.0071794871794874e-05,
      "loss": 0.069,
      "step": 1936
    },
    {
      "epoch": 24.75399361022364,
      "grad_norm": 4.799659729003906,
      "learning_rate": 1.0066666666666666e-05,
      "loss": 0.1001,
      "step": 1937
    },
    {
      "epoch": 24.766773162939298,
      "grad_norm": 4.191411018371582,
      "learning_rate": 1.0061538461538462e-05,
      "loss": 0.0508,
      "step": 1938
    },
    {
      "epoch": 24.77955271565495,
      "grad_norm": 2.4692318439483643,
      "learning_rate": 1.0056410256410256e-05,
      "loss": 0.0412,
      "step": 1939
    },
    {
      "epoch": 24.792332268370608,
      "grad_norm": 2.660069465637207,
      "learning_rate": 1.0051282051282052e-05,
      "loss": 0.0844,
      "step": 1940
    },
    {
      "epoch": 24.80511182108626,
      "grad_norm": 3.0512633323669434,
      "learning_rate": 1.0046153846153848e-05,
      "loss": 0.0372,
      "step": 1941
    },
    {
      "epoch": 24.817891373801917,
      "grad_norm": 4.098048210144043,
      "learning_rate": 1.0041025641025642e-05,
      "loss": 0.0682,
      "step": 1942
    },
    {
      "epoch": 24.830670926517573,
      "grad_norm": 2.6245460510253906,
      "learning_rate": 1.0035897435897437e-05,
      "loss": 0.0284,
      "step": 1943
    },
    {
      "epoch": 24.843450479233226,
      "grad_norm": 3.8218538761138916,
      "learning_rate": 1.0030769230769231e-05,
      "loss": 0.0786,
      "step": 1944
    },
    {
      "epoch": 24.856230031948883,
      "grad_norm": 4.563466548919678,
      "learning_rate": 1.0025641025641027e-05,
      "loss": 0.0896,
      "step": 1945
    },
    {
      "epoch": 24.869009584664536,
      "grad_norm": 2.8993422985076904,
      "learning_rate": 1.0020512820512823e-05,
      "loss": 0.0362,
      "step": 1946
    },
    {
      "epoch": 24.881789137380192,
      "grad_norm": 2.655593156814575,
      "learning_rate": 1.0015384615384615e-05,
      "loss": 0.0303,
      "step": 1947
    },
    {
      "epoch": 24.894568690095845,
      "grad_norm": 4.739352703094482,
      "learning_rate": 1.0010256410256411e-05,
      "loss": 0.0614,
      "step": 1948
    },
    {
      "epoch": 24.9073482428115,
      "grad_norm": 6.2021803855896,
      "learning_rate": 1.0005128205128205e-05,
      "loss": 0.0882,
      "step": 1949
    },
    {
      "epoch": 24.920127795527158,
      "grad_norm": 2.9262235164642334,
      "learning_rate": 1e-05,
      "loss": 0.0549,
      "step": 1950
    },
    {
      "epoch": 24.93290734824281,
      "grad_norm": 3.3608341217041016,
      "learning_rate": 9.994871794871795e-06,
      "loss": 0.0598,
      "step": 1951
    },
    {
      "epoch": 24.945686900958467,
      "grad_norm": 4.746756553649902,
      "learning_rate": 9.98974358974359e-06,
      "loss": 0.0448,
      "step": 1952
    },
    {
      "epoch": 24.95846645367412,
      "grad_norm": 5.411993026733398,
      "learning_rate": 9.984615384615386e-06,
      "loss": 0.0582,
      "step": 1953
    },
    {
      "epoch": 24.971246006389777,
      "grad_norm": 3.2365033626556396,
      "learning_rate": 9.97948717948718e-06,
      "loss": 0.07,
      "step": 1954
    },
    {
      "epoch": 24.984025559105433,
      "grad_norm": 2.531045913696289,
      "learning_rate": 9.974358974358974e-06,
      "loss": 0.0566,
      "step": 1955
    },
    {
      "epoch": 24.996805111821086,
      "grad_norm": 3.0786638259887695,
      "learning_rate": 9.96923076923077e-06,
      "loss": 0.0439,
      "step": 1956
    },
    {
      "epoch": 25.009584664536742,
      "grad_norm": 3.8696084022521973,
      "learning_rate": 9.964102564102564e-06,
      "loss": 0.0452,
      "step": 1957
    },
    {
      "epoch": 25.022364217252395,
      "grad_norm": 3.3101108074188232,
      "learning_rate": 9.95897435897436e-06,
      "loss": 0.0404,
      "step": 1958
    },
    {
      "epoch": 25.03514376996805,
      "grad_norm": 2.594020128250122,
      "learning_rate": 9.953846153846156e-06,
      "loss": 0.0346,
      "step": 1959
    },
    {
      "epoch": 25.047923322683705,
      "grad_norm": 3.4840526580810547,
      "learning_rate": 9.94871794871795e-06,
      "loss": 0.0588,
      "step": 1960
    },
    {
      "epoch": 25.06070287539936,
      "grad_norm": 1.951032042503357,
      "learning_rate": 9.943589743589744e-06,
      "loss": 0.0249,
      "step": 1961
    },
    {
      "epoch": 25.073482428115017,
      "grad_norm": 3.4109833240509033,
      "learning_rate": 9.93846153846154e-06,
      "loss": 0.0451,
      "step": 1962
    },
    {
      "epoch": 25.08626198083067,
      "grad_norm": 2.452439785003662,
      "learning_rate": 9.933333333333334e-06,
      "loss": 0.0328,
      "step": 1963
    },
    {
      "epoch": 25.099041533546327,
      "grad_norm": 3.8744537830352783,
      "learning_rate": 9.92820512820513e-06,
      "loss": 0.0558,
      "step": 1964
    },
    {
      "epoch": 25.11182108626198,
      "grad_norm": 2.6519272327423096,
      "learning_rate": 9.923076923076923e-06,
      "loss": 0.0304,
      "step": 1965
    },
    {
      "epoch": 25.124600638977636,
      "grad_norm": 4.713734149932861,
      "learning_rate": 9.91794871794872e-06,
      "loss": 0.043,
      "step": 1966
    },
    {
      "epoch": 25.13738019169329,
      "grad_norm": 1.7487976551055908,
      "learning_rate": 9.912820512820513e-06,
      "loss": 0.0248,
      "step": 1967
    },
    {
      "epoch": 25.150159744408946,
      "grad_norm": 3.6812212467193604,
      "learning_rate": 9.907692307692309e-06,
      "loss": 0.035,
      "step": 1968
    },
    {
      "epoch": 25.162939297124602,
      "grad_norm": 2.805856704711914,
      "learning_rate": 9.902564102564103e-06,
      "loss": 0.0358,
      "step": 1969
    },
    {
      "epoch": 25.175718849840255,
      "grad_norm": 2.701242446899414,
      "learning_rate": 9.897435897435899e-06,
      "loss": 0.0419,
      "step": 1970
    },
    {
      "epoch": 25.18849840255591,
      "grad_norm": 2.5276713371276855,
      "learning_rate": 9.892307692307693e-06,
      "loss": 0.0266,
      "step": 1971
    },
    {
      "epoch": 25.201277955271564,
      "grad_norm": 5.18175745010376,
      "learning_rate": 9.887179487179489e-06,
      "loss": 0.0282,
      "step": 1972
    },
    {
      "epoch": 25.21405750798722,
      "grad_norm": 8.315861701965332,
      "learning_rate": 9.882051282051283e-06,
      "loss": 0.0904,
      "step": 1973
    },
    {
      "epoch": 25.226837060702877,
      "grad_norm": 3.513111114501953,
      "learning_rate": 9.876923076923077e-06,
      "loss": 0.0458,
      "step": 1974
    },
    {
      "epoch": 25.23961661341853,
      "grad_norm": 3.5734593868255615,
      "learning_rate": 9.871794871794872e-06,
      "loss": 0.0688,
      "step": 1975
    },
    {
      "epoch": 25.252396166134186,
      "grad_norm": 2.2913684844970703,
      "learning_rate": 9.866666666666668e-06,
      "loss": 0.0305,
      "step": 1976
    },
    {
      "epoch": 25.26517571884984,
      "grad_norm": 4.16497802734375,
      "learning_rate": 9.861538461538462e-06,
      "loss": 0.0455,
      "step": 1977
    },
    {
      "epoch": 25.277955271565496,
      "grad_norm": 4.468432426452637,
      "learning_rate": 9.856410256410256e-06,
      "loss": 0.0491,
      "step": 1978
    },
    {
      "epoch": 25.29073482428115,
      "grad_norm": 2.0939643383026123,
      "learning_rate": 9.851282051282052e-06,
      "loss": 0.0303,
      "step": 1979
    },
    {
      "epoch": 25.303514376996805,
      "grad_norm": 1.9866693019866943,
      "learning_rate": 9.846153846153848e-06,
      "loss": 0.0208,
      "step": 1980
    },
    {
      "epoch": 25.31629392971246,
      "grad_norm": 2.8087332248687744,
      "learning_rate": 9.841025641025642e-06,
      "loss": 0.0328,
      "step": 1981
    },
    {
      "epoch": 25.329073482428115,
      "grad_norm": 3.123957633972168,
      "learning_rate": 9.835897435897438e-06,
      "loss": 0.034,
      "step": 1982
    },
    {
      "epoch": 25.34185303514377,
      "grad_norm": 2.317481279373169,
      "learning_rate": 9.830769230769232e-06,
      "loss": 0.0371,
      "step": 1983
    },
    {
      "epoch": 25.354632587859424,
      "grad_norm": 3.0201683044433594,
      "learning_rate": 9.825641025641026e-06,
      "loss": 0.038,
      "step": 1984
    },
    {
      "epoch": 25.36741214057508,
      "grad_norm": 3.6297476291656494,
      "learning_rate": 9.820512820512821e-06,
      "loss": 0.0508,
      "step": 1985
    },
    {
      "epoch": 25.380191693290733,
      "grad_norm": 2.2146573066711426,
      "learning_rate": 9.815384615384617e-06,
      "loss": 0.0329,
      "step": 1986
    },
    {
      "epoch": 25.39297124600639,
      "grad_norm": 2.0576748847961426,
      "learning_rate": 9.810256410256411e-06,
      "loss": 0.0383,
      "step": 1987
    },
    {
      "epoch": 25.405750798722046,
      "grad_norm": 2.456467390060425,
      "learning_rate": 9.805128205128205e-06,
      "loss": 0.0398,
      "step": 1988
    },
    {
      "epoch": 25.4185303514377,
      "grad_norm": 2.6752357482910156,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.0401,
      "step": 1989
    },
    {
      "epoch": 25.431309904153355,
      "grad_norm": 2.9409120082855225,
      "learning_rate": 9.794871794871795e-06,
      "loss": 0.0426,
      "step": 1990
    },
    {
      "epoch": 25.44408945686901,
      "grad_norm": 2.9854166507720947,
      "learning_rate": 9.78974358974359e-06,
      "loss": 0.0403,
      "step": 1991
    },
    {
      "epoch": 25.456869009584665,
      "grad_norm": 3.2052009105682373,
      "learning_rate": 9.784615384615387e-06,
      "loss": 0.0493,
      "step": 1992
    },
    {
      "epoch": 25.46964856230032,
      "grad_norm": 1.7671165466308594,
      "learning_rate": 9.77948717948718e-06,
      "loss": 0.0201,
      "step": 1993
    },
    {
      "epoch": 25.482428115015974,
      "grad_norm": 8.041123390197754,
      "learning_rate": 9.774358974358975e-06,
      "loss": 0.0452,
      "step": 1994
    },
    {
      "epoch": 25.49520766773163,
      "grad_norm": 2.9875450134277344,
      "learning_rate": 9.76923076923077e-06,
      "loss": 0.0777,
      "step": 1995
    },
    {
      "epoch": 25.507987220447284,
      "grad_norm": 3.887556552886963,
      "learning_rate": 9.764102564102564e-06,
      "loss": 0.0542,
      "step": 1996
    },
    {
      "epoch": 25.52076677316294,
      "grad_norm": 2.4877817630767822,
      "learning_rate": 9.75897435897436e-06,
      "loss": 0.0342,
      "step": 1997
    },
    {
      "epoch": 25.533546325878593,
      "grad_norm": 3.525219440460205,
      "learning_rate": 9.753846153846154e-06,
      "loss": 0.0647,
      "step": 1998
    },
    {
      "epoch": 25.54632587859425,
      "grad_norm": 3.614077568054199,
      "learning_rate": 9.74871794871795e-06,
      "loss": 0.0333,
      "step": 1999
    },
    {
      "epoch": 25.559105431309906,
      "grad_norm": 3.021026134490967,
      "learning_rate": 9.743589743589744e-06,
      "loss": 0.0458,
      "step": 2000
    },
    {
      "epoch": 25.559105431309906,
      "eval_loss": 0.7721496224403381,
      "eval_runtime": 183.9025,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 2000
    },
    {
      "epoch": 25.57188498402556,
      "grad_norm": 7.759271144866943,
      "learning_rate": 9.738461538461538e-06,
      "loss": 0.0931,
      "step": 2001
    },
    {
      "epoch": 25.584664536741215,
      "grad_norm": 2.6857120990753174,
      "learning_rate": 9.733333333333334e-06,
      "loss": 0.0421,
      "step": 2002
    },
    {
      "epoch": 25.597444089456868,
      "grad_norm": 2.8657867908477783,
      "learning_rate": 9.72820512820513e-06,
      "loss": 0.0413,
      "step": 2003
    },
    {
      "epoch": 25.610223642172524,
      "grad_norm": 2.1326651573181152,
      "learning_rate": 9.723076923076924e-06,
      "loss": 0.0257,
      "step": 2004
    },
    {
      "epoch": 25.623003194888177,
      "grad_norm": 3.7716381549835205,
      "learning_rate": 9.71794871794872e-06,
      "loss": 0.0297,
      "step": 2005
    },
    {
      "epoch": 25.635782747603834,
      "grad_norm": 2.4711177349090576,
      "learning_rate": 9.712820512820513e-06,
      "loss": 0.0344,
      "step": 2006
    },
    {
      "epoch": 25.64856230031949,
      "grad_norm": 2.9155871868133545,
      "learning_rate": 9.707692307692308e-06,
      "loss": 0.0447,
      "step": 2007
    },
    {
      "epoch": 25.661341853035143,
      "grad_norm": 2.907470703125,
      "learning_rate": 9.702564102564103e-06,
      "loss": 0.042,
      "step": 2008
    },
    {
      "epoch": 25.6741214057508,
      "grad_norm": 1.783984899520874,
      "learning_rate": 9.697435897435899e-06,
      "loss": 0.0172,
      "step": 2009
    },
    {
      "epoch": 25.686900958466452,
      "grad_norm": 3.024845600128174,
      "learning_rate": 9.692307692307693e-06,
      "loss": 0.0293,
      "step": 2010
    },
    {
      "epoch": 25.69968051118211,
      "grad_norm": 4.1440815925598145,
      "learning_rate": 9.687179487179487e-06,
      "loss": 0.0412,
      "step": 2011
    },
    {
      "epoch": 25.712460063897765,
      "grad_norm": 3.1864733695983887,
      "learning_rate": 9.682051282051283e-06,
      "loss": 0.0324,
      "step": 2012
    },
    {
      "epoch": 25.72523961661342,
      "grad_norm": 3.390996217727661,
      "learning_rate": 9.676923076923079e-06,
      "loss": 0.0287,
      "step": 2013
    },
    {
      "epoch": 25.738019169329075,
      "grad_norm": 5.366546630859375,
      "learning_rate": 9.671794871794873e-06,
      "loss": 0.0562,
      "step": 2014
    },
    {
      "epoch": 25.750798722044728,
      "grad_norm": 3.494955539703369,
      "learning_rate": 9.666666666666667e-06,
      "loss": 0.0489,
      "step": 2015
    },
    {
      "epoch": 25.763578274760384,
      "grad_norm": 4.710078239440918,
      "learning_rate": 9.661538461538462e-06,
      "loss": 0.0694,
      "step": 2016
    },
    {
      "epoch": 25.776357827476037,
      "grad_norm": 2.927586555480957,
      "learning_rate": 9.656410256410257e-06,
      "loss": 0.0492,
      "step": 2017
    },
    {
      "epoch": 25.789137380191693,
      "grad_norm": 2.5779459476470947,
      "learning_rate": 9.651282051282052e-06,
      "loss": 0.0371,
      "step": 2018
    },
    {
      "epoch": 25.80191693290735,
      "grad_norm": 7.566952705383301,
      "learning_rate": 9.646153846153848e-06,
      "loss": 0.0404,
      "step": 2019
    },
    {
      "epoch": 25.814696485623003,
      "grad_norm": 2.655378580093384,
      "learning_rate": 9.641025641025642e-06,
      "loss": 0.0301,
      "step": 2020
    },
    {
      "epoch": 25.82747603833866,
      "grad_norm": 9.1538724899292,
      "learning_rate": 9.635897435897436e-06,
      "loss": 0.0374,
      "step": 2021
    },
    {
      "epoch": 25.840255591054312,
      "grad_norm": 2.7713680267333984,
      "learning_rate": 9.630769230769232e-06,
      "loss": 0.0348,
      "step": 2022
    },
    {
      "epoch": 25.85303514376997,
      "grad_norm": 5.083139896392822,
      "learning_rate": 9.625641025641026e-06,
      "loss": 0.0648,
      "step": 2023
    },
    {
      "epoch": 25.86581469648562,
      "grad_norm": 3.456340789794922,
      "learning_rate": 9.620512820512822e-06,
      "loss": 0.0465,
      "step": 2024
    },
    {
      "epoch": 25.878594249201278,
      "grad_norm": 2.917619466781616,
      "learning_rate": 9.615384615384616e-06,
      "loss": 0.0486,
      "step": 2025
    },
    {
      "epoch": 25.891373801916934,
      "grad_norm": 4.159769058227539,
      "learning_rate": 9.610256410256411e-06,
      "loss": 0.0458,
      "step": 2026
    },
    {
      "epoch": 25.904153354632587,
      "grad_norm": 5.426984786987305,
      "learning_rate": 9.605128205128206e-06,
      "loss": 0.0679,
      "step": 2027
    },
    {
      "epoch": 25.916932907348244,
      "grad_norm": 2.974804639816284,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.0251,
      "step": 2028
    },
    {
      "epoch": 25.929712460063897,
      "grad_norm": 2.336801290512085,
      "learning_rate": 9.594871794871797e-06,
      "loss": 0.0337,
      "step": 2029
    },
    {
      "epoch": 25.942492012779553,
      "grad_norm": 3.529595375061035,
      "learning_rate": 9.589743589743591e-06,
      "loss": 0.0487,
      "step": 2030
    },
    {
      "epoch": 25.955271565495206,
      "grad_norm": 3.1817433834075928,
      "learning_rate": 9.584615384615385e-06,
      "loss": 0.042,
      "step": 2031
    },
    {
      "epoch": 25.968051118210862,
      "grad_norm": 4.293992042541504,
      "learning_rate": 9.579487179487181e-06,
      "loss": 0.0245,
      "step": 2032
    },
    {
      "epoch": 25.98083067092652,
      "grad_norm": 3.56746506690979,
      "learning_rate": 9.574358974358975e-06,
      "loss": 0.0403,
      "step": 2033
    },
    {
      "epoch": 25.99361022364217,
      "grad_norm": 2.8027122020721436,
      "learning_rate": 9.569230769230769e-06,
      "loss": 0.0311,
      "step": 2034
    },
    {
      "epoch": 26.00638977635783,
      "grad_norm": 3.442718267440796,
      "learning_rate": 9.564102564102565e-06,
      "loss": 0.0521,
      "step": 2035
    },
    {
      "epoch": 26.01916932907348,
      "grad_norm": 2.420042037963867,
      "learning_rate": 9.55897435897436e-06,
      "loss": 0.0343,
      "step": 2036
    },
    {
      "epoch": 26.031948881789138,
      "grad_norm": 3.9595730304718018,
      "learning_rate": 9.553846153846155e-06,
      "loss": 0.0401,
      "step": 2037
    },
    {
      "epoch": 26.044728434504794,
      "grad_norm": 3.0006279945373535,
      "learning_rate": 9.548717948717949e-06,
      "loss": 0.0319,
      "step": 2038
    },
    {
      "epoch": 26.057507987220447,
      "grad_norm": 2.6789944171905518,
      "learning_rate": 9.543589743589744e-06,
      "loss": 0.0245,
      "step": 2039
    },
    {
      "epoch": 26.070287539936103,
      "grad_norm": 1.75850510597229,
      "learning_rate": 9.53846153846154e-06,
      "loss": 0.0294,
      "step": 2040
    },
    {
      "epoch": 26.083067092651756,
      "grad_norm": 2.2789313793182373,
      "learning_rate": 9.533333333333334e-06,
      "loss": 0.0196,
      "step": 2041
    },
    {
      "epoch": 26.095846645367413,
      "grad_norm": 1.8305696249008179,
      "learning_rate": 9.52820512820513e-06,
      "loss": 0.0249,
      "step": 2042
    },
    {
      "epoch": 26.108626198083066,
      "grad_norm": 1.6572062969207764,
      "learning_rate": 9.523076923076924e-06,
      "loss": 0.0193,
      "step": 2043
    },
    {
      "epoch": 26.121405750798722,
      "grad_norm": 2.569342851638794,
      "learning_rate": 9.517948717948718e-06,
      "loss": 0.0449,
      "step": 2044
    },
    {
      "epoch": 26.13418530351438,
      "grad_norm": 4.079476356506348,
      "learning_rate": 9.512820512820514e-06,
      "loss": 0.0287,
      "step": 2045
    },
    {
      "epoch": 26.14696485623003,
      "grad_norm": 3.0456132888793945,
      "learning_rate": 9.50769230769231e-06,
      "loss": 0.0448,
      "step": 2046
    },
    {
      "epoch": 26.159744408945688,
      "grad_norm": 2.417229413986206,
      "learning_rate": 9.502564102564103e-06,
      "loss": 0.047,
      "step": 2047
    },
    {
      "epoch": 26.17252396166134,
      "grad_norm": 2.991654872894287,
      "learning_rate": 9.497435897435898e-06,
      "loss": 0.0459,
      "step": 2048
    },
    {
      "epoch": 26.185303514376997,
      "grad_norm": 2.4096415042877197,
      "learning_rate": 9.492307692307693e-06,
      "loss": 0.0259,
      "step": 2049
    },
    {
      "epoch": 26.19808306709265,
      "grad_norm": 2.3776893615722656,
      "learning_rate": 9.487179487179487e-06,
      "loss": 0.0269,
      "step": 2050
    },
    {
      "epoch": 26.210862619808307,
      "grad_norm": 3.019699811935425,
      "learning_rate": 9.482051282051283e-06,
      "loss": 0.026,
      "step": 2051
    },
    {
      "epoch": 26.223642172523963,
      "grad_norm": 3.0545332431793213,
      "learning_rate": 9.476923076923079e-06,
      "loss": 0.0377,
      "step": 2052
    },
    {
      "epoch": 26.236421725239616,
      "grad_norm": 4.2493062019348145,
      "learning_rate": 9.471794871794873e-06,
      "loss": 0.0342,
      "step": 2053
    },
    {
      "epoch": 26.249201277955272,
      "grad_norm": 3.70605206489563,
      "learning_rate": 9.466666666666667e-06,
      "loss": 0.0452,
      "step": 2054
    },
    {
      "epoch": 26.261980830670925,
      "grad_norm": 1.6265230178833008,
      "learning_rate": 9.461538461538463e-06,
      "loss": 0.0203,
      "step": 2055
    },
    {
      "epoch": 26.27476038338658,
      "grad_norm": 3.664158821105957,
      "learning_rate": 9.456410256410257e-06,
      "loss": 0.0199,
      "step": 2056
    },
    {
      "epoch": 26.287539936102238,
      "grad_norm": 2.3494155406951904,
      "learning_rate": 9.451282051282052e-06,
      "loss": 0.026,
      "step": 2057
    },
    {
      "epoch": 26.30031948881789,
      "grad_norm": 3.5722639560699463,
      "learning_rate": 9.446153846153847e-06,
      "loss": 0.0364,
      "step": 2058
    },
    {
      "epoch": 26.313099041533548,
      "grad_norm": 4.406216621398926,
      "learning_rate": 9.441025641025642e-06,
      "loss": 0.0331,
      "step": 2059
    },
    {
      "epoch": 26.3258785942492,
      "grad_norm": 1.9493281841278076,
      "learning_rate": 9.435897435897436e-06,
      "loss": 0.0352,
      "step": 2060
    },
    {
      "epoch": 26.338658146964857,
      "grad_norm": 2.5774435997009277,
      "learning_rate": 9.43076923076923e-06,
      "loss": 0.0268,
      "step": 2061
    },
    {
      "epoch": 26.35143769968051,
      "grad_norm": 2.915691375732422,
      "learning_rate": 9.425641025641026e-06,
      "loss": 0.0378,
      "step": 2062
    },
    {
      "epoch": 26.364217252396166,
      "grad_norm": 3.6906867027282715,
      "learning_rate": 9.420512820512822e-06,
      "loss": 0.0294,
      "step": 2063
    },
    {
      "epoch": 26.376996805111823,
      "grad_norm": 2.9189324378967285,
      "learning_rate": 9.415384615384616e-06,
      "loss": 0.0212,
      "step": 2064
    },
    {
      "epoch": 26.389776357827476,
      "grad_norm": 4.310835361480713,
      "learning_rate": 9.410256410256412e-06,
      "loss": 0.0604,
      "step": 2065
    },
    {
      "epoch": 26.402555910543132,
      "grad_norm": 2.1591827869415283,
      "learning_rate": 9.405128205128206e-06,
      "loss": 0.0287,
      "step": 2066
    },
    {
      "epoch": 26.415335463258785,
      "grad_norm": 4.369766712188721,
      "learning_rate": 9.4e-06,
      "loss": 0.036,
      "step": 2067
    },
    {
      "epoch": 26.42811501597444,
      "grad_norm": 4.2029876708984375,
      "learning_rate": 9.394871794871796e-06,
      "loss": 0.0441,
      "step": 2068
    },
    {
      "epoch": 26.440894568690094,
      "grad_norm": 3.996307849884033,
      "learning_rate": 9.389743589743591e-06,
      "loss": 0.0749,
      "step": 2069
    },
    {
      "epoch": 26.45367412140575,
      "grad_norm": 2.7402710914611816,
      "learning_rate": 9.384615384615385e-06,
      "loss": 0.026,
      "step": 2070
    },
    {
      "epoch": 26.466453674121407,
      "grad_norm": 1.8309948444366455,
      "learning_rate": 9.37948717948718e-06,
      "loss": 0.0223,
      "step": 2071
    },
    {
      "epoch": 26.47923322683706,
      "grad_norm": 2.682682752609253,
      "learning_rate": 9.374358974358975e-06,
      "loss": 0.0416,
      "step": 2072
    },
    {
      "epoch": 26.492012779552716,
      "grad_norm": 2.569913864135742,
      "learning_rate": 9.369230769230771e-06,
      "loss": 0.0348,
      "step": 2073
    },
    {
      "epoch": 26.50479233226837,
      "grad_norm": 2.0622801780700684,
      "learning_rate": 9.364102564102565e-06,
      "loss": 0.0552,
      "step": 2074
    },
    {
      "epoch": 26.517571884984026,
      "grad_norm": 3.3317339420318604,
      "learning_rate": 9.358974358974359e-06,
      "loss": 0.0274,
      "step": 2075
    },
    {
      "epoch": 26.53035143769968,
      "grad_norm": 2.5767219066619873,
      "learning_rate": 9.353846153846155e-06,
      "loss": 0.0391,
      "step": 2076
    },
    {
      "epoch": 26.543130990415335,
      "grad_norm": 1.9681037664413452,
      "learning_rate": 9.348717948717949e-06,
      "loss": 0.025,
      "step": 2077
    },
    {
      "epoch": 26.55591054313099,
      "grad_norm": 2.127232074737549,
      "learning_rate": 9.343589743589745e-06,
      "loss": 0.0322,
      "step": 2078
    },
    {
      "epoch": 26.568690095846645,
      "grad_norm": 4.699692249298096,
      "learning_rate": 9.33846153846154e-06,
      "loss": 0.0384,
      "step": 2079
    },
    {
      "epoch": 26.5814696485623,
      "grad_norm": 3.699765682220459,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.0491,
      "step": 2080
    },
    {
      "epoch": 26.594249201277954,
      "grad_norm": 2.842067241668701,
      "learning_rate": 9.328205128205128e-06,
      "loss": 0.0244,
      "step": 2081
    },
    {
      "epoch": 26.60702875399361,
      "grad_norm": 2.6938209533691406,
      "learning_rate": 9.323076923076924e-06,
      "loss": 0.0235,
      "step": 2082
    },
    {
      "epoch": 26.619808306709267,
      "grad_norm": 4.348531723022461,
      "learning_rate": 9.317948717948718e-06,
      "loss": 0.0521,
      "step": 2083
    },
    {
      "epoch": 26.63258785942492,
      "grad_norm": 2.8389244079589844,
      "learning_rate": 9.312820512820514e-06,
      "loss": 0.0423,
      "step": 2084
    },
    {
      "epoch": 26.645367412140576,
      "grad_norm": 1.9097580909729004,
      "learning_rate": 9.307692307692308e-06,
      "loss": 0.0278,
      "step": 2085
    },
    {
      "epoch": 26.65814696485623,
      "grad_norm": 3.038785696029663,
      "learning_rate": 9.302564102564104e-06,
      "loss": 0.0372,
      "step": 2086
    },
    {
      "epoch": 26.670926517571885,
      "grad_norm": 2.6908535957336426,
      "learning_rate": 9.297435897435898e-06,
      "loss": 0.0315,
      "step": 2087
    },
    {
      "epoch": 26.68370607028754,
      "grad_norm": 4.660581111907959,
      "learning_rate": 9.292307692307694e-06,
      "loss": 0.0456,
      "step": 2088
    },
    {
      "epoch": 26.696485623003195,
      "grad_norm": 4.069143772125244,
      "learning_rate": 9.28717948717949e-06,
      "loss": 0.0474,
      "step": 2089
    },
    {
      "epoch": 26.70926517571885,
      "grad_norm": 2.327202558517456,
      "learning_rate": 9.282051282051283e-06,
      "loss": 0.0213,
      "step": 2090
    },
    {
      "epoch": 26.722044728434504,
      "grad_norm": 2.2371463775634766,
      "learning_rate": 9.276923076923077e-06,
      "loss": 0.0355,
      "step": 2091
    },
    {
      "epoch": 26.73482428115016,
      "grad_norm": 3.8798370361328125,
      "learning_rate": 9.271794871794873e-06,
      "loss": 0.0299,
      "step": 2092
    },
    {
      "epoch": 26.747603833865814,
      "grad_norm": 2.8291265964508057,
      "learning_rate": 9.266666666666667e-06,
      "loss": 0.0333,
      "step": 2093
    },
    {
      "epoch": 26.76038338658147,
      "grad_norm": 2.7667057514190674,
      "learning_rate": 9.261538461538461e-06,
      "loss": 0.0348,
      "step": 2094
    },
    {
      "epoch": 26.773162939297123,
      "grad_norm": 3.423550844192505,
      "learning_rate": 9.256410256410257e-06,
      "loss": 0.0444,
      "step": 2095
    },
    {
      "epoch": 26.78594249201278,
      "grad_norm": 2.025509834289551,
      "learning_rate": 9.251282051282053e-06,
      "loss": 0.0254,
      "step": 2096
    },
    {
      "epoch": 26.798722044728436,
      "grad_norm": 3.826721668243408,
      "learning_rate": 9.246153846153847e-06,
      "loss": 0.0405,
      "step": 2097
    },
    {
      "epoch": 26.81150159744409,
      "grad_norm": 1.5340789556503296,
      "learning_rate": 9.24102564102564e-06,
      "loss": 0.0141,
      "step": 2098
    },
    {
      "epoch": 26.824281150159745,
      "grad_norm": 2.702284574508667,
      "learning_rate": 9.235897435897437e-06,
      "loss": 0.0344,
      "step": 2099
    },
    {
      "epoch": 26.837060702875398,
      "grad_norm": 3.0476367473602295,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.0385,
      "step": 2100
    },
    {
      "epoch": 26.837060702875398,
      "eval_loss": 0.8120326399803162,
      "eval_runtime": 183.4599,
      "eval_samples_per_second": 0.856,
      "eval_steps_per_second": 0.109,
      "step": 2100
    },
    {
      "epoch": 26.849840255591054,
      "grad_norm": 1.8112848997116089,
      "learning_rate": 9.225641025641026e-06,
      "loss": 0.0173,
      "step": 2101
    },
    {
      "epoch": 26.86261980830671,
      "grad_norm": 2.793485641479492,
      "learning_rate": 9.220512820512822e-06,
      "loss": 0.0228,
      "step": 2102
    },
    {
      "epoch": 26.875399361022364,
      "grad_norm": 3.2606515884399414,
      "learning_rate": 9.215384615384616e-06,
      "loss": 0.0425,
      "step": 2103
    },
    {
      "epoch": 26.88817891373802,
      "grad_norm": 3.649292469024658,
      "learning_rate": 9.21025641025641e-06,
      "loss": 0.0281,
      "step": 2104
    },
    {
      "epoch": 26.900958466453673,
      "grad_norm": 2.2313618659973145,
      "learning_rate": 9.205128205128206e-06,
      "loss": 0.0281,
      "step": 2105
    },
    {
      "epoch": 26.91373801916933,
      "grad_norm": 5.063486099243164,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.0466,
      "step": 2106
    },
    {
      "epoch": 26.926517571884983,
      "grad_norm": 3.2838034629821777,
      "learning_rate": 9.194871794871796e-06,
      "loss": 0.0365,
      "step": 2107
    },
    {
      "epoch": 26.93929712460064,
      "grad_norm": 3.712921142578125,
      "learning_rate": 9.18974358974359e-06,
      "loss": 0.0281,
      "step": 2108
    },
    {
      "epoch": 26.952076677316295,
      "grad_norm": 4.407195091247559,
      "learning_rate": 9.184615384615386e-06,
      "loss": 0.032,
      "step": 2109
    },
    {
      "epoch": 26.96485623003195,
      "grad_norm": 5.876495361328125,
      "learning_rate": 9.17948717948718e-06,
      "loss": 0.0655,
      "step": 2110
    },
    {
      "epoch": 26.977635782747605,
      "grad_norm": 1.9610528945922852,
      "learning_rate": 9.174358974358975e-06,
      "loss": 0.0204,
      "step": 2111
    },
    {
      "epoch": 26.990415335463258,
      "grad_norm": 2.3272671699523926,
      "learning_rate": 9.169230769230771e-06,
      "loss": 0.0247,
      "step": 2112
    },
    {
      "epoch": 27.003194888178914,
      "grad_norm": 2.827895402908325,
      "learning_rate": 9.164102564102565e-06,
      "loss": 0.0215,
      "step": 2113
    },
    {
      "epoch": 27.015974440894567,
      "grad_norm": 1.6048749685287476,
      "learning_rate": 9.15897435897436e-06,
      "loss": 0.0244,
      "step": 2114
    },
    {
      "epoch": 27.028753993610223,
      "grad_norm": 2.2859067916870117,
      "learning_rate": 9.153846153846155e-06,
      "loss": 0.0206,
      "step": 2115
    },
    {
      "epoch": 27.04153354632588,
      "grad_norm": 2.696887969970703,
      "learning_rate": 9.148717948717949e-06,
      "loss": 0.0251,
      "step": 2116
    },
    {
      "epoch": 27.054313099041533,
      "grad_norm": 1.4276155233383179,
      "learning_rate": 9.143589743589745e-06,
      "loss": 0.0207,
      "step": 2117
    },
    {
      "epoch": 27.06709265175719,
      "grad_norm": 5.58026123046875,
      "learning_rate": 9.138461538461539e-06,
      "loss": 0.0437,
      "step": 2118
    },
    {
      "epoch": 27.079872204472842,
      "grad_norm": 5.844724655151367,
      "learning_rate": 9.133333333333335e-06,
      "loss": 0.0436,
      "step": 2119
    },
    {
      "epoch": 27.0926517571885,
      "grad_norm": 8.150542259216309,
      "learning_rate": 9.128205128205129e-06,
      "loss": 0.0422,
      "step": 2120
    },
    {
      "epoch": 27.105431309904155,
      "grad_norm": 5.604848384857178,
      "learning_rate": 9.123076923076923e-06,
      "loss": 0.0382,
      "step": 2121
    },
    {
      "epoch": 27.118210862619808,
      "grad_norm": 3.6250290870666504,
      "learning_rate": 9.117948717948718e-06,
      "loss": 0.0324,
      "step": 2122
    },
    {
      "epoch": 27.130990415335464,
      "grad_norm": 5.809665679931641,
      "learning_rate": 9.112820512820514e-06,
      "loss": 0.0319,
      "step": 2123
    },
    {
      "epoch": 27.143769968051117,
      "grad_norm": 2.2189152240753174,
      "learning_rate": 9.107692307692308e-06,
      "loss": 0.0208,
      "step": 2124
    },
    {
      "epoch": 27.156549520766774,
      "grad_norm": 2.7135326862335205,
      "learning_rate": 9.102564102564104e-06,
      "loss": 0.0243,
      "step": 2125
    },
    {
      "epoch": 27.169329073482427,
      "grad_norm": 1.4363244771957397,
      "learning_rate": 9.097435897435898e-06,
      "loss": 0.014,
      "step": 2126
    },
    {
      "epoch": 27.182108626198083,
      "grad_norm": 4.288629055023193,
      "learning_rate": 9.092307692307692e-06,
      "loss": 0.0569,
      "step": 2127
    },
    {
      "epoch": 27.19488817891374,
      "grad_norm": 4.174808502197266,
      "learning_rate": 9.087179487179488e-06,
      "loss": 0.0383,
      "step": 2128
    },
    {
      "epoch": 27.207667731629392,
      "grad_norm": 2.7566633224487305,
      "learning_rate": 9.082051282051284e-06,
      "loss": 0.0336,
      "step": 2129
    },
    {
      "epoch": 27.22044728434505,
      "grad_norm": 1.8850542306900024,
      "learning_rate": 9.076923076923078e-06,
      "loss": 0.0168,
      "step": 2130
    },
    {
      "epoch": 27.233226837060702,
      "grad_norm": 1.9132078886032104,
      "learning_rate": 9.071794871794872e-06,
      "loss": 0.0155,
      "step": 2131
    },
    {
      "epoch": 27.24600638977636,
      "grad_norm": 19.71868324279785,
      "learning_rate": 9.066666666666667e-06,
      "loss": 0.0332,
      "step": 2132
    },
    {
      "epoch": 27.25878594249201,
      "grad_norm": 1.4613136053085327,
      "learning_rate": 9.061538461538463e-06,
      "loss": 0.0161,
      "step": 2133
    },
    {
      "epoch": 27.271565495207668,
      "grad_norm": 2.2136759757995605,
      "learning_rate": 9.056410256410257e-06,
      "loss": 0.0301,
      "step": 2134
    },
    {
      "epoch": 27.284345047923324,
      "grad_norm": 2.1000702381134033,
      "learning_rate": 9.051282051282051e-06,
      "loss": 0.032,
      "step": 2135
    },
    {
      "epoch": 27.297124600638977,
      "grad_norm": 2.5691099166870117,
      "learning_rate": 9.046153846153847e-06,
      "loss": 0.0538,
      "step": 2136
    },
    {
      "epoch": 27.309904153354633,
      "grad_norm": 1.641770839691162,
      "learning_rate": 9.041025641025641e-06,
      "loss": 0.0149,
      "step": 2137
    },
    {
      "epoch": 27.322683706070286,
      "grad_norm": 4.829441070556641,
      "learning_rate": 9.035897435897437e-06,
      "loss": 0.0441,
      "step": 2138
    },
    {
      "epoch": 27.335463258785943,
      "grad_norm": 2.505126476287842,
      "learning_rate": 9.030769230769233e-06,
      "loss": 0.0289,
      "step": 2139
    },
    {
      "epoch": 27.3482428115016,
      "grad_norm": 6.310408115386963,
      "learning_rate": 9.025641025641027e-06,
      "loss": 0.0431,
      "step": 2140
    },
    {
      "epoch": 27.361022364217252,
      "grad_norm": 2.2306759357452393,
      "learning_rate": 9.02051282051282e-06,
      "loss": 0.0153,
      "step": 2141
    },
    {
      "epoch": 27.37380191693291,
      "grad_norm": 1.9416497945785522,
      "learning_rate": 9.015384615384616e-06,
      "loss": 0.0199,
      "step": 2142
    },
    {
      "epoch": 27.38658146964856,
      "grad_norm": 4.289925575256348,
      "learning_rate": 9.01025641025641e-06,
      "loss": 0.0346,
      "step": 2143
    },
    {
      "epoch": 27.399361022364218,
      "grad_norm": 1.3997843265533447,
      "learning_rate": 9.005128205128206e-06,
      "loss": 0.0153,
      "step": 2144
    },
    {
      "epoch": 27.41214057507987,
      "grad_norm": 3.835322380065918,
      "learning_rate": 9e-06,
      "loss": 0.0603,
      "step": 2145
    },
    {
      "epoch": 27.424920127795527,
      "grad_norm": 2.464630365371704,
      "learning_rate": 8.994871794871796e-06,
      "loss": 0.0285,
      "step": 2146
    },
    {
      "epoch": 27.437699680511184,
      "grad_norm": 2.7458603382110596,
      "learning_rate": 8.98974358974359e-06,
      "loss": 0.0379,
      "step": 2147
    },
    {
      "epoch": 27.450479233226837,
      "grad_norm": 1.4156213998794556,
      "learning_rate": 8.984615384615386e-06,
      "loss": 0.0164,
      "step": 2148
    },
    {
      "epoch": 27.463258785942493,
      "grad_norm": 1.647200584411621,
      "learning_rate": 8.979487179487182e-06,
      "loss": 0.0216,
      "step": 2149
    },
    {
      "epoch": 27.476038338658146,
      "grad_norm": 2.2965891361236572,
      "learning_rate": 8.974358974358976e-06,
      "loss": 0.0231,
      "step": 2150
    },
    {
      "epoch": 27.488817891373802,
      "grad_norm": 3.635070562362671,
      "learning_rate": 8.96923076923077e-06,
      "loss": 0.0221,
      "step": 2151
    },
    {
      "epoch": 27.501597444089455,
      "grad_norm": 2.297700881958008,
      "learning_rate": 8.964102564102565e-06,
      "loss": 0.0231,
      "step": 2152
    },
    {
      "epoch": 27.51437699680511,
      "grad_norm": 1.8114467859268188,
      "learning_rate": 8.95897435897436e-06,
      "loss": 0.0185,
      "step": 2153
    },
    {
      "epoch": 27.527156549520768,
      "grad_norm": 4.470355987548828,
      "learning_rate": 8.953846153846153e-06,
      "loss": 0.0464,
      "step": 2154
    },
    {
      "epoch": 27.53993610223642,
      "grad_norm": 1.3283096551895142,
      "learning_rate": 8.94871794871795e-06,
      "loss": 0.0161,
      "step": 2155
    },
    {
      "epoch": 27.552715654952078,
      "grad_norm": 7.926822662353516,
      "learning_rate": 8.943589743589745e-06,
      "loss": 0.0662,
      "step": 2156
    },
    {
      "epoch": 27.56549520766773,
      "grad_norm": 2.1085832118988037,
      "learning_rate": 8.938461538461539e-06,
      "loss": 0.0244,
      "step": 2157
    },
    {
      "epoch": 27.578274760383387,
      "grad_norm": 7.27145528793335,
      "learning_rate": 8.933333333333333e-06,
      "loss": 0.0482,
      "step": 2158
    },
    {
      "epoch": 27.591054313099043,
      "grad_norm": 2.787456750869751,
      "learning_rate": 8.928205128205129e-06,
      "loss": 0.0379,
      "step": 2159
    },
    {
      "epoch": 27.603833865814696,
      "grad_norm": 2.238116502761841,
      "learning_rate": 8.923076923076925e-06,
      "loss": 0.024,
      "step": 2160
    },
    {
      "epoch": 27.616613418530353,
      "grad_norm": 5.366321086883545,
      "learning_rate": 8.917948717948719e-06,
      "loss": 0.0464,
      "step": 2161
    },
    {
      "epoch": 27.629392971246006,
      "grad_norm": 4.097918510437012,
      "learning_rate": 8.912820512820514e-06,
      "loss": 0.0262,
      "step": 2162
    },
    {
      "epoch": 27.642172523961662,
      "grad_norm": 4.417905807495117,
      "learning_rate": 8.907692307692308e-06,
      "loss": 0.043,
      "step": 2163
    },
    {
      "epoch": 27.654952076677315,
      "grad_norm": 2.8238251209259033,
      "learning_rate": 8.902564102564102e-06,
      "loss": 0.0376,
      "step": 2164
    },
    {
      "epoch": 27.66773162939297,
      "grad_norm": 1.3213348388671875,
      "learning_rate": 8.897435897435898e-06,
      "loss": 0.0146,
      "step": 2165
    },
    {
      "epoch": 27.680511182108628,
      "grad_norm": 3.444852828979492,
      "learning_rate": 8.892307692307694e-06,
      "loss": 0.0278,
      "step": 2166
    },
    {
      "epoch": 27.69329073482428,
      "grad_norm": 1.3029857873916626,
      "learning_rate": 8.887179487179488e-06,
      "loss": 0.0129,
      "step": 2167
    },
    {
      "epoch": 27.706070287539937,
      "grad_norm": 5.024180889129639,
      "learning_rate": 8.882051282051282e-06,
      "loss": 0.0363,
      "step": 2168
    },
    {
      "epoch": 27.71884984025559,
      "grad_norm": 1.4591314792633057,
      "learning_rate": 8.876923076923078e-06,
      "loss": 0.0145,
      "step": 2169
    },
    {
      "epoch": 27.731629392971247,
      "grad_norm": 2.011525869369507,
      "learning_rate": 8.871794871794872e-06,
      "loss": 0.0233,
      "step": 2170
    },
    {
      "epoch": 27.7444089456869,
      "grad_norm": 2.6536760330200195,
      "learning_rate": 8.866666666666668e-06,
      "loss": 0.0259,
      "step": 2171
    },
    {
      "epoch": 27.757188498402556,
      "grad_norm": 2.8977763652801514,
      "learning_rate": 8.861538461538463e-06,
      "loss": 0.0406,
      "step": 2172
    },
    {
      "epoch": 27.769968051118212,
      "grad_norm": 2.8562774658203125,
      "learning_rate": 8.856410256410257e-06,
      "loss": 0.0288,
      "step": 2173
    },
    {
      "epoch": 27.782747603833865,
      "grad_norm": 5.174256801605225,
      "learning_rate": 8.851282051282051e-06,
      "loss": 0.0222,
      "step": 2174
    },
    {
      "epoch": 27.79552715654952,
      "grad_norm": 1.915551781654358,
      "learning_rate": 8.846153846153847e-06,
      "loss": 0.035,
      "step": 2175
    },
    {
      "epoch": 27.808306709265175,
      "grad_norm": 3.622258424758911,
      "learning_rate": 8.841025641025641e-06,
      "loss": 0.0271,
      "step": 2176
    },
    {
      "epoch": 27.82108626198083,
      "grad_norm": 2.134631633758545,
      "learning_rate": 8.835897435897437e-06,
      "loss": 0.0273,
      "step": 2177
    },
    {
      "epoch": 27.833865814696484,
      "grad_norm": 7.068699359893799,
      "learning_rate": 8.830769230769231e-06,
      "loss": 0.0362,
      "step": 2178
    },
    {
      "epoch": 27.84664536741214,
      "grad_norm": 3.2578909397125244,
      "learning_rate": 8.825641025641027e-06,
      "loss": 0.0328,
      "step": 2179
    },
    {
      "epoch": 27.859424920127797,
      "grad_norm": 2.324989080429077,
      "learning_rate": 8.820512820512821e-06,
      "loss": 0.025,
      "step": 2180
    },
    {
      "epoch": 27.87220447284345,
      "grad_norm": 1.5827807188034058,
      "learning_rate": 8.815384615384615e-06,
      "loss": 0.0219,
      "step": 2181
    },
    {
      "epoch": 27.884984025559106,
      "grad_norm": 4.256125450134277,
      "learning_rate": 8.81025641025641e-06,
      "loss": 0.0267,
      "step": 2182
    },
    {
      "epoch": 27.89776357827476,
      "grad_norm": 3.0664892196655273,
      "learning_rate": 8.805128205128206e-06,
      "loss": 0.0263,
      "step": 2183
    },
    {
      "epoch": 27.910543130990416,
      "grad_norm": 4.755489826202393,
      "learning_rate": 8.8e-06,
      "loss": 0.029,
      "step": 2184
    },
    {
      "epoch": 27.923322683706072,
      "grad_norm": 3.9942352771759033,
      "learning_rate": 8.794871794871796e-06,
      "loss": 0.0378,
      "step": 2185
    },
    {
      "epoch": 27.936102236421725,
      "grad_norm": 1.4267417192459106,
      "learning_rate": 8.78974358974359e-06,
      "loss": 0.0142,
      "step": 2186
    },
    {
      "epoch": 27.94888178913738,
      "grad_norm": 3.233738899230957,
      "learning_rate": 8.784615384615386e-06,
      "loss": 0.0268,
      "step": 2187
    },
    {
      "epoch": 27.961661341853034,
      "grad_norm": 3.739074468612671,
      "learning_rate": 8.77948717948718e-06,
      "loss": 0.0438,
      "step": 2188
    },
    {
      "epoch": 27.97444089456869,
      "grad_norm": 3.132491111755371,
      "learning_rate": 8.774358974358976e-06,
      "loss": 0.0315,
      "step": 2189
    },
    {
      "epoch": 27.987220447284344,
      "grad_norm": 1.5290716886520386,
      "learning_rate": 8.76923076923077e-06,
      "loss": 0.0224,
      "step": 2190
    },
    {
      "epoch": 28.0,
      "grad_norm": 1.6961673498153687,
      "learning_rate": 8.764102564102564e-06,
      "loss": 0.0171,
      "step": 2191
    },
    {
      "epoch": 28.012779552715656,
      "grad_norm": 1.394661545753479,
      "learning_rate": 8.75897435897436e-06,
      "loss": 0.0148,
      "step": 2192
    },
    {
      "epoch": 28.02555910543131,
      "grad_norm": 1.8422940969467163,
      "learning_rate": 8.753846153846155e-06,
      "loss": 0.0188,
      "step": 2193
    },
    {
      "epoch": 28.038338658146966,
      "grad_norm": 1.0886930227279663,
      "learning_rate": 8.74871794871795e-06,
      "loss": 0.0112,
      "step": 2194
    },
    {
      "epoch": 28.05111821086262,
      "grad_norm": 1.127504587173462,
      "learning_rate": 8.743589743589743e-06,
      "loss": 0.0122,
      "step": 2195
    },
    {
      "epoch": 28.063897763578275,
      "grad_norm": 1.3174153566360474,
      "learning_rate": 8.73846153846154e-06,
      "loss": 0.0149,
      "step": 2196
    },
    {
      "epoch": 28.076677316293928,
      "grad_norm": 0.9179958701133728,
      "learning_rate": 8.733333333333333e-06,
      "loss": 0.0109,
      "step": 2197
    },
    {
      "epoch": 28.089456869009584,
      "grad_norm": 1.8915237188339233,
      "learning_rate": 8.728205128205129e-06,
      "loss": 0.0164,
      "step": 2198
    },
    {
      "epoch": 28.10223642172524,
      "grad_norm": 1.8076939582824707,
      "learning_rate": 8.723076923076925e-06,
      "loss": 0.0155,
      "step": 2199
    },
    {
      "epoch": 28.115015974440894,
      "grad_norm": 1.4523608684539795,
      "learning_rate": 8.717948717948719e-06,
      "loss": 0.0163,
      "step": 2200
    },
    {
      "epoch": 28.115015974440894,
      "eval_loss": 0.8285750150680542,
      "eval_runtime": 183.9379,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 2200
    },
    {
      "epoch": 28.12779552715655,
      "grad_norm": 3.155454635620117,
      "learning_rate": 8.712820512820513e-06,
      "loss": 0.0242,
      "step": 2201
    },
    {
      "epoch": 28.140575079872203,
      "grad_norm": 0.9504920244216919,
      "learning_rate": 8.707692307692309e-06,
      "loss": 0.0113,
      "step": 2202
    },
    {
      "epoch": 28.15335463258786,
      "grad_norm": 2.4801573753356934,
      "learning_rate": 8.702564102564103e-06,
      "loss": 0.0291,
      "step": 2203
    },
    {
      "epoch": 28.166134185303516,
      "grad_norm": 1.5551204681396484,
      "learning_rate": 8.697435897435898e-06,
      "loss": 0.0249,
      "step": 2204
    },
    {
      "epoch": 28.17891373801917,
      "grad_norm": 1.50502347946167,
      "learning_rate": 8.692307692307692e-06,
      "loss": 0.0129,
      "step": 2205
    },
    {
      "epoch": 28.191693290734825,
      "grad_norm": 3.653285503387451,
      "learning_rate": 8.687179487179488e-06,
      "loss": 0.0239,
      "step": 2206
    },
    {
      "epoch": 28.20447284345048,
      "grad_norm": 2.517505168914795,
      "learning_rate": 8.682051282051282e-06,
      "loss": 0.0401,
      "step": 2207
    },
    {
      "epoch": 28.217252396166135,
      "grad_norm": 3.3538625240325928,
      "learning_rate": 8.676923076923078e-06,
      "loss": 0.0457,
      "step": 2208
    },
    {
      "epoch": 28.230031948881788,
      "grad_norm": 1.3679107427597046,
      "learning_rate": 8.671794871794874e-06,
      "loss": 0.0149,
      "step": 2209
    },
    {
      "epoch": 28.242811501597444,
      "grad_norm": 5.592023849487305,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.0218,
      "step": 2210
    },
    {
      "epoch": 28.2555910543131,
      "grad_norm": 0.7429912090301514,
      "learning_rate": 8.661538461538462e-06,
      "loss": 0.0083,
      "step": 2211
    },
    {
      "epoch": 28.268370607028753,
      "grad_norm": 1.2271093130111694,
      "learning_rate": 8.656410256410258e-06,
      "loss": 0.0131,
      "step": 2212
    },
    {
      "epoch": 28.28115015974441,
      "grad_norm": 2.5322265625,
      "learning_rate": 8.651282051282052e-06,
      "loss": 0.0254,
      "step": 2213
    },
    {
      "epoch": 28.293929712460063,
      "grad_norm": 2.8084962368011475,
      "learning_rate": 8.646153846153846e-06,
      "loss": 0.0189,
      "step": 2214
    },
    {
      "epoch": 28.30670926517572,
      "grad_norm": 4.381146430969238,
      "learning_rate": 8.641025641025641e-06,
      "loss": 0.0203,
      "step": 2215
    },
    {
      "epoch": 28.319488817891372,
      "grad_norm": 3.287971258163452,
      "learning_rate": 8.635897435897437e-06,
      "loss": 0.0324,
      "step": 2216
    },
    {
      "epoch": 28.33226837060703,
      "grad_norm": 2.3260955810546875,
      "learning_rate": 8.630769230769231e-06,
      "loss": 0.0262,
      "step": 2217
    },
    {
      "epoch": 28.345047923322685,
      "grad_norm": 2.256308078765869,
      "learning_rate": 8.625641025641025e-06,
      "loss": 0.0339,
      "step": 2218
    },
    {
      "epoch": 28.357827476038338,
      "grad_norm": 9.611410140991211,
      "learning_rate": 8.620512820512821e-06,
      "loss": 0.0633,
      "step": 2219
    },
    {
      "epoch": 28.370607028753994,
      "grad_norm": 4.757236957550049,
      "learning_rate": 8.615384615384617e-06,
      "loss": 0.0282,
      "step": 2220
    },
    {
      "epoch": 28.383386581469647,
      "grad_norm": 2.538170337677002,
      "learning_rate": 8.610256410256411e-06,
      "loss": 0.0169,
      "step": 2221
    },
    {
      "epoch": 28.396166134185304,
      "grad_norm": 1.2487812042236328,
      "learning_rate": 8.605128205128207e-06,
      "loss": 0.0153,
      "step": 2222
    },
    {
      "epoch": 28.408945686900957,
      "grad_norm": 3.4212915897369385,
      "learning_rate": 8.6e-06,
      "loss": 0.0224,
      "step": 2223
    },
    {
      "epoch": 28.421725239616613,
      "grad_norm": 2.1509947776794434,
      "learning_rate": 8.594871794871795e-06,
      "loss": 0.0217,
      "step": 2224
    },
    {
      "epoch": 28.43450479233227,
      "grad_norm": 1.3148765563964844,
      "learning_rate": 8.58974358974359e-06,
      "loss": 0.0345,
      "step": 2225
    },
    {
      "epoch": 28.447284345047922,
      "grad_norm": 2.901209592819214,
      "learning_rate": 8.584615384615386e-06,
      "loss": 0.0248,
      "step": 2226
    },
    {
      "epoch": 28.46006389776358,
      "grad_norm": 2.8941667079925537,
      "learning_rate": 8.57948717948718e-06,
      "loss": 0.0244,
      "step": 2227
    },
    {
      "epoch": 28.472843450479232,
      "grad_norm": 5.521927356719971,
      "learning_rate": 8.574358974358974e-06,
      "loss": 0.0432,
      "step": 2228
    },
    {
      "epoch": 28.48562300319489,
      "grad_norm": 1.0991194248199463,
      "learning_rate": 8.56923076923077e-06,
      "loss": 0.0133,
      "step": 2229
    },
    {
      "epoch": 28.498402555910545,
      "grad_norm": 1.867282748222351,
      "learning_rate": 8.564102564102564e-06,
      "loss": 0.0163,
      "step": 2230
    },
    {
      "epoch": 28.511182108626198,
      "grad_norm": 2.533832311630249,
      "learning_rate": 8.55897435897436e-06,
      "loss": 0.03,
      "step": 2231
    },
    {
      "epoch": 28.523961661341854,
      "grad_norm": 2.8608005046844482,
      "learning_rate": 8.553846153846156e-06,
      "loss": 0.0364,
      "step": 2232
    },
    {
      "epoch": 28.536741214057507,
      "grad_norm": 1.3630461692810059,
      "learning_rate": 8.54871794871795e-06,
      "loss": 0.0164,
      "step": 2233
    },
    {
      "epoch": 28.549520766773163,
      "grad_norm": 1.4716297388076782,
      "learning_rate": 8.543589743589744e-06,
      "loss": 0.0189,
      "step": 2234
    },
    {
      "epoch": 28.562300319488816,
      "grad_norm": 3.877401113510132,
      "learning_rate": 8.53846153846154e-06,
      "loss": 0.0324,
      "step": 2235
    },
    {
      "epoch": 28.575079872204473,
      "grad_norm": 1.6375051736831665,
      "learning_rate": 8.533333333333335e-06,
      "loss": 0.0164,
      "step": 2236
    },
    {
      "epoch": 28.58785942492013,
      "grad_norm": 1.801792860031128,
      "learning_rate": 8.52820512820513e-06,
      "loss": 0.0188,
      "step": 2237
    },
    {
      "epoch": 28.600638977635782,
      "grad_norm": 3.3205134868621826,
      "learning_rate": 8.523076923076923e-06,
      "loss": 0.0266,
      "step": 2238
    },
    {
      "epoch": 28.61341853035144,
      "grad_norm": 1.5833179950714111,
      "learning_rate": 8.517948717948719e-06,
      "loss": 0.0161,
      "step": 2239
    },
    {
      "epoch": 28.62619808306709,
      "grad_norm": 2.784162998199463,
      "learning_rate": 8.512820512820513e-06,
      "loss": 0.0225,
      "step": 2240
    },
    {
      "epoch": 28.638977635782748,
      "grad_norm": 4.481935977935791,
      "learning_rate": 8.507692307692307e-06,
      "loss": 0.0506,
      "step": 2241
    },
    {
      "epoch": 28.6517571884984,
      "grad_norm": 1.981535792350769,
      "learning_rate": 8.502564102564103e-06,
      "loss": 0.0213,
      "step": 2242
    },
    {
      "epoch": 28.664536741214057,
      "grad_norm": 1.7869662046432495,
      "learning_rate": 8.497435897435899e-06,
      "loss": 0.028,
      "step": 2243
    },
    {
      "epoch": 28.677316293929714,
      "grad_norm": 2.6905384063720703,
      "learning_rate": 8.492307692307693e-06,
      "loss": 0.0259,
      "step": 2244
    },
    {
      "epoch": 28.690095846645367,
      "grad_norm": 4.806192874908447,
      "learning_rate": 8.487179487179488e-06,
      "loss": 0.0369,
      "step": 2245
    },
    {
      "epoch": 28.702875399361023,
      "grad_norm": 2.636756181716919,
      "learning_rate": 8.482051282051283e-06,
      "loss": 0.0254,
      "step": 2246
    },
    {
      "epoch": 28.715654952076676,
      "grad_norm": 2.5176260471343994,
      "learning_rate": 8.476923076923078e-06,
      "loss": 0.0205,
      "step": 2247
    },
    {
      "epoch": 28.728434504792332,
      "grad_norm": 3.5982775688171387,
      "learning_rate": 8.471794871794872e-06,
      "loss": 0.0317,
      "step": 2248
    },
    {
      "epoch": 28.74121405750799,
      "grad_norm": 2.210441827774048,
      "learning_rate": 8.466666666666668e-06,
      "loss": 0.0197,
      "step": 2249
    },
    {
      "epoch": 28.75399361022364,
      "grad_norm": 1.3656779527664185,
      "learning_rate": 8.461538461538462e-06,
      "loss": 0.0125,
      "step": 2250
    },
    {
      "epoch": 28.766773162939298,
      "grad_norm": 1.8053436279296875,
      "learning_rate": 8.456410256410256e-06,
      "loss": 0.0121,
      "step": 2251
    },
    {
      "epoch": 28.77955271565495,
      "grad_norm": 2.3068583011627197,
      "learning_rate": 8.451282051282052e-06,
      "loss": 0.017,
      "step": 2252
    },
    {
      "epoch": 28.792332268370608,
      "grad_norm": 3.386035442352295,
      "learning_rate": 8.446153846153848e-06,
      "loss": 0.0307,
      "step": 2253
    },
    {
      "epoch": 28.80511182108626,
      "grad_norm": 1.64307701587677,
      "learning_rate": 8.441025641025642e-06,
      "loss": 0.017,
      "step": 2254
    },
    {
      "epoch": 28.817891373801917,
      "grad_norm": 3.6835851669311523,
      "learning_rate": 8.435897435897436e-06,
      "loss": 0.0422,
      "step": 2255
    },
    {
      "epoch": 28.830670926517573,
      "grad_norm": 5.623953819274902,
      "learning_rate": 8.430769230769231e-06,
      "loss": 0.0452,
      "step": 2256
    },
    {
      "epoch": 28.843450479233226,
      "grad_norm": 2.691822052001953,
      "learning_rate": 8.425641025641026e-06,
      "loss": 0.017,
      "step": 2257
    },
    {
      "epoch": 28.856230031948883,
      "grad_norm": 3.718013286590576,
      "learning_rate": 8.420512820512821e-06,
      "loss": 0.028,
      "step": 2258
    },
    {
      "epoch": 28.869009584664536,
      "grad_norm": 14.377763748168945,
      "learning_rate": 8.415384615384617e-06,
      "loss": 0.0477,
      "step": 2259
    },
    {
      "epoch": 28.881789137380192,
      "grad_norm": 2.0842843055725098,
      "learning_rate": 8.410256410256411e-06,
      "loss": 0.0174,
      "step": 2260
    },
    {
      "epoch": 28.894568690095845,
      "grad_norm": 6.70881986618042,
      "learning_rate": 8.405128205128205e-06,
      "loss": 0.0365,
      "step": 2261
    },
    {
      "epoch": 28.9073482428115,
      "grad_norm": 2.3938944339752197,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.0291,
      "step": 2262
    },
    {
      "epoch": 28.920127795527158,
      "grad_norm": 10.553678512573242,
      "learning_rate": 8.394871794871795e-06,
      "loss": 0.0226,
      "step": 2263
    },
    {
      "epoch": 28.93290734824281,
      "grad_norm": 2.9902539253234863,
      "learning_rate": 8.38974358974359e-06,
      "loss": 0.0339,
      "step": 2264
    },
    {
      "epoch": 28.945686900958467,
      "grad_norm": 1.8003623485565186,
      "learning_rate": 8.384615384615385e-06,
      "loss": 0.0236,
      "step": 2265
    },
    {
      "epoch": 28.95846645367412,
      "grad_norm": 3.184990882873535,
      "learning_rate": 8.37948717948718e-06,
      "loss": 0.0317,
      "step": 2266
    },
    {
      "epoch": 28.971246006389777,
      "grad_norm": 2.8180389404296875,
      "learning_rate": 8.374358974358975e-06,
      "loss": 0.0206,
      "step": 2267
    },
    {
      "epoch": 28.984025559105433,
      "grad_norm": 2.6118054389953613,
      "learning_rate": 8.36923076923077e-06,
      "loss": 0.0248,
      "step": 2268
    },
    {
      "epoch": 28.996805111821086,
      "grad_norm": 4.395126819610596,
      "learning_rate": 8.364102564102566e-06,
      "loss": 0.0305,
      "step": 2269
    },
    {
      "epoch": 29.009584664536742,
      "grad_norm": 1.494640588760376,
      "learning_rate": 8.35897435897436e-06,
      "loss": 0.0128,
      "step": 2270
    },
    {
      "epoch": 29.022364217252395,
      "grad_norm": 2.212526321411133,
      "learning_rate": 8.353846153846154e-06,
      "loss": 0.0347,
      "step": 2271
    },
    {
      "epoch": 29.03514376996805,
      "grad_norm": 1.449722409248352,
      "learning_rate": 8.34871794871795e-06,
      "loss": 0.0134,
      "step": 2272
    },
    {
      "epoch": 29.047923322683705,
      "grad_norm": 1.4646117687225342,
      "learning_rate": 8.343589743589744e-06,
      "loss": 0.0191,
      "step": 2273
    },
    {
      "epoch": 29.06070287539936,
      "grad_norm": 2.184894561767578,
      "learning_rate": 8.338461538461538e-06,
      "loss": 0.0188,
      "step": 2274
    },
    {
      "epoch": 29.073482428115017,
      "grad_norm": 0.5592262148857117,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0063,
      "step": 2275
    },
    {
      "epoch": 29.08626198083067,
      "grad_norm": 1.5937706232070923,
      "learning_rate": 8.32820512820513e-06,
      "loss": 0.0164,
      "step": 2276
    },
    {
      "epoch": 29.099041533546327,
      "grad_norm": 0.9277424812316895,
      "learning_rate": 8.323076923076924e-06,
      "loss": 0.0111,
      "step": 2277
    },
    {
      "epoch": 29.11182108626198,
      "grad_norm": 2.7014410495758057,
      "learning_rate": 8.317948717948718e-06,
      "loss": 0.0151,
      "step": 2278
    },
    {
      "epoch": 29.124600638977636,
      "grad_norm": 3.2854347229003906,
      "learning_rate": 8.312820512820513e-06,
      "loss": 0.0126,
      "step": 2279
    },
    {
      "epoch": 29.13738019169329,
      "grad_norm": 1.0488464832305908,
      "learning_rate": 8.307692307692309e-06,
      "loss": 0.0116,
      "step": 2280
    },
    {
      "epoch": 29.150159744408946,
      "grad_norm": 1.374603509902954,
      "learning_rate": 8.302564102564103e-06,
      "loss": 0.0144,
      "step": 2281
    },
    {
      "epoch": 29.162939297124602,
      "grad_norm": 0.8280434608459473,
      "learning_rate": 8.297435897435899e-06,
      "loss": 0.0081,
      "step": 2282
    },
    {
      "epoch": 29.175718849840255,
      "grad_norm": 1.7680490016937256,
      "learning_rate": 8.292307692307693e-06,
      "loss": 0.0194,
      "step": 2283
    },
    {
      "epoch": 29.18849840255591,
      "grad_norm": 1.6616615056991577,
      "learning_rate": 8.287179487179487e-06,
      "loss": 0.0153,
      "step": 2284
    },
    {
      "epoch": 29.201277955271564,
      "grad_norm": 1.9223436117172241,
      "learning_rate": 8.282051282051283e-06,
      "loss": 0.0189,
      "step": 2285
    },
    {
      "epoch": 29.21405750798722,
      "grad_norm": 1.9705466032028198,
      "learning_rate": 8.276923076923078e-06,
      "loss": 0.0138,
      "step": 2286
    },
    {
      "epoch": 29.226837060702877,
      "grad_norm": 1.6173869371414185,
      "learning_rate": 8.271794871794873e-06,
      "loss": 0.0175,
      "step": 2287
    },
    {
      "epoch": 29.23961661341853,
      "grad_norm": 1.4238561391830444,
      "learning_rate": 8.266666666666667e-06,
      "loss": 0.0154,
      "step": 2288
    },
    {
      "epoch": 29.252396166134186,
      "grad_norm": 9.690901756286621,
      "learning_rate": 8.261538461538462e-06,
      "loss": 0.0228,
      "step": 2289
    },
    {
      "epoch": 29.26517571884984,
      "grad_norm": 1.6259404420852661,
      "learning_rate": 8.256410256410256e-06,
      "loss": 0.0138,
      "step": 2290
    },
    {
      "epoch": 29.277955271565496,
      "grad_norm": 2.3683691024780273,
      "learning_rate": 8.251282051282052e-06,
      "loss": 0.0145,
      "step": 2291
    },
    {
      "epoch": 29.29073482428115,
      "grad_norm": 2.1553542613983154,
      "learning_rate": 8.246153846153848e-06,
      "loss": 0.0177,
      "step": 2292
    },
    {
      "epoch": 29.303514376996805,
      "grad_norm": 2.189448833465576,
      "learning_rate": 8.241025641025642e-06,
      "loss": 0.0196,
      "step": 2293
    },
    {
      "epoch": 29.31629392971246,
      "grad_norm": 1.2564376592636108,
      "learning_rate": 8.235897435897436e-06,
      "loss": 0.0138,
      "step": 2294
    },
    {
      "epoch": 29.329073482428115,
      "grad_norm": 2.2801411151885986,
      "learning_rate": 8.230769230769232e-06,
      "loss": 0.0215,
      "step": 2295
    },
    {
      "epoch": 29.34185303514377,
      "grad_norm": 1.3593823909759521,
      "learning_rate": 8.225641025641027e-06,
      "loss": 0.0139,
      "step": 2296
    },
    {
      "epoch": 29.354632587859424,
      "grad_norm": 1.1615716218948364,
      "learning_rate": 8.220512820512822e-06,
      "loss": 0.0124,
      "step": 2297
    },
    {
      "epoch": 29.36741214057508,
      "grad_norm": 4.084254264831543,
      "learning_rate": 8.215384615384616e-06,
      "loss": 0.0576,
      "step": 2298
    },
    {
      "epoch": 29.380191693290733,
      "grad_norm": 2.3387715816497803,
      "learning_rate": 8.210256410256411e-06,
      "loss": 0.0231,
      "step": 2299
    },
    {
      "epoch": 29.39297124600639,
      "grad_norm": 1.4509865045547485,
      "learning_rate": 8.205128205128205e-06,
      "loss": 0.0177,
      "step": 2300
    },
    {
      "epoch": 29.39297124600639,
      "eval_loss": 0.8588030338287354,
      "eval_runtime": 183.6274,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 2300
    },
    {
      "epoch": 29.405750798722046,
      "grad_norm": 0.7722755074501038,
      "learning_rate": 8.2e-06,
      "loss": 0.0082,
      "step": 2301
    },
    {
      "epoch": 29.4185303514377,
      "grad_norm": 2.2694647312164307,
      "learning_rate": 8.194871794871795e-06,
      "loss": 0.0214,
      "step": 2302
    },
    {
      "epoch": 29.431309904153355,
      "grad_norm": 3.7953007221221924,
      "learning_rate": 8.189743589743591e-06,
      "loss": 0.0396,
      "step": 2303
    },
    {
      "epoch": 29.44408945686901,
      "grad_norm": 1.1408644914627075,
      "learning_rate": 8.184615384615385e-06,
      "loss": 0.0151,
      "step": 2304
    },
    {
      "epoch": 29.456869009584665,
      "grad_norm": 1.9948118925094604,
      "learning_rate": 8.17948717948718e-06,
      "loss": 0.0157,
      "step": 2305
    },
    {
      "epoch": 29.46964856230032,
      "grad_norm": 0.918438732624054,
      "learning_rate": 8.174358974358975e-06,
      "loss": 0.0101,
      "step": 2306
    },
    {
      "epoch": 29.482428115015974,
      "grad_norm": 1.8009915351867676,
      "learning_rate": 8.16923076923077e-06,
      "loss": 0.0173,
      "step": 2307
    },
    {
      "epoch": 29.49520766773163,
      "grad_norm": 1.3889222145080566,
      "learning_rate": 8.164102564102565e-06,
      "loss": 0.0131,
      "step": 2308
    },
    {
      "epoch": 29.507987220447284,
      "grad_norm": 1.0000081062316895,
      "learning_rate": 8.15897435897436e-06,
      "loss": 0.0117,
      "step": 2309
    },
    {
      "epoch": 29.52076677316294,
      "grad_norm": 3.719639778137207,
      "learning_rate": 8.153846153846154e-06,
      "loss": 0.0305,
      "step": 2310
    },
    {
      "epoch": 29.533546325878593,
      "grad_norm": 4.300498962402344,
      "learning_rate": 8.148717948717948e-06,
      "loss": 0.0467,
      "step": 2311
    },
    {
      "epoch": 29.54632587859425,
      "grad_norm": 3.517829418182373,
      "learning_rate": 8.143589743589744e-06,
      "loss": 0.0433,
      "step": 2312
    },
    {
      "epoch": 29.559105431309906,
      "grad_norm": 1.670379877090454,
      "learning_rate": 8.13846153846154e-06,
      "loss": 0.0203,
      "step": 2313
    },
    {
      "epoch": 29.57188498402556,
      "grad_norm": 2.073993682861328,
      "learning_rate": 8.133333333333334e-06,
      "loss": 0.0222,
      "step": 2314
    },
    {
      "epoch": 29.584664536741215,
      "grad_norm": 2.1130599975585938,
      "learning_rate": 8.12820512820513e-06,
      "loss": 0.0234,
      "step": 2315
    },
    {
      "epoch": 29.597444089456868,
      "grad_norm": 1.0086466073989868,
      "learning_rate": 8.123076923076924e-06,
      "loss": 0.012,
      "step": 2316
    },
    {
      "epoch": 29.610223642172524,
      "grad_norm": 2.0270586013793945,
      "learning_rate": 8.117948717948718e-06,
      "loss": 0.0137,
      "step": 2317
    },
    {
      "epoch": 29.623003194888177,
      "grad_norm": 1.058440089225769,
      "learning_rate": 8.112820512820514e-06,
      "loss": 0.0104,
      "step": 2318
    },
    {
      "epoch": 29.635782747603834,
      "grad_norm": 2.1392710208892822,
      "learning_rate": 8.10769230769231e-06,
      "loss": 0.0136,
      "step": 2319
    },
    {
      "epoch": 29.64856230031949,
      "grad_norm": 2.7256596088409424,
      "learning_rate": 8.102564102564103e-06,
      "loss": 0.0161,
      "step": 2320
    },
    {
      "epoch": 29.661341853035143,
      "grad_norm": 3.502432107925415,
      "learning_rate": 8.097435897435897e-06,
      "loss": 0.0237,
      "step": 2321
    },
    {
      "epoch": 29.6741214057508,
      "grad_norm": 4.068586826324463,
      "learning_rate": 8.092307692307693e-06,
      "loss": 0.0366,
      "step": 2322
    },
    {
      "epoch": 29.686900958466452,
      "grad_norm": 2.0909790992736816,
      "learning_rate": 8.087179487179487e-06,
      "loss": 0.0148,
      "step": 2323
    },
    {
      "epoch": 29.69968051118211,
      "grad_norm": 1.4317904710769653,
      "learning_rate": 8.082051282051283e-06,
      "loss": 0.0106,
      "step": 2324
    },
    {
      "epoch": 29.712460063897765,
      "grad_norm": 4.125987529754639,
      "learning_rate": 8.076923076923077e-06,
      "loss": 0.0261,
      "step": 2325
    },
    {
      "epoch": 29.72523961661342,
      "grad_norm": 3.255063533782959,
      "learning_rate": 8.071794871794873e-06,
      "loss": 0.0176,
      "step": 2326
    },
    {
      "epoch": 29.738019169329075,
      "grad_norm": 0.9352941513061523,
      "learning_rate": 8.066666666666667e-06,
      "loss": 0.0095,
      "step": 2327
    },
    {
      "epoch": 29.750798722044728,
      "grad_norm": 2.9449105262756348,
      "learning_rate": 8.061538461538463e-06,
      "loss": 0.0155,
      "step": 2328
    },
    {
      "epoch": 29.763578274760384,
      "grad_norm": 2.225759506225586,
      "learning_rate": 8.056410256410258e-06,
      "loss": 0.0124,
      "step": 2329
    },
    {
      "epoch": 29.776357827476037,
      "grad_norm": 2.320310115814209,
      "learning_rate": 8.051282051282052e-06,
      "loss": 0.0177,
      "step": 2330
    },
    {
      "epoch": 29.789137380191693,
      "grad_norm": 1.8711694478988647,
      "learning_rate": 8.046153846153846e-06,
      "loss": 0.0236,
      "step": 2331
    },
    {
      "epoch": 29.80191693290735,
      "grad_norm": 1.8947467803955078,
      "learning_rate": 8.041025641025642e-06,
      "loss": 0.0123,
      "step": 2332
    },
    {
      "epoch": 29.814696485623003,
      "grad_norm": 1.9755359888076782,
      "learning_rate": 8.035897435897436e-06,
      "loss": 0.0133,
      "step": 2333
    },
    {
      "epoch": 29.82747603833866,
      "grad_norm": 2.1724817752838135,
      "learning_rate": 8.03076923076923e-06,
      "loss": 0.0189,
      "step": 2334
    },
    {
      "epoch": 29.840255591054312,
      "grad_norm": 2.910400629043579,
      "learning_rate": 8.025641025641026e-06,
      "loss": 0.0262,
      "step": 2335
    },
    {
      "epoch": 29.85303514376997,
      "grad_norm": 2.85970139503479,
      "learning_rate": 8.020512820512822e-06,
      "loss": 0.0272,
      "step": 2336
    },
    {
      "epoch": 29.86581469648562,
      "grad_norm": 3.4145076274871826,
      "learning_rate": 8.015384615384616e-06,
      "loss": 0.0154,
      "step": 2337
    },
    {
      "epoch": 29.878594249201278,
      "grad_norm": 2.902740240097046,
      "learning_rate": 8.01025641025641e-06,
      "loss": 0.0183,
      "step": 2338
    },
    {
      "epoch": 29.891373801916934,
      "grad_norm": 3.356585741043091,
      "learning_rate": 8.005128205128206e-06,
      "loss": 0.0363,
      "step": 2339
    },
    {
      "epoch": 29.904153354632587,
      "grad_norm": 50.76426315307617,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0385,
      "step": 2340
    },
    {
      "epoch": 29.916932907348244,
      "grad_norm": 3.5773375034332275,
      "learning_rate": 7.994871794871795e-06,
      "loss": 0.0267,
      "step": 2341
    },
    {
      "epoch": 29.929712460063897,
      "grad_norm": 3.0727083683013916,
      "learning_rate": 7.989743589743591e-06,
      "loss": 0.0273,
      "step": 2342
    },
    {
      "epoch": 29.942492012779553,
      "grad_norm": 3.3302500247955322,
      "learning_rate": 7.984615384615385e-06,
      "loss": 0.017,
      "step": 2343
    },
    {
      "epoch": 29.955271565495206,
      "grad_norm": 1.594466209411621,
      "learning_rate": 7.97948717948718e-06,
      "loss": 0.0186,
      "step": 2344
    },
    {
      "epoch": 29.968051118210862,
      "grad_norm": 1.5575155019760132,
      "learning_rate": 7.974358974358975e-06,
      "loss": 0.0271,
      "step": 2345
    },
    {
      "epoch": 29.98083067092652,
      "grad_norm": 1.5307331085205078,
      "learning_rate": 7.96923076923077e-06,
      "loss": 0.0144,
      "step": 2346
    },
    {
      "epoch": 29.99361022364217,
      "grad_norm": 2.816906690597534,
      "learning_rate": 7.964102564102565e-06,
      "loss": 0.0155,
      "step": 2347
    },
    {
      "epoch": 30.00638977635783,
      "grad_norm": 1.9447122812271118,
      "learning_rate": 7.958974358974359e-06,
      "loss": 0.0191,
      "step": 2348
    },
    {
      "epoch": 30.01916932907348,
      "grad_norm": 1.1538459062576294,
      "learning_rate": 7.953846153846155e-06,
      "loss": 0.0111,
      "step": 2349
    },
    {
      "epoch": 30.031948881789138,
      "grad_norm": 1.9048233032226562,
      "learning_rate": 7.948717948717949e-06,
      "loss": 0.0108,
      "step": 2350
    },
    {
      "epoch": 30.044728434504794,
      "grad_norm": 0.9791685938835144,
      "learning_rate": 7.943589743589744e-06,
      "loss": 0.0123,
      "step": 2351
    },
    {
      "epoch": 30.057507987220447,
      "grad_norm": 1.4688340425491333,
      "learning_rate": 7.93846153846154e-06,
      "loss": 0.0099,
      "step": 2352
    },
    {
      "epoch": 30.070287539936103,
      "grad_norm": 11.503856658935547,
      "learning_rate": 7.933333333333334e-06,
      "loss": 0.0124,
      "step": 2353
    },
    {
      "epoch": 30.083067092651756,
      "grad_norm": 1.4641727209091187,
      "learning_rate": 7.928205128205128e-06,
      "loss": 0.0087,
      "step": 2354
    },
    {
      "epoch": 30.095846645367413,
      "grad_norm": 1.5176548957824707,
      "learning_rate": 7.923076923076924e-06,
      "loss": 0.0133,
      "step": 2355
    },
    {
      "epoch": 30.108626198083066,
      "grad_norm": 1.0059568881988525,
      "learning_rate": 7.91794871794872e-06,
      "loss": 0.0082,
      "step": 2356
    },
    {
      "epoch": 30.121405750798722,
      "grad_norm": 1.561390995979309,
      "learning_rate": 7.912820512820514e-06,
      "loss": 0.0223,
      "step": 2357
    },
    {
      "epoch": 30.13418530351438,
      "grad_norm": 0.5975431799888611,
      "learning_rate": 7.907692307692308e-06,
      "loss": 0.0055,
      "step": 2358
    },
    {
      "epoch": 30.14696485623003,
      "grad_norm": 0.4060616195201874,
      "learning_rate": 7.902564102564104e-06,
      "loss": 0.0044,
      "step": 2359
    },
    {
      "epoch": 30.159744408945688,
      "grad_norm": 3.557813882827759,
      "learning_rate": 7.897435897435898e-06,
      "loss": 0.0321,
      "step": 2360
    },
    {
      "epoch": 30.17252396166134,
      "grad_norm": 0.8282617926597595,
      "learning_rate": 7.892307692307692e-06,
      "loss": 0.0068,
      "step": 2361
    },
    {
      "epoch": 30.185303514376997,
      "grad_norm": 2.1953506469726562,
      "learning_rate": 7.887179487179487e-06,
      "loss": 0.0209,
      "step": 2362
    },
    {
      "epoch": 30.19808306709265,
      "grad_norm": 0.8145207166671753,
      "learning_rate": 7.882051282051283e-06,
      "loss": 0.0072,
      "step": 2363
    },
    {
      "epoch": 30.210862619808307,
      "grad_norm": 2.264665365219116,
      "learning_rate": 7.876923076923077e-06,
      "loss": 0.0113,
      "step": 2364
    },
    {
      "epoch": 30.223642172523963,
      "grad_norm": 1.1967779397964478,
      "learning_rate": 7.871794871794873e-06,
      "loss": 0.0112,
      "step": 2365
    },
    {
      "epoch": 30.236421725239616,
      "grad_norm": 2.0275306701660156,
      "learning_rate": 7.866666666666667e-06,
      "loss": 0.0176,
      "step": 2366
    },
    {
      "epoch": 30.249201277955272,
      "grad_norm": 0.7311588525772095,
      "learning_rate": 7.861538461538463e-06,
      "loss": 0.0084,
      "step": 2367
    },
    {
      "epoch": 30.261980830670925,
      "grad_norm": 7.656560897827148,
      "learning_rate": 7.856410256410257e-06,
      "loss": 0.0419,
      "step": 2368
    },
    {
      "epoch": 30.27476038338658,
      "grad_norm": 2.2187883853912354,
      "learning_rate": 7.851282051282053e-06,
      "loss": 0.0166,
      "step": 2369
    },
    {
      "epoch": 30.287539936102238,
      "grad_norm": 1.6911665201187134,
      "learning_rate": 7.846153846153847e-06,
      "loss": 0.0312,
      "step": 2370
    },
    {
      "epoch": 30.30031948881789,
      "grad_norm": 1.257012963294983,
      "learning_rate": 7.84102564102564e-06,
      "loss": 0.0142,
      "step": 2371
    },
    {
      "epoch": 30.313099041533548,
      "grad_norm": 0.9386212825775146,
      "learning_rate": 7.835897435897436e-06,
      "loss": 0.0094,
      "step": 2372
    },
    {
      "epoch": 30.3258785942492,
      "grad_norm": 1.295942783355713,
      "learning_rate": 7.830769230769232e-06,
      "loss": 0.0094,
      "step": 2373
    },
    {
      "epoch": 30.338658146964857,
      "grad_norm": 5.017524242401123,
      "learning_rate": 7.825641025641026e-06,
      "loss": 0.0184,
      "step": 2374
    },
    {
      "epoch": 30.35143769968051,
      "grad_norm": 1.9900352954864502,
      "learning_rate": 7.820512820512822e-06,
      "loss": 0.0169,
      "step": 2375
    },
    {
      "epoch": 30.364217252396166,
      "grad_norm": 1.0820534229278564,
      "learning_rate": 7.815384615384616e-06,
      "loss": 0.0136,
      "step": 2376
    },
    {
      "epoch": 30.376996805111823,
      "grad_norm": 4.093630313873291,
      "learning_rate": 7.81025641025641e-06,
      "loss": 0.041,
      "step": 2377
    },
    {
      "epoch": 30.389776357827476,
      "grad_norm": 1.1857229471206665,
      "learning_rate": 7.805128205128206e-06,
      "loss": 0.0118,
      "step": 2378
    },
    {
      "epoch": 30.402555910543132,
      "grad_norm": 3.726752758026123,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.0137,
      "step": 2379
    },
    {
      "epoch": 30.415335463258785,
      "grad_norm": 1.2537568807601929,
      "learning_rate": 7.794871794871796e-06,
      "loss": 0.0088,
      "step": 2380
    },
    {
      "epoch": 30.42811501597444,
      "grad_norm": 1.6540377140045166,
      "learning_rate": 7.78974358974359e-06,
      "loss": 0.0074,
      "step": 2381
    },
    {
      "epoch": 30.440894568690094,
      "grad_norm": 0.9351989030838013,
      "learning_rate": 7.784615384615385e-06,
      "loss": 0.0101,
      "step": 2382
    },
    {
      "epoch": 30.45367412140575,
      "grad_norm": 1.9741319417953491,
      "learning_rate": 7.77948717948718e-06,
      "loss": 0.016,
      "step": 2383
    },
    {
      "epoch": 30.466453674121407,
      "grad_norm": 0.8153372406959534,
      "learning_rate": 7.774358974358975e-06,
      "loss": 0.0079,
      "step": 2384
    },
    {
      "epoch": 30.47923322683706,
      "grad_norm": 2.5167553424835205,
      "learning_rate": 7.76923076923077e-06,
      "loss": 0.0145,
      "step": 2385
    },
    {
      "epoch": 30.492012779552716,
      "grad_norm": 0.9691523313522339,
      "learning_rate": 7.764102564102565e-06,
      "loss": 0.0089,
      "step": 2386
    },
    {
      "epoch": 30.50479233226837,
      "grad_norm": 2.4006729125976562,
      "learning_rate": 7.758974358974359e-06,
      "loss": 0.0301,
      "step": 2387
    },
    {
      "epoch": 30.517571884984026,
      "grad_norm": 1.315584659576416,
      "learning_rate": 7.753846153846155e-06,
      "loss": 0.0139,
      "step": 2388
    },
    {
      "epoch": 30.53035143769968,
      "grad_norm": 0.833059549331665,
      "learning_rate": 7.74871794871795e-06,
      "loss": 0.0084,
      "step": 2389
    },
    {
      "epoch": 30.543130990415335,
      "grad_norm": 0.7655555605888367,
      "learning_rate": 7.743589743589745e-06,
      "loss": 0.0075,
      "step": 2390
    },
    {
      "epoch": 30.55591054313099,
      "grad_norm": 2.363828659057617,
      "learning_rate": 7.738461538461539e-06,
      "loss": 0.0162,
      "step": 2391
    },
    {
      "epoch": 30.568690095846645,
      "grad_norm": 1.7295135259628296,
      "learning_rate": 7.733333333333334e-06,
      "loss": 0.0133,
      "step": 2392
    },
    {
      "epoch": 30.5814696485623,
      "grad_norm": 0.7795832753181458,
      "learning_rate": 7.728205128205128e-06,
      "loss": 0.0079,
      "step": 2393
    },
    {
      "epoch": 30.594249201277954,
      "grad_norm": 4.533823013305664,
      "learning_rate": 7.723076923076924e-06,
      "loss": 0.06,
      "step": 2394
    },
    {
      "epoch": 30.60702875399361,
      "grad_norm": 1.1290048360824585,
      "learning_rate": 7.717948717948718e-06,
      "loss": 0.0083,
      "step": 2395
    },
    {
      "epoch": 30.619808306709267,
      "grad_norm": 1.0387988090515137,
      "learning_rate": 7.712820512820514e-06,
      "loss": 0.0126,
      "step": 2396
    },
    {
      "epoch": 30.63258785942492,
      "grad_norm": 1.8188968896865845,
      "learning_rate": 7.707692307692308e-06,
      "loss": 0.0141,
      "step": 2397
    },
    {
      "epoch": 30.645367412140576,
      "grad_norm": 1.7080775499343872,
      "learning_rate": 7.702564102564102e-06,
      "loss": 0.0122,
      "step": 2398
    },
    {
      "epoch": 30.65814696485623,
      "grad_norm": 3.921550750732422,
      "learning_rate": 7.697435897435898e-06,
      "loss": 0.0255,
      "step": 2399
    },
    {
      "epoch": 30.670926517571885,
      "grad_norm": 1.774515986442566,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.0156,
      "step": 2400
    },
    {
      "epoch": 30.670926517571885,
      "eval_loss": 0.8820593953132629,
      "eval_runtime": 183.7504,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 2400
    },
    {
      "epoch": 30.68370607028754,
      "grad_norm": 1.382828950881958,
      "learning_rate": 7.687179487179488e-06,
      "loss": 0.0152,
      "step": 2401
    },
    {
      "epoch": 30.696485623003195,
      "grad_norm": 1.670392394065857,
      "learning_rate": 7.682051282051283e-06,
      "loss": 0.0134,
      "step": 2402
    },
    {
      "epoch": 30.70926517571885,
      "grad_norm": 2.938199520111084,
      "learning_rate": 7.676923076923077e-06,
      "loss": 0.028,
      "step": 2403
    },
    {
      "epoch": 30.722044728434504,
      "grad_norm": 1.7198128700256348,
      "learning_rate": 7.671794871794871e-06,
      "loss": 0.0116,
      "step": 2404
    },
    {
      "epoch": 30.73482428115016,
      "grad_norm": 1.027203917503357,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.0083,
      "step": 2405
    },
    {
      "epoch": 30.747603833865814,
      "grad_norm": 1.1331881284713745,
      "learning_rate": 7.661538461538463e-06,
      "loss": 0.0108,
      "step": 2406
    },
    {
      "epoch": 30.76038338658147,
      "grad_norm": 2.0122077465057373,
      "learning_rate": 7.656410256410257e-06,
      "loss": 0.0151,
      "step": 2407
    },
    {
      "epoch": 30.773162939297123,
      "grad_norm": 1.0752384662628174,
      "learning_rate": 7.651282051282051e-06,
      "loss": 0.0087,
      "step": 2408
    },
    {
      "epoch": 30.78594249201278,
      "grad_norm": 0.7439934015274048,
      "learning_rate": 7.646153846153847e-06,
      "loss": 0.0069,
      "step": 2409
    },
    {
      "epoch": 30.798722044728436,
      "grad_norm": 1.182193636894226,
      "learning_rate": 7.641025641025641e-06,
      "loss": 0.0089,
      "step": 2410
    },
    {
      "epoch": 30.81150159744409,
      "grad_norm": 1.1009929180145264,
      "learning_rate": 7.635897435897437e-06,
      "loss": 0.0117,
      "step": 2411
    },
    {
      "epoch": 30.824281150159745,
      "grad_norm": 1.8570705652236938,
      "learning_rate": 7.630769230769232e-06,
      "loss": 0.0107,
      "step": 2412
    },
    {
      "epoch": 30.837060702875398,
      "grad_norm": 1.9369806051254272,
      "learning_rate": 7.6256410256410264e-06,
      "loss": 0.0102,
      "step": 2413
    },
    {
      "epoch": 30.849840255591054,
      "grad_norm": 2.9887959957122803,
      "learning_rate": 7.620512820512821e-06,
      "loss": 0.0207,
      "step": 2414
    },
    {
      "epoch": 30.86261980830671,
      "grad_norm": 1.9918948411941528,
      "learning_rate": 7.615384615384615e-06,
      "loss": 0.0203,
      "step": 2415
    },
    {
      "epoch": 30.875399361022364,
      "grad_norm": 2.1750328540802,
      "learning_rate": 7.610256410256411e-06,
      "loss": 0.0208,
      "step": 2416
    },
    {
      "epoch": 30.88817891373802,
      "grad_norm": 1.5640554428100586,
      "learning_rate": 7.605128205128206e-06,
      "loss": 0.0166,
      "step": 2417
    },
    {
      "epoch": 30.900958466453673,
      "grad_norm": 1.598807692527771,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.0104,
      "step": 2418
    },
    {
      "epoch": 30.91373801916933,
      "grad_norm": 40.95674133300781,
      "learning_rate": 7.594871794871795e-06,
      "loss": 0.0492,
      "step": 2419
    },
    {
      "epoch": 30.926517571884983,
      "grad_norm": 2.3611934185028076,
      "learning_rate": 7.58974358974359e-06,
      "loss": 0.016,
      "step": 2420
    },
    {
      "epoch": 30.93929712460064,
      "grad_norm": 1.6652703285217285,
      "learning_rate": 7.584615384615385e-06,
      "loss": 0.0128,
      "step": 2421
    },
    {
      "epoch": 30.952076677316295,
      "grad_norm": 3.1491055488586426,
      "learning_rate": 7.5794871794871805e-06,
      "loss": 0.0193,
      "step": 2422
    },
    {
      "epoch": 30.96485623003195,
      "grad_norm": 3.1785836219787598,
      "learning_rate": 7.574358974358975e-06,
      "loss": 0.0239,
      "step": 2423
    },
    {
      "epoch": 30.977635782747605,
      "grad_norm": 1.2700327634811401,
      "learning_rate": 7.5692307692307695e-06,
      "loss": 0.014,
      "step": 2424
    },
    {
      "epoch": 30.990415335463258,
      "grad_norm": 2.9445688724517822,
      "learning_rate": 7.564102564102564e-06,
      "loss": 0.0289,
      "step": 2425
    },
    {
      "epoch": 31.003194888178914,
      "grad_norm": 1.4139279127120972,
      "learning_rate": 7.558974358974359e-06,
      "loss": 0.0157,
      "step": 2426
    },
    {
      "epoch": 31.015974440894567,
      "grad_norm": 1.359743356704712,
      "learning_rate": 7.553846153846155e-06,
      "loss": 0.0114,
      "step": 2427
    },
    {
      "epoch": 31.028753993610223,
      "grad_norm": 1.6461807489395142,
      "learning_rate": 7.54871794871795e-06,
      "loss": 0.0102,
      "step": 2428
    },
    {
      "epoch": 31.04153354632588,
      "grad_norm": 1.2406097650527954,
      "learning_rate": 7.543589743589744e-06,
      "loss": 0.0096,
      "step": 2429
    },
    {
      "epoch": 31.054313099041533,
      "grad_norm": 0.6446853876113892,
      "learning_rate": 7.538461538461539e-06,
      "loss": 0.0067,
      "step": 2430
    },
    {
      "epoch": 31.06709265175719,
      "grad_norm": 0.4024180471897125,
      "learning_rate": 7.533333333333334e-06,
      "loss": 0.0044,
      "step": 2431
    },
    {
      "epoch": 31.079872204472842,
      "grad_norm": 0.8510559797286987,
      "learning_rate": 7.528205128205129e-06,
      "loss": 0.0064,
      "step": 2432
    },
    {
      "epoch": 31.0926517571885,
      "grad_norm": 2.8283066749572754,
      "learning_rate": 7.523076923076924e-06,
      "loss": 0.0271,
      "step": 2433
    },
    {
      "epoch": 31.105431309904155,
      "grad_norm": 0.76324462890625,
      "learning_rate": 7.5179487179487185e-06,
      "loss": 0.0086,
      "step": 2434
    },
    {
      "epoch": 31.118210862619808,
      "grad_norm": 0.6604694128036499,
      "learning_rate": 7.512820512820513e-06,
      "loss": 0.0076,
      "step": 2435
    },
    {
      "epoch": 31.130990415335464,
      "grad_norm": 0.856167197227478,
      "learning_rate": 7.507692307692308e-06,
      "loss": 0.0081,
      "step": 2436
    },
    {
      "epoch": 31.143769968051117,
      "grad_norm": 0.48480814695358276,
      "learning_rate": 7.502564102564102e-06,
      "loss": 0.0053,
      "step": 2437
    },
    {
      "epoch": 31.156549520766774,
      "grad_norm": 4.314848899841309,
      "learning_rate": 7.497435897435899e-06,
      "loss": 0.0498,
      "step": 2438
    },
    {
      "epoch": 31.169329073482427,
      "grad_norm": 3.0163071155548096,
      "learning_rate": 7.492307692307693e-06,
      "loss": 0.0158,
      "step": 2439
    },
    {
      "epoch": 31.182108626198083,
      "grad_norm": 1.104270577430725,
      "learning_rate": 7.487179487179488e-06,
      "loss": 0.0081,
      "step": 2440
    },
    {
      "epoch": 31.19488817891374,
      "grad_norm": 1.8899494409561157,
      "learning_rate": 7.482051282051283e-06,
      "loss": 0.0149,
      "step": 2441
    },
    {
      "epoch": 31.207667731629392,
      "grad_norm": 2.2089486122131348,
      "learning_rate": 7.476923076923077e-06,
      "loss": 0.0136,
      "step": 2442
    },
    {
      "epoch": 31.22044728434505,
      "grad_norm": 0.7608822584152222,
      "learning_rate": 7.4717948717948726e-06,
      "loss": 0.0064,
      "step": 2443
    },
    {
      "epoch": 31.233226837060702,
      "grad_norm": 1.0831680297851562,
      "learning_rate": 7.4666666666666675e-06,
      "loss": 0.0108,
      "step": 2444
    },
    {
      "epoch": 31.24600638977636,
      "grad_norm": 0.8487801551818848,
      "learning_rate": 7.461538461538462e-06,
      "loss": 0.0065,
      "step": 2445
    },
    {
      "epoch": 31.25878594249201,
      "grad_norm": 0.7963667511940002,
      "learning_rate": 7.456410256410257e-06,
      "loss": 0.0066,
      "step": 2446
    },
    {
      "epoch": 31.271565495207668,
      "grad_norm": 0.7370432019233704,
      "learning_rate": 7.451282051282051e-06,
      "loss": 0.0083,
      "step": 2447
    },
    {
      "epoch": 31.284345047923324,
      "grad_norm": 1.1382296085357666,
      "learning_rate": 7.446153846153846e-06,
      "loss": 0.0083,
      "step": 2448
    },
    {
      "epoch": 31.297124600638977,
      "grad_norm": 1.2064332962036133,
      "learning_rate": 7.441025641025642e-06,
      "loss": 0.008,
      "step": 2449
    },
    {
      "epoch": 31.309904153354633,
      "grad_norm": 1.4648174047470093,
      "learning_rate": 7.435897435897437e-06,
      "loss": 0.0073,
      "step": 2450
    },
    {
      "epoch": 31.322683706070286,
      "grad_norm": 1.0092828273773193,
      "learning_rate": 7.430769230769232e-06,
      "loss": 0.0085,
      "step": 2451
    },
    {
      "epoch": 31.335463258785943,
      "grad_norm": 2.0985512733459473,
      "learning_rate": 7.425641025641026e-06,
      "loss": 0.0088,
      "step": 2452
    },
    {
      "epoch": 31.3482428115016,
      "grad_norm": 2.0856070518493652,
      "learning_rate": 7.420512820512821e-06,
      "loss": 0.0115,
      "step": 2453
    },
    {
      "epoch": 31.361022364217252,
      "grad_norm": 3.9369513988494873,
      "learning_rate": 7.4153846153846164e-06,
      "loss": 0.0291,
      "step": 2454
    },
    {
      "epoch": 31.37380191693291,
      "grad_norm": 0.8650026917457581,
      "learning_rate": 7.410256410256411e-06,
      "loss": 0.0085,
      "step": 2455
    },
    {
      "epoch": 31.38658146964856,
      "grad_norm": 2.597470760345459,
      "learning_rate": 7.405128205128206e-06,
      "loss": 0.0131,
      "step": 2456
    },
    {
      "epoch": 31.399361022364218,
      "grad_norm": 1.0550893545150757,
      "learning_rate": 7.4e-06,
      "loss": 0.0071,
      "step": 2457
    },
    {
      "epoch": 31.41214057507987,
      "grad_norm": 1.2370715141296387,
      "learning_rate": 7.394871794871795e-06,
      "loss": 0.0079,
      "step": 2458
    },
    {
      "epoch": 31.424920127795527,
      "grad_norm": 2.8915867805480957,
      "learning_rate": 7.38974358974359e-06,
      "loss": 0.0274,
      "step": 2459
    },
    {
      "epoch": 31.437699680511184,
      "grad_norm": 1.7772983312606812,
      "learning_rate": 7.384615384615386e-06,
      "loss": 0.0082,
      "step": 2460
    },
    {
      "epoch": 31.450479233226837,
      "grad_norm": 2.2857325077056885,
      "learning_rate": 7.37948717948718e-06,
      "loss": 0.0135,
      "step": 2461
    },
    {
      "epoch": 31.463258785942493,
      "grad_norm": 3.7650675773620605,
      "learning_rate": 7.374358974358975e-06,
      "loss": 0.0247,
      "step": 2462
    },
    {
      "epoch": 31.476038338658146,
      "grad_norm": 2.1733803749084473,
      "learning_rate": 7.36923076923077e-06,
      "loss": 0.0179,
      "step": 2463
    },
    {
      "epoch": 31.488817891373802,
      "grad_norm": 3.0555264949798584,
      "learning_rate": 7.364102564102565e-06,
      "loss": 0.0138,
      "step": 2464
    },
    {
      "epoch": 31.501597444089455,
      "grad_norm": 4.623447895050049,
      "learning_rate": 7.35897435897436e-06,
      "loss": 0.0163,
      "step": 2465
    },
    {
      "epoch": 31.51437699680511,
      "grad_norm": 1.1830363273620605,
      "learning_rate": 7.353846153846154e-06,
      "loss": 0.0064,
      "step": 2466
    },
    {
      "epoch": 31.527156549520768,
      "grad_norm": 1.612902283668518,
      "learning_rate": 7.348717948717949e-06,
      "loss": 0.0148,
      "step": 2467
    },
    {
      "epoch": 31.53993610223642,
      "grad_norm": 2.0270512104034424,
      "learning_rate": 7.343589743589744e-06,
      "loss": 0.0143,
      "step": 2468
    },
    {
      "epoch": 31.552715654952078,
      "grad_norm": 4.224149227142334,
      "learning_rate": 7.338461538461539e-06,
      "loss": 0.041,
      "step": 2469
    },
    {
      "epoch": 31.56549520766773,
      "grad_norm": 1.3380467891693115,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.0072,
      "step": 2470
    },
    {
      "epoch": 31.578274760383387,
      "grad_norm": 0.8487076163291931,
      "learning_rate": 7.328205128205129e-06,
      "loss": 0.011,
      "step": 2471
    },
    {
      "epoch": 31.591054313099043,
      "grad_norm": 0.8845389485359192,
      "learning_rate": 7.323076923076924e-06,
      "loss": 0.0092,
      "step": 2472
    },
    {
      "epoch": 31.603833865814696,
      "grad_norm": 2.505274772644043,
      "learning_rate": 7.317948717948719e-06,
      "loss": 0.0123,
      "step": 2473
    },
    {
      "epoch": 31.616613418530353,
      "grad_norm": 1.9374302625656128,
      "learning_rate": 7.312820512820514e-06,
      "loss": 0.0086,
      "step": 2474
    },
    {
      "epoch": 31.629392971246006,
      "grad_norm": 0.6446139216423035,
      "learning_rate": 7.307692307692308e-06,
      "loss": 0.0055,
      "step": 2475
    },
    {
      "epoch": 31.642172523961662,
      "grad_norm": 0.738622784614563,
      "learning_rate": 7.302564102564103e-06,
      "loss": 0.0076,
      "step": 2476
    },
    {
      "epoch": 31.654952076677315,
      "grad_norm": 1.9127014875411987,
      "learning_rate": 7.297435897435898e-06,
      "loss": 0.0121,
      "step": 2477
    },
    {
      "epoch": 31.66773162939297,
      "grad_norm": 3.925105571746826,
      "learning_rate": 7.292307692307693e-06,
      "loss": 0.0257,
      "step": 2478
    },
    {
      "epoch": 31.680511182108628,
      "grad_norm": 1.7984658479690552,
      "learning_rate": 7.287179487179487e-06,
      "loss": 0.011,
      "step": 2479
    },
    {
      "epoch": 31.69329073482428,
      "grad_norm": 2.5924363136291504,
      "learning_rate": 7.282051282051282e-06,
      "loss": 0.0202,
      "step": 2480
    },
    {
      "epoch": 31.706070287539937,
      "grad_norm": 3.3326330184936523,
      "learning_rate": 7.276923076923077e-06,
      "loss": 0.0312,
      "step": 2481
    },
    {
      "epoch": 31.71884984025559,
      "grad_norm": 2.6017725467681885,
      "learning_rate": 7.271794871794873e-06,
      "loss": 0.0229,
      "step": 2482
    },
    {
      "epoch": 31.731629392971247,
      "grad_norm": 2.4673171043395996,
      "learning_rate": 7.266666666666668e-06,
      "loss": 0.0214,
      "step": 2483
    },
    {
      "epoch": 31.7444089456869,
      "grad_norm": 1.627975583076477,
      "learning_rate": 7.261538461538462e-06,
      "loss": 0.0215,
      "step": 2484
    },
    {
      "epoch": 31.757188498402556,
      "grad_norm": 2.3996481895446777,
      "learning_rate": 7.256410256410257e-06,
      "loss": 0.0184,
      "step": 2485
    },
    {
      "epoch": 31.769968051118212,
      "grad_norm": 1.8067755699157715,
      "learning_rate": 7.2512820512820515e-06,
      "loss": 0.0076,
      "step": 2486
    },
    {
      "epoch": 31.782747603833865,
      "grad_norm": 0.9003115892410278,
      "learning_rate": 7.246153846153847e-06,
      "loss": 0.0089,
      "step": 2487
    },
    {
      "epoch": 31.79552715654952,
      "grad_norm": 1.9725600481033325,
      "learning_rate": 7.241025641025642e-06,
      "loss": 0.0175,
      "step": 2488
    },
    {
      "epoch": 31.808306709265175,
      "grad_norm": 0.8610435128211975,
      "learning_rate": 7.235897435897436e-06,
      "loss": 0.0214,
      "step": 2489
    },
    {
      "epoch": 31.82108626198083,
      "grad_norm": 1.769567847251892,
      "learning_rate": 7.230769230769231e-06,
      "loss": 0.0149,
      "step": 2490
    },
    {
      "epoch": 31.833865814696484,
      "grad_norm": 1.4577288627624512,
      "learning_rate": 7.225641025641026e-06,
      "loss": 0.0114,
      "step": 2491
    },
    {
      "epoch": 31.84664536741214,
      "grad_norm": 1.6688158512115479,
      "learning_rate": 7.220512820512822e-06,
      "loss": 0.0151,
      "step": 2492
    },
    {
      "epoch": 31.859424920127797,
      "grad_norm": 1.0488781929016113,
      "learning_rate": 7.215384615384617e-06,
      "loss": 0.0078,
      "step": 2493
    },
    {
      "epoch": 31.87220447284345,
      "grad_norm": 1.9858828783035278,
      "learning_rate": 7.210256410256411e-06,
      "loss": 0.0101,
      "step": 2494
    },
    {
      "epoch": 31.884984025559106,
      "grad_norm": 2.6534643173217773,
      "learning_rate": 7.205128205128206e-06,
      "loss": 0.0147,
      "step": 2495
    },
    {
      "epoch": 31.89776357827476,
      "grad_norm": 0.849526047706604,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.0081,
      "step": 2496
    },
    {
      "epoch": 31.910543130990416,
      "grad_norm": 1.108740210533142,
      "learning_rate": 7.1948717948717946e-06,
      "loss": 0.0106,
      "step": 2497
    },
    {
      "epoch": 31.923322683706072,
      "grad_norm": 1.1372129917144775,
      "learning_rate": 7.189743589743591e-06,
      "loss": 0.0086,
      "step": 2498
    },
    {
      "epoch": 31.936102236421725,
      "grad_norm": 0.8639397621154785,
      "learning_rate": 7.184615384615385e-06,
      "loss": 0.0097,
      "step": 2499
    },
    {
      "epoch": 31.94888178913738,
      "grad_norm": 1.0387763977050781,
      "learning_rate": 7.17948717948718e-06,
      "loss": 0.0107,
      "step": 2500
    },
    {
      "epoch": 31.94888178913738,
      "eval_loss": 0.9020013213157654,
      "eval_runtime": 183.5845,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 2500
    },
    {
      "epoch": 31.961661341853034,
      "grad_norm": 1.7540512084960938,
      "learning_rate": 7.174358974358975e-06,
      "loss": 0.0101,
      "step": 2501
    },
    {
      "epoch": 31.97444089456869,
      "grad_norm": 1.5571906566619873,
      "learning_rate": 7.169230769230769e-06,
      "loss": 0.0119,
      "step": 2502
    },
    {
      "epoch": 31.987220447284344,
      "grad_norm": 3.689589023590088,
      "learning_rate": 7.164102564102565e-06,
      "loss": 0.0188,
      "step": 2503
    },
    {
      "epoch": 32.0,
      "grad_norm": 9.199419975280762,
      "learning_rate": 7.15897435897436e-06,
      "loss": 0.0465,
      "step": 2504
    },
    {
      "epoch": 32.01277955271566,
      "grad_norm": 47.8953857421875,
      "learning_rate": 7.153846153846155e-06,
      "loss": 0.0376,
      "step": 2505
    },
    {
      "epoch": 32.02555910543131,
      "grad_norm": 0.7875796556472778,
      "learning_rate": 7.1487179487179495e-06,
      "loss": 0.0102,
      "step": 2506
    },
    {
      "epoch": 32.03833865814696,
      "grad_norm": 1.4419995546340942,
      "learning_rate": 7.1435897435897436e-06,
      "loss": 0.0114,
      "step": 2507
    },
    {
      "epoch": 32.05111821086262,
      "grad_norm": 2.094075918197632,
      "learning_rate": 7.1384615384615385e-06,
      "loss": 0.0124,
      "step": 2508
    },
    {
      "epoch": 32.063897763578275,
      "grad_norm": 0.5076902508735657,
      "learning_rate": 7.133333333333334e-06,
      "loss": 0.0048,
      "step": 2509
    },
    {
      "epoch": 32.07667731629393,
      "grad_norm": 1.2808444499969482,
      "learning_rate": 7.128205128205129e-06,
      "loss": 0.0195,
      "step": 2510
    },
    {
      "epoch": 32.08945686900959,
      "grad_norm": 1.2106831073760986,
      "learning_rate": 7.123076923076924e-06,
      "loss": 0.0126,
      "step": 2511
    },
    {
      "epoch": 32.10223642172524,
      "grad_norm": 2.212151288986206,
      "learning_rate": 7.117948717948718e-06,
      "loss": 0.0068,
      "step": 2512
    },
    {
      "epoch": 32.115015974440894,
      "grad_norm": 1.6798101663589478,
      "learning_rate": 7.112820512820513e-06,
      "loss": 0.0091,
      "step": 2513
    },
    {
      "epoch": 32.12779552715655,
      "grad_norm": 0.4618903696537018,
      "learning_rate": 7.107692307692309e-06,
      "loss": 0.005,
      "step": 2514
    },
    {
      "epoch": 32.14057507987221,
      "grad_norm": 2.6432876586914062,
      "learning_rate": 7.102564102564104e-06,
      "loss": 0.0137,
      "step": 2515
    },
    {
      "epoch": 32.153354632587856,
      "grad_norm": 0.6872953772544861,
      "learning_rate": 7.0974358974358985e-06,
      "loss": 0.0082,
      "step": 2516
    },
    {
      "epoch": 32.16613418530351,
      "grad_norm": 0.45526954531669617,
      "learning_rate": 7.0923076923076926e-06,
      "loss": 0.0057,
      "step": 2517
    },
    {
      "epoch": 32.17891373801917,
      "grad_norm": 1.115525245666504,
      "learning_rate": 7.0871794871794875e-06,
      "loss": 0.0101,
      "step": 2518
    },
    {
      "epoch": 32.191693290734825,
      "grad_norm": 1.0506788492202759,
      "learning_rate": 7.082051282051282e-06,
      "loss": 0.0073,
      "step": 2519
    },
    {
      "epoch": 32.20447284345048,
      "grad_norm": 0.7957150936126709,
      "learning_rate": 7.076923076923078e-06,
      "loss": 0.0063,
      "step": 2520
    },
    {
      "epoch": 32.21725239616613,
      "grad_norm": 0.6349469423294067,
      "learning_rate": 7.071794871794872e-06,
      "loss": 0.0054,
      "step": 2521
    },
    {
      "epoch": 32.23003194888179,
      "grad_norm": 2.232160806655884,
      "learning_rate": 7.066666666666667e-06,
      "loss": 0.012,
      "step": 2522
    },
    {
      "epoch": 32.242811501597444,
      "grad_norm": 0.7386549115180969,
      "learning_rate": 7.061538461538462e-06,
      "loss": 0.0059,
      "step": 2523
    },
    {
      "epoch": 32.2555910543131,
      "grad_norm": 0.6190231442451477,
      "learning_rate": 7.056410256410257e-06,
      "loss": 0.0063,
      "step": 2524
    },
    {
      "epoch": 32.26837060702876,
      "grad_norm": 0.7233198881149292,
      "learning_rate": 7.051282051282053e-06,
      "loss": 0.0091,
      "step": 2525
    },
    {
      "epoch": 32.281150159744406,
      "grad_norm": 0.612321138381958,
      "learning_rate": 7.046153846153847e-06,
      "loss": 0.0068,
      "step": 2526
    },
    {
      "epoch": 32.29392971246006,
      "grad_norm": 2.0452654361724854,
      "learning_rate": 7.0410256410256415e-06,
      "loss": 0.0131,
      "step": 2527
    },
    {
      "epoch": 32.30670926517572,
      "grad_norm": 1.4713987112045288,
      "learning_rate": 7.0358974358974364e-06,
      "loss": 0.0089,
      "step": 2528
    },
    {
      "epoch": 32.319488817891376,
      "grad_norm": 2.9305877685546875,
      "learning_rate": 7.030769230769231e-06,
      "loss": 0.0205,
      "step": 2529
    },
    {
      "epoch": 32.33226837060703,
      "grad_norm": 3.299680233001709,
      "learning_rate": 7.025641025641025e-06,
      "loss": 0.0293,
      "step": 2530
    },
    {
      "epoch": 32.34504792332268,
      "grad_norm": 1.3041348457336426,
      "learning_rate": 7.020512820512821e-06,
      "loss": 0.0115,
      "step": 2531
    },
    {
      "epoch": 32.35782747603834,
      "grad_norm": 1.1142361164093018,
      "learning_rate": 7.015384615384616e-06,
      "loss": 0.0093,
      "step": 2532
    },
    {
      "epoch": 32.370607028753994,
      "grad_norm": 2.617877721786499,
      "learning_rate": 7.010256410256411e-06,
      "loss": 0.0168,
      "step": 2533
    },
    {
      "epoch": 32.38338658146965,
      "grad_norm": 1.5802733898162842,
      "learning_rate": 7.005128205128206e-06,
      "loss": 0.012,
      "step": 2534
    },
    {
      "epoch": 32.3961661341853,
      "grad_norm": 3.298232316970825,
      "learning_rate": 7e-06,
      "loss": 0.0196,
      "step": 2535
    },
    {
      "epoch": 32.40894568690096,
      "grad_norm": 1.9718133211135864,
      "learning_rate": 6.994871794871796e-06,
      "loss": 0.011,
      "step": 2536
    },
    {
      "epoch": 32.42172523961661,
      "grad_norm": 2.5648529529571533,
      "learning_rate": 6.9897435897435905e-06,
      "loss": 0.0283,
      "step": 2537
    },
    {
      "epoch": 32.43450479233227,
      "grad_norm": 1.1077498197555542,
      "learning_rate": 6.9846153846153854e-06,
      "loss": 0.0076,
      "step": 2538
    },
    {
      "epoch": 32.447284345047926,
      "grad_norm": 0.5472296476364136,
      "learning_rate": 6.9794871794871795e-06,
      "loss": 0.0048,
      "step": 2539
    },
    {
      "epoch": 32.460063897763575,
      "grad_norm": 3.183478832244873,
      "learning_rate": 6.974358974358974e-06,
      "loss": 0.0139,
      "step": 2540
    },
    {
      "epoch": 32.47284345047923,
      "grad_norm": 1.6973299980163574,
      "learning_rate": 6.96923076923077e-06,
      "loss": 0.0178,
      "step": 2541
    },
    {
      "epoch": 32.48562300319489,
      "grad_norm": 0.7870661020278931,
      "learning_rate": 6.964102564102565e-06,
      "loss": 0.007,
      "step": 2542
    },
    {
      "epoch": 32.498402555910545,
      "grad_norm": 1.308314561843872,
      "learning_rate": 6.95897435897436e-06,
      "loss": 0.0075,
      "step": 2543
    },
    {
      "epoch": 32.5111821086262,
      "grad_norm": 0.9130819439888,
      "learning_rate": 6.953846153846154e-06,
      "loss": 0.009,
      "step": 2544
    },
    {
      "epoch": 32.52396166134185,
      "grad_norm": 0.6322726607322693,
      "learning_rate": 6.948717948717949e-06,
      "loss": 0.006,
      "step": 2545
    },
    {
      "epoch": 32.53674121405751,
      "grad_norm": 1.411386251449585,
      "learning_rate": 6.943589743589744e-06,
      "loss": 0.02,
      "step": 2546
    },
    {
      "epoch": 32.54952076677316,
      "grad_norm": 0.5440376400947571,
      "learning_rate": 6.9384615384615395e-06,
      "loss": 0.0052,
      "step": 2547
    },
    {
      "epoch": 32.56230031948882,
      "grad_norm": 1.4267332553863525,
      "learning_rate": 6.9333333333333344e-06,
      "loss": 0.0069,
      "step": 2548
    },
    {
      "epoch": 32.575079872204476,
      "grad_norm": 0.9322569966316223,
      "learning_rate": 6.9282051282051285e-06,
      "loss": 0.0083,
      "step": 2549
    },
    {
      "epoch": 32.587859424920126,
      "grad_norm": 3.507944107055664,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.0276,
      "step": 2550
    },
    {
      "epoch": 32.60063897763578,
      "grad_norm": 0.6120370030403137,
      "learning_rate": 6.917948717948718e-06,
      "loss": 0.0054,
      "step": 2551
    },
    {
      "epoch": 32.61341853035144,
      "grad_norm": 1.4462029933929443,
      "learning_rate": 6.912820512820514e-06,
      "loss": 0.0112,
      "step": 2552
    },
    {
      "epoch": 32.626198083067095,
      "grad_norm": 9.039031028747559,
      "learning_rate": 6.907692307692309e-06,
      "loss": 0.0083,
      "step": 2553
    },
    {
      "epoch": 32.638977635782744,
      "grad_norm": 1.2227258682250977,
      "learning_rate": 6.902564102564103e-06,
      "loss": 0.0088,
      "step": 2554
    },
    {
      "epoch": 32.6517571884984,
      "grad_norm": 2.9861364364624023,
      "learning_rate": 6.897435897435898e-06,
      "loss": 0.0166,
      "step": 2555
    },
    {
      "epoch": 32.66453674121406,
      "grad_norm": 0.9244511127471924,
      "learning_rate": 6.892307692307693e-06,
      "loss": 0.0069,
      "step": 2556
    },
    {
      "epoch": 32.677316293929714,
      "grad_norm": 2.979329824447632,
      "learning_rate": 6.887179487179488e-06,
      "loss": 0.012,
      "step": 2557
    },
    {
      "epoch": 32.69009584664537,
      "grad_norm": 1.4540989398956299,
      "learning_rate": 6.882051282051283e-06,
      "loss": 0.0108,
      "step": 2558
    },
    {
      "epoch": 32.70287539936102,
      "grad_norm": 2.675171375274658,
      "learning_rate": 6.8769230769230775e-06,
      "loss": 0.0207,
      "step": 2559
    },
    {
      "epoch": 32.715654952076676,
      "grad_norm": 1.284254550933838,
      "learning_rate": 6.871794871794872e-06,
      "loss": 0.0063,
      "step": 2560
    },
    {
      "epoch": 32.72843450479233,
      "grad_norm": 2.4144058227539062,
      "learning_rate": 6.866666666666667e-06,
      "loss": 0.0146,
      "step": 2561
    },
    {
      "epoch": 32.74121405750799,
      "grad_norm": 2.105947971343994,
      "learning_rate": 6.861538461538461e-06,
      "loss": 0.0167,
      "step": 2562
    },
    {
      "epoch": 32.753993610223645,
      "grad_norm": 0.9299395084381104,
      "learning_rate": 6.856410256410257e-06,
      "loss": 0.0088,
      "step": 2563
    },
    {
      "epoch": 32.766773162939295,
      "grad_norm": 1.4186254739761353,
      "learning_rate": 6.851282051282052e-06,
      "loss": 0.0081,
      "step": 2564
    },
    {
      "epoch": 32.77955271565495,
      "grad_norm": 2.574317216873169,
      "learning_rate": 6.846153846153847e-06,
      "loss": 0.0192,
      "step": 2565
    },
    {
      "epoch": 32.79233226837061,
      "grad_norm": 0.4454169273376465,
      "learning_rate": 6.841025641025642e-06,
      "loss": 0.0039,
      "step": 2566
    },
    {
      "epoch": 32.805111821086264,
      "grad_norm": 3.510132312774658,
      "learning_rate": 6.835897435897436e-06,
      "loss": 0.0174,
      "step": 2567
    },
    {
      "epoch": 32.81789137380191,
      "grad_norm": 2.4825921058654785,
      "learning_rate": 6.830769230769231e-06,
      "loss": 0.0243,
      "step": 2568
    },
    {
      "epoch": 32.83067092651757,
      "grad_norm": 0.8653742671012878,
      "learning_rate": 6.8256410256410265e-06,
      "loss": 0.0069,
      "step": 2569
    },
    {
      "epoch": 32.843450479233226,
      "grad_norm": 2.524458408355713,
      "learning_rate": 6.820512820512821e-06,
      "loss": 0.0157,
      "step": 2570
    },
    {
      "epoch": 32.85623003194888,
      "grad_norm": 1.3834733963012695,
      "learning_rate": 6.815384615384616e-06,
      "loss": 0.0059,
      "step": 2571
    },
    {
      "epoch": 32.86900958466454,
      "grad_norm": 0.8807143568992615,
      "learning_rate": 6.81025641025641e-06,
      "loss": 0.0076,
      "step": 2572
    },
    {
      "epoch": 32.88178913738019,
      "grad_norm": 4.4518561363220215,
      "learning_rate": 6.805128205128205e-06,
      "loss": 0.0194,
      "step": 2573
    },
    {
      "epoch": 32.894568690095845,
      "grad_norm": 0.6095708608627319,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.0064,
      "step": 2574
    },
    {
      "epoch": 32.9073482428115,
      "grad_norm": 4.341555595397949,
      "learning_rate": 6.794871794871796e-06,
      "loss": 0.0369,
      "step": 2575
    },
    {
      "epoch": 32.92012779552716,
      "grad_norm": 3.9712815284729004,
      "learning_rate": 6.789743589743591e-06,
      "loss": 0.0314,
      "step": 2576
    },
    {
      "epoch": 32.932907348242814,
      "grad_norm": 0.6758043766021729,
      "learning_rate": 6.784615384615385e-06,
      "loss": 0.0063,
      "step": 2577
    },
    {
      "epoch": 32.945686900958464,
      "grad_norm": 3.790515422821045,
      "learning_rate": 6.77948717948718e-06,
      "loss": 0.0175,
      "step": 2578
    },
    {
      "epoch": 32.95846645367412,
      "grad_norm": 1.8134866952896118,
      "learning_rate": 6.774358974358975e-06,
      "loss": 0.0131,
      "step": 2579
    },
    {
      "epoch": 32.97124600638978,
      "grad_norm": 0.7580832839012146,
      "learning_rate": 6.76923076923077e-06,
      "loss": 0.0068,
      "step": 2580
    },
    {
      "epoch": 32.98402555910543,
      "grad_norm": 1.2273781299591064,
      "learning_rate": 6.764102564102564e-06,
      "loss": 0.0068,
      "step": 2581
    },
    {
      "epoch": 32.99680511182109,
      "grad_norm": 3.0153393745422363,
      "learning_rate": 6.758974358974359e-06,
      "loss": 0.0205,
      "step": 2582
    },
    {
      "epoch": 33.00958466453674,
      "grad_norm": 0.7186205983161926,
      "learning_rate": 6.753846153846154e-06,
      "loss": 0.0071,
      "step": 2583
    },
    {
      "epoch": 33.022364217252395,
      "grad_norm": 0.7623568773269653,
      "learning_rate": 6.748717948717949e-06,
      "loss": 0.0067,
      "step": 2584
    },
    {
      "epoch": 33.03514376996805,
      "grad_norm": 1.622061848640442,
      "learning_rate": 6.743589743589745e-06,
      "loss": 0.0097,
      "step": 2585
    },
    {
      "epoch": 33.04792332268371,
      "grad_norm": 0.7285746335983276,
      "learning_rate": 6.738461538461539e-06,
      "loss": 0.0053,
      "step": 2586
    },
    {
      "epoch": 33.06070287539936,
      "grad_norm": 0.7648468613624573,
      "learning_rate": 6.733333333333334e-06,
      "loss": 0.0052,
      "step": 2587
    },
    {
      "epoch": 33.073482428115014,
      "grad_norm": 2.6559994220733643,
      "learning_rate": 6.728205128205129e-06,
      "loss": 0.0138,
      "step": 2588
    },
    {
      "epoch": 33.08626198083067,
      "grad_norm": 2.8083527088165283,
      "learning_rate": 6.723076923076924e-06,
      "loss": 0.0149,
      "step": 2589
    },
    {
      "epoch": 33.09904153354633,
      "grad_norm": 0.924113392829895,
      "learning_rate": 6.717948717948718e-06,
      "loss": 0.0086,
      "step": 2590
    },
    {
      "epoch": 33.11182108626198,
      "grad_norm": 0.6254075765609741,
      "learning_rate": 6.712820512820513e-06,
      "loss": 0.0035,
      "step": 2591
    },
    {
      "epoch": 33.12460063897763,
      "grad_norm": 0.8559412956237793,
      "learning_rate": 6.707692307692308e-06,
      "loss": 0.0067,
      "step": 2592
    },
    {
      "epoch": 33.13738019169329,
      "grad_norm": 0.5838778614997864,
      "learning_rate": 6.702564102564103e-06,
      "loss": 0.0041,
      "step": 2593
    },
    {
      "epoch": 33.150159744408946,
      "grad_norm": 0.6391317844390869,
      "learning_rate": 6.697435897435898e-06,
      "loss": 0.0057,
      "step": 2594
    },
    {
      "epoch": 33.1629392971246,
      "grad_norm": 2.1996893882751465,
      "learning_rate": 6.692307692307692e-06,
      "loss": 0.0124,
      "step": 2595
    },
    {
      "epoch": 33.17571884984026,
      "grad_norm": 3.635840892791748,
      "learning_rate": 6.687179487179488e-06,
      "loss": 0.0202,
      "step": 2596
    },
    {
      "epoch": 33.18849840255591,
      "grad_norm": 1.2818849086761475,
      "learning_rate": 6.682051282051283e-06,
      "loss": 0.0076,
      "step": 2597
    },
    {
      "epoch": 33.201277955271564,
      "grad_norm": 0.7113415002822876,
      "learning_rate": 6.676923076923078e-06,
      "loss": 0.0054,
      "step": 2598
    },
    {
      "epoch": 33.21405750798722,
      "grad_norm": 0.8988485932350159,
      "learning_rate": 6.671794871794873e-06,
      "loss": 0.0053,
      "step": 2599
    },
    {
      "epoch": 33.22683706070288,
      "grad_norm": 1.191688895225525,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.009,
      "step": 2600
    },
    {
      "epoch": 33.22683706070288,
      "eval_loss": 0.9255790710449219,
      "eval_runtime": 183.4128,
      "eval_samples_per_second": 0.856,
      "eval_steps_per_second": 0.109,
      "step": 2600
    },
    {
      "epoch": 33.239616613418534,
      "grad_norm": 0.3343746066093445,
      "learning_rate": 6.661538461538462e-06,
      "loss": 0.0036,
      "step": 2601
    },
    {
      "epoch": 33.25239616613418,
      "grad_norm": 0.6919339895248413,
      "learning_rate": 6.656410256410257e-06,
      "loss": 0.0059,
      "step": 2602
    },
    {
      "epoch": 33.26517571884984,
      "grad_norm": 0.7552350163459778,
      "learning_rate": 6.651282051282052e-06,
      "loss": 0.0079,
      "step": 2603
    },
    {
      "epoch": 33.277955271565496,
      "grad_norm": 1.3328794240951538,
      "learning_rate": 6.646153846153846e-06,
      "loss": 0.0091,
      "step": 2604
    },
    {
      "epoch": 33.29073482428115,
      "grad_norm": 0.6286408305168152,
      "learning_rate": 6.641025641025641e-06,
      "loss": 0.0056,
      "step": 2605
    },
    {
      "epoch": 33.3035143769968,
      "grad_norm": 0.548167884349823,
      "learning_rate": 6.635897435897436e-06,
      "loss": 0.0062,
      "step": 2606
    },
    {
      "epoch": 33.31629392971246,
      "grad_norm": 1.6738111972808838,
      "learning_rate": 6.630769230769232e-06,
      "loss": 0.0102,
      "step": 2607
    },
    {
      "epoch": 33.329073482428115,
      "grad_norm": 0.5843906402587891,
      "learning_rate": 6.625641025641027e-06,
      "loss": 0.0035,
      "step": 2608
    },
    {
      "epoch": 33.34185303514377,
      "grad_norm": 0.33983996510505676,
      "learning_rate": 6.620512820512821e-06,
      "loss": 0.0036,
      "step": 2609
    },
    {
      "epoch": 33.35463258785943,
      "grad_norm": 0.44117605686187744,
      "learning_rate": 6.615384615384616e-06,
      "loss": 0.0048,
      "step": 2610
    },
    {
      "epoch": 33.36741214057508,
      "grad_norm": 0.572445809841156,
      "learning_rate": 6.6102564102564105e-06,
      "loss": 0.0039,
      "step": 2611
    },
    {
      "epoch": 33.38019169329073,
      "grad_norm": 1.3580918312072754,
      "learning_rate": 6.605128205128206e-06,
      "loss": 0.0081,
      "step": 2612
    },
    {
      "epoch": 33.39297124600639,
      "grad_norm": 2.4212586879730225,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.0238,
      "step": 2613
    },
    {
      "epoch": 33.405750798722046,
      "grad_norm": 3.360891342163086,
      "learning_rate": 6.594871794871795e-06,
      "loss": 0.0278,
      "step": 2614
    },
    {
      "epoch": 33.4185303514377,
      "grad_norm": 2.3740994930267334,
      "learning_rate": 6.58974358974359e-06,
      "loss": 0.0125,
      "step": 2615
    },
    {
      "epoch": 33.43130990415335,
      "grad_norm": 3.995941638946533,
      "learning_rate": 6.584615384615385e-06,
      "loss": 0.009,
      "step": 2616
    },
    {
      "epoch": 33.44408945686901,
      "grad_norm": 0.8252402544021606,
      "learning_rate": 6.57948717948718e-06,
      "loss": 0.0059,
      "step": 2617
    },
    {
      "epoch": 33.456869009584665,
      "grad_norm": 3.215841054916382,
      "learning_rate": 6.574358974358976e-06,
      "loss": 0.0196,
      "step": 2618
    },
    {
      "epoch": 33.46964856230032,
      "grad_norm": 2.137525796890259,
      "learning_rate": 6.56923076923077e-06,
      "loss": 0.0167,
      "step": 2619
    },
    {
      "epoch": 33.48242811501598,
      "grad_norm": 0.39872461557388306,
      "learning_rate": 6.564102564102565e-06,
      "loss": 0.0046,
      "step": 2620
    },
    {
      "epoch": 33.49520766773163,
      "grad_norm": 0.7364473342895508,
      "learning_rate": 6.5589743589743595e-06,
      "loss": 0.0087,
      "step": 2621
    },
    {
      "epoch": 33.50798722044728,
      "grad_norm": 0.5490440130233765,
      "learning_rate": 6.553846153846154e-06,
      "loss": 0.0062,
      "step": 2622
    },
    {
      "epoch": 33.52076677316294,
      "grad_norm": 0.509581446647644,
      "learning_rate": 6.548717948717949e-06,
      "loss": 0.0059,
      "step": 2623
    },
    {
      "epoch": 33.533546325878596,
      "grad_norm": 0.47174230217933655,
      "learning_rate": 6.543589743589744e-06,
      "loss": 0.0038,
      "step": 2624
    },
    {
      "epoch": 33.546325878594246,
      "grad_norm": 0.7682559490203857,
      "learning_rate": 6.538461538461539e-06,
      "loss": 0.0063,
      "step": 2625
    },
    {
      "epoch": 33.5591054313099,
      "grad_norm": 1.427538514137268,
      "learning_rate": 6.533333333333334e-06,
      "loss": 0.0118,
      "step": 2626
    },
    {
      "epoch": 33.57188498402556,
      "grad_norm": 0.7195096611976624,
      "learning_rate": 6.528205128205128e-06,
      "loss": 0.0063,
      "step": 2627
    },
    {
      "epoch": 33.584664536741215,
      "grad_norm": 0.47817856073379517,
      "learning_rate": 6.523076923076923e-06,
      "loss": 0.0053,
      "step": 2628
    },
    {
      "epoch": 33.59744408945687,
      "grad_norm": 0.9087768197059631,
      "learning_rate": 6.517948717948719e-06,
      "loss": 0.0086,
      "step": 2629
    },
    {
      "epoch": 33.61022364217252,
      "grad_norm": 0.6149556040763855,
      "learning_rate": 6.512820512820514e-06,
      "loss": 0.0069,
      "step": 2630
    },
    {
      "epoch": 33.62300319488818,
      "grad_norm": 1.6814866065979004,
      "learning_rate": 6.5076923076923085e-06,
      "loss": 0.0165,
      "step": 2631
    },
    {
      "epoch": 33.635782747603834,
      "grad_norm": 2.084594249725342,
      "learning_rate": 6.5025641025641026e-06,
      "loss": 0.0138,
      "step": 2632
    },
    {
      "epoch": 33.64856230031949,
      "grad_norm": 0.8016097545623779,
      "learning_rate": 6.4974358974358975e-06,
      "loss": 0.0073,
      "step": 2633
    },
    {
      "epoch": 33.66134185303515,
      "grad_norm": 0.9108177423477173,
      "learning_rate": 6.492307692307693e-06,
      "loss": 0.0098,
      "step": 2634
    },
    {
      "epoch": 33.674121405750796,
      "grad_norm": 0.7251105904579163,
      "learning_rate": 6.487179487179488e-06,
      "loss": 0.0069,
      "step": 2635
    },
    {
      "epoch": 33.68690095846645,
      "grad_norm": 2.2520105838775635,
      "learning_rate": 6.482051282051283e-06,
      "loss": 0.0141,
      "step": 2636
    },
    {
      "epoch": 33.69968051118211,
      "grad_norm": 5.15416955947876,
      "learning_rate": 6.476923076923077e-06,
      "loss": 0.0279,
      "step": 2637
    },
    {
      "epoch": 33.712460063897765,
      "grad_norm": 2.7999393939971924,
      "learning_rate": 6.471794871794872e-06,
      "loss": 0.0195,
      "step": 2638
    },
    {
      "epoch": 33.72523961661342,
      "grad_norm": 1.4104244709014893,
      "learning_rate": 6.466666666666667e-06,
      "loss": 0.01,
      "step": 2639
    },
    {
      "epoch": 33.73801916932907,
      "grad_norm": 3.1084578037261963,
      "learning_rate": 6.461538461538463e-06,
      "loss": 0.007,
      "step": 2640
    },
    {
      "epoch": 33.75079872204473,
      "grad_norm": 1.16458261013031,
      "learning_rate": 6.4564102564102575e-06,
      "loss": 0.0102,
      "step": 2641
    },
    {
      "epoch": 33.763578274760384,
      "grad_norm": 2.8983755111694336,
      "learning_rate": 6.4512820512820516e-06,
      "loss": 0.015,
      "step": 2642
    },
    {
      "epoch": 33.77635782747604,
      "grad_norm": 0.38780269026756287,
      "learning_rate": 6.4461538461538465e-06,
      "loss": 0.0037,
      "step": 2643
    },
    {
      "epoch": 33.78913738019169,
      "grad_norm": 4.92404317855835,
      "learning_rate": 6.441025641025641e-06,
      "loss": 0.0244,
      "step": 2644
    },
    {
      "epoch": 33.801916932907346,
      "grad_norm": 3.1654813289642334,
      "learning_rate": 6.435897435897437e-06,
      "loss": 0.017,
      "step": 2645
    },
    {
      "epoch": 33.814696485623,
      "grad_norm": 0.5431066155433655,
      "learning_rate": 6.430769230769231e-06,
      "loss": 0.0051,
      "step": 2646
    },
    {
      "epoch": 33.82747603833866,
      "grad_norm": 0.9253678917884827,
      "learning_rate": 6.425641025641026e-06,
      "loss": 0.0075,
      "step": 2647
    },
    {
      "epoch": 33.840255591054316,
      "grad_norm": 1.2948482036590576,
      "learning_rate": 6.420512820512821e-06,
      "loss": 0.0074,
      "step": 2648
    },
    {
      "epoch": 33.853035143769965,
      "grad_norm": 1.8760825395584106,
      "learning_rate": 6.415384615384616e-06,
      "loss": 0.01,
      "step": 2649
    },
    {
      "epoch": 33.86581469648562,
      "grad_norm": 0.7007573246955872,
      "learning_rate": 6.410256410256412e-06,
      "loss": 0.0053,
      "step": 2650
    },
    {
      "epoch": 33.87859424920128,
      "grad_norm": 0.7296006083488464,
      "learning_rate": 6.405128205128206e-06,
      "loss": 0.005,
      "step": 2651
    },
    {
      "epoch": 33.891373801916934,
      "grad_norm": 0.9634267091751099,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.0076,
      "step": 2652
    },
    {
      "epoch": 33.90415335463259,
      "grad_norm": 2.8756632804870605,
      "learning_rate": 6.3948717948717955e-06,
      "loss": 0.0147,
      "step": 2653
    },
    {
      "epoch": 33.91693290734824,
      "grad_norm": 1.979262351989746,
      "learning_rate": 6.38974358974359e-06,
      "loss": 0.0205,
      "step": 2654
    },
    {
      "epoch": 33.9297124600639,
      "grad_norm": 1.4359487295150757,
      "learning_rate": 6.384615384615384e-06,
      "loss": 0.0085,
      "step": 2655
    },
    {
      "epoch": 33.94249201277955,
      "grad_norm": 0.5436068773269653,
      "learning_rate": 6.37948717948718e-06,
      "loss": 0.004,
      "step": 2656
    },
    {
      "epoch": 33.95527156549521,
      "grad_norm": 0.5183521509170532,
      "learning_rate": 6.374358974358975e-06,
      "loss": 0.0042,
      "step": 2657
    },
    {
      "epoch": 33.968051118210866,
      "grad_norm": 0.49053263664245605,
      "learning_rate": 6.36923076923077e-06,
      "loss": 0.0046,
      "step": 2658
    },
    {
      "epoch": 33.980830670926515,
      "grad_norm": 4.7706499099731445,
      "learning_rate": 6.364102564102565e-06,
      "loss": 0.0579,
      "step": 2659
    },
    {
      "epoch": 33.99361022364217,
      "grad_norm": 0.7718542814254761,
      "learning_rate": 6.358974358974359e-06,
      "loss": 0.0053,
      "step": 2660
    },
    {
      "epoch": 34.00638977635783,
      "grad_norm": 0.5025060176849365,
      "learning_rate": 6.353846153846155e-06,
      "loss": 0.0053,
      "step": 2661
    },
    {
      "epoch": 34.019169329073485,
      "grad_norm": 2.5457983016967773,
      "learning_rate": 6.3487179487179495e-06,
      "loss": 0.0118,
      "step": 2662
    },
    {
      "epoch": 34.031948881789134,
      "grad_norm": 0.7873379588127136,
      "learning_rate": 6.3435897435897444e-06,
      "loss": 0.0051,
      "step": 2663
    },
    {
      "epoch": 34.04472843450479,
      "grad_norm": 0.3507178723812103,
      "learning_rate": 6.3384615384615385e-06,
      "loss": 0.0038,
      "step": 2664
    },
    {
      "epoch": 34.05750798722045,
      "grad_norm": 1.270276665687561,
      "learning_rate": 6.333333333333333e-06,
      "loss": 0.0061,
      "step": 2665
    },
    {
      "epoch": 34.0702875399361,
      "grad_norm": 1.035638689994812,
      "learning_rate": 6.328205128205128e-06,
      "loss": 0.0065,
      "step": 2666
    },
    {
      "epoch": 34.08306709265176,
      "grad_norm": 0.6638619303703308,
      "learning_rate": 6.323076923076924e-06,
      "loss": 0.0061,
      "step": 2667
    },
    {
      "epoch": 34.09584664536741,
      "grad_norm": 1.7834850549697876,
      "learning_rate": 6.317948717948719e-06,
      "loss": 0.0077,
      "step": 2668
    },
    {
      "epoch": 34.108626198083066,
      "grad_norm": 0.4163866639137268,
      "learning_rate": 6.312820512820513e-06,
      "loss": 0.0042,
      "step": 2669
    },
    {
      "epoch": 34.12140575079872,
      "grad_norm": 3.6426656246185303,
      "learning_rate": 6.307692307692308e-06,
      "loss": 0.0245,
      "step": 2670
    },
    {
      "epoch": 34.13418530351438,
      "grad_norm": 0.363334983587265,
      "learning_rate": 6.302564102564103e-06,
      "loss": 0.0042,
      "step": 2671
    },
    {
      "epoch": 34.146964856230035,
      "grad_norm": 0.37875476479530334,
      "learning_rate": 6.2974358974358985e-06,
      "loss": 0.004,
      "step": 2672
    },
    {
      "epoch": 34.159744408945684,
      "grad_norm": 3.785109758377075,
      "learning_rate": 6.2923076923076934e-06,
      "loss": 0.0077,
      "step": 2673
    },
    {
      "epoch": 34.17252396166134,
      "grad_norm": 3.752488374710083,
      "learning_rate": 6.2871794871794875e-06,
      "loss": 0.0091,
      "step": 2674
    },
    {
      "epoch": 34.185303514377,
      "grad_norm": 0.7148936986923218,
      "learning_rate": 6.282051282051282e-06,
      "loss": 0.0071,
      "step": 2675
    },
    {
      "epoch": 34.198083067092654,
      "grad_norm": 2.916012763977051,
      "learning_rate": 6.276923076923077e-06,
      "loss": 0.0149,
      "step": 2676
    },
    {
      "epoch": 34.21086261980831,
      "grad_norm": 2.615849494934082,
      "learning_rate": 6.271794871794872e-06,
      "loss": 0.0114,
      "step": 2677
    },
    {
      "epoch": 34.22364217252396,
      "grad_norm": 0.3755146563053131,
      "learning_rate": 6.266666666666668e-06,
      "loss": 0.0045,
      "step": 2678
    },
    {
      "epoch": 34.236421725239616,
      "grad_norm": 1.1889039278030396,
      "learning_rate": 6.261538461538462e-06,
      "loss": 0.0083,
      "step": 2679
    },
    {
      "epoch": 34.24920127795527,
      "grad_norm": 0.518305242061615,
      "learning_rate": 6.256410256410257e-06,
      "loss": 0.0037,
      "step": 2680
    },
    {
      "epoch": 34.26198083067093,
      "grad_norm": 1.432346224784851,
      "learning_rate": 6.251282051282052e-06,
      "loss": 0.0112,
      "step": 2681
    },
    {
      "epoch": 34.27476038338658,
      "grad_norm": 0.7005894780158997,
      "learning_rate": 6.246153846153846e-06,
      "loss": 0.0056,
      "step": 2682
    },
    {
      "epoch": 34.287539936102235,
      "grad_norm": 0.5781270861625671,
      "learning_rate": 6.2410256410256424e-06,
      "loss": 0.0042,
      "step": 2683
    },
    {
      "epoch": 34.30031948881789,
      "grad_norm": 0.3330841660499573,
      "learning_rate": 6.2358974358974365e-06,
      "loss": 0.0028,
      "step": 2684
    },
    {
      "epoch": 34.31309904153355,
      "grad_norm": 4.505059242248535,
      "learning_rate": 6.230769230769231e-06,
      "loss": 0.0157,
      "step": 2685
    },
    {
      "epoch": 34.325878594249204,
      "grad_norm": 2.908541440963745,
      "learning_rate": 6.225641025641026e-06,
      "loss": 0.0185,
      "step": 2686
    },
    {
      "epoch": 34.33865814696485,
      "grad_norm": 1.3968557119369507,
      "learning_rate": 6.22051282051282e-06,
      "loss": 0.0059,
      "step": 2687
    },
    {
      "epoch": 34.35143769968051,
      "grad_norm": 0.5780673623085022,
      "learning_rate": 6.215384615384615e-06,
      "loss": 0.0059,
      "step": 2688
    },
    {
      "epoch": 34.364217252396166,
      "grad_norm": 0.33554375171661377,
      "learning_rate": 6.210256410256411e-06,
      "loss": 0.0038,
      "step": 2689
    },
    {
      "epoch": 34.37699680511182,
      "grad_norm": 0.6034390926361084,
      "learning_rate": 6.205128205128206e-06,
      "loss": 0.0052,
      "step": 2690
    },
    {
      "epoch": 34.38977635782748,
      "grad_norm": 0.8513287305831909,
      "learning_rate": 6.200000000000001e-06,
      "loss": 0.0052,
      "step": 2691
    },
    {
      "epoch": 34.40255591054313,
      "grad_norm": 0.47783803939819336,
      "learning_rate": 6.194871794871795e-06,
      "loss": 0.004,
      "step": 2692
    },
    {
      "epoch": 34.415335463258785,
      "grad_norm": 0.5561738014221191,
      "learning_rate": 6.18974358974359e-06,
      "loss": 0.0057,
      "step": 2693
    },
    {
      "epoch": 34.42811501597444,
      "grad_norm": 0.8518915772438049,
      "learning_rate": 6.1846153846153855e-06,
      "loss": 0.0081,
      "step": 2694
    },
    {
      "epoch": 34.4408945686901,
      "grad_norm": 30.251968383789062,
      "learning_rate": 6.17948717948718e-06,
      "loss": 0.0125,
      "step": 2695
    },
    {
      "epoch": 34.453674121405754,
      "grad_norm": 0.7382756471633911,
      "learning_rate": 6.174358974358975e-06,
      "loss": 0.0057,
      "step": 2696
    },
    {
      "epoch": 34.466453674121404,
      "grad_norm": 0.9222352504730225,
      "learning_rate": 6.169230769230769e-06,
      "loss": 0.0057,
      "step": 2697
    },
    {
      "epoch": 34.47923322683706,
      "grad_norm": 1.679905891418457,
      "learning_rate": 6.164102564102564e-06,
      "loss": 0.007,
      "step": 2698
    },
    {
      "epoch": 34.49201277955272,
      "grad_norm": 0.7995689511299133,
      "learning_rate": 6.15897435897436e-06,
      "loss": 0.008,
      "step": 2699
    },
    {
      "epoch": 34.50479233226837,
      "grad_norm": 0.8102620840072632,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.0055,
      "step": 2700
    },
    {
      "epoch": 34.50479233226837,
      "eval_loss": 0.9436935186386108,
      "eval_runtime": 183.897,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 2700
    },
    {
      "epoch": 34.51757188498402,
      "grad_norm": 0.44937846064567566,
      "learning_rate": 6.14871794871795e-06,
      "loss": 0.0038,
      "step": 2701
    },
    {
      "epoch": 34.53035143769968,
      "grad_norm": 0.4699499309062958,
      "learning_rate": 6.143589743589744e-06,
      "loss": 0.0055,
      "step": 2702
    },
    {
      "epoch": 34.543130990415335,
      "grad_norm": 1.1993327140808105,
      "learning_rate": 6.138461538461539e-06,
      "loss": 0.0065,
      "step": 2703
    },
    {
      "epoch": 34.55591054313099,
      "grad_norm": 0.4858746826648712,
      "learning_rate": 6.133333333333334e-06,
      "loss": 0.0041,
      "step": 2704
    },
    {
      "epoch": 34.56869009584665,
      "grad_norm": 0.6977940797805786,
      "learning_rate": 6.128205128205129e-06,
      "loss": 0.0051,
      "step": 2705
    },
    {
      "epoch": 34.5814696485623,
      "grad_norm": 0.36483317613601685,
      "learning_rate": 6.123076923076923e-06,
      "loss": 0.0032,
      "step": 2706
    },
    {
      "epoch": 34.594249201277954,
      "grad_norm": 0.4539944529533386,
      "learning_rate": 6.117948717948718e-06,
      "loss": 0.0049,
      "step": 2707
    },
    {
      "epoch": 34.60702875399361,
      "grad_norm": 1.643522024154663,
      "learning_rate": 6.112820512820513e-06,
      "loss": 0.0091,
      "step": 2708
    },
    {
      "epoch": 34.61980830670927,
      "grad_norm": 0.47860094904899597,
      "learning_rate": 6.107692307692308e-06,
      "loss": 0.0042,
      "step": 2709
    },
    {
      "epoch": 34.63258785942492,
      "grad_norm": 1.3490616083145142,
      "learning_rate": 6.102564102564104e-06,
      "loss": 0.0091,
      "step": 2710
    },
    {
      "epoch": 34.64536741214057,
      "grad_norm": 3.0723254680633545,
      "learning_rate": 6.097435897435898e-06,
      "loss": 0.0131,
      "step": 2711
    },
    {
      "epoch": 34.65814696485623,
      "grad_norm": 0.4609510600566864,
      "learning_rate": 6.092307692307693e-06,
      "loss": 0.0038,
      "step": 2712
    },
    {
      "epoch": 34.670926517571885,
      "grad_norm": 0.3855610489845276,
      "learning_rate": 6.087179487179488e-06,
      "loss": 0.0038,
      "step": 2713
    },
    {
      "epoch": 34.68370607028754,
      "grad_norm": 3.6798195838928223,
      "learning_rate": 6.082051282051283e-06,
      "loss": 0.0134,
      "step": 2714
    },
    {
      "epoch": 34.6964856230032,
      "grad_norm": 0.545027494430542,
      "learning_rate": 6.076923076923077e-06,
      "loss": 0.0054,
      "step": 2715
    },
    {
      "epoch": 34.70926517571885,
      "grad_norm": 1.6062592267990112,
      "learning_rate": 6.071794871794872e-06,
      "loss": 0.0174,
      "step": 2716
    },
    {
      "epoch": 34.722044728434504,
      "grad_norm": 0.5605974197387695,
      "learning_rate": 6.066666666666667e-06,
      "loss": 0.005,
      "step": 2717
    },
    {
      "epoch": 34.73482428115016,
      "grad_norm": 0.6980181932449341,
      "learning_rate": 6.061538461538462e-06,
      "loss": 0.0059,
      "step": 2718
    },
    {
      "epoch": 34.74760383386582,
      "grad_norm": 0.5648006796836853,
      "learning_rate": 6.056410256410257e-06,
      "loss": 0.0054,
      "step": 2719
    },
    {
      "epoch": 34.760383386581466,
      "grad_norm": 0.6961598992347717,
      "learning_rate": 6.051282051282051e-06,
      "loss": 0.0052,
      "step": 2720
    },
    {
      "epoch": 34.77316293929712,
      "grad_norm": 0.5445839166641235,
      "learning_rate": 6.046153846153847e-06,
      "loss": 0.0053,
      "step": 2721
    },
    {
      "epoch": 34.78594249201278,
      "grad_norm": 0.4801168143749237,
      "learning_rate": 6.041025641025642e-06,
      "loss": 0.0054,
      "step": 2722
    },
    {
      "epoch": 34.798722044728436,
      "grad_norm": 0.3502941429615021,
      "learning_rate": 6.035897435897437e-06,
      "loss": 0.0031,
      "step": 2723
    },
    {
      "epoch": 34.81150159744409,
      "grad_norm": 2.878469228744507,
      "learning_rate": 6.030769230769231e-06,
      "loss": 0.0258,
      "step": 2724
    },
    {
      "epoch": 34.82428115015974,
      "grad_norm": 1.5865216255187988,
      "learning_rate": 6.025641025641026e-06,
      "loss": 0.0127,
      "step": 2725
    },
    {
      "epoch": 34.8370607028754,
      "grad_norm": 1.8358023166656494,
      "learning_rate": 6.0205128205128206e-06,
      "loss": 0.0146,
      "step": 2726
    },
    {
      "epoch": 34.849840255591054,
      "grad_norm": 2.642219305038452,
      "learning_rate": 6.015384615384616e-06,
      "loss": 0.0126,
      "step": 2727
    },
    {
      "epoch": 34.86261980830671,
      "grad_norm": 2.9801673889160156,
      "learning_rate": 6.010256410256411e-06,
      "loss": 0.0151,
      "step": 2728
    },
    {
      "epoch": 34.87539936102237,
      "grad_norm": 2.9196527004241943,
      "learning_rate": 6.005128205128205e-06,
      "loss": 0.012,
      "step": 2729
    },
    {
      "epoch": 34.88817891373802,
      "grad_norm": 1.0630669593811035,
      "learning_rate": 6e-06,
      "loss": 0.0089,
      "step": 2730
    },
    {
      "epoch": 34.90095846645367,
      "grad_norm": 0.7134859561920166,
      "learning_rate": 5.994871794871795e-06,
      "loss": 0.0044,
      "step": 2731
    },
    {
      "epoch": 34.91373801916933,
      "grad_norm": 3.1851651668548584,
      "learning_rate": 5.989743589743591e-06,
      "loss": 0.0175,
      "step": 2732
    },
    {
      "epoch": 34.926517571884986,
      "grad_norm": 0.4819214642047882,
      "learning_rate": 5.984615384615386e-06,
      "loss": 0.0046,
      "step": 2733
    },
    {
      "epoch": 34.93929712460064,
      "grad_norm": 1.816530466079712,
      "learning_rate": 5.97948717948718e-06,
      "loss": 0.0157,
      "step": 2734
    },
    {
      "epoch": 34.95207667731629,
      "grad_norm": 0.5850124955177307,
      "learning_rate": 5.974358974358975e-06,
      "loss": 0.0073,
      "step": 2735
    },
    {
      "epoch": 34.96485623003195,
      "grad_norm": 8.858750343322754,
      "learning_rate": 5.9692307692307695e-06,
      "loss": 0.0335,
      "step": 2736
    },
    {
      "epoch": 34.977635782747605,
      "grad_norm": 2.54228138923645,
      "learning_rate": 5.9641025641025644e-06,
      "loss": 0.0212,
      "step": 2737
    },
    {
      "epoch": 34.99041533546326,
      "grad_norm": 0.37504932284355164,
      "learning_rate": 5.95897435897436e-06,
      "loss": 0.0038,
      "step": 2738
    },
    {
      "epoch": 35.00319488817891,
      "grad_norm": 0.5887596011161804,
      "learning_rate": 5.953846153846154e-06,
      "loss": 0.0049,
      "step": 2739
    },
    {
      "epoch": 35.01597444089457,
      "grad_norm": 0.47470083832740784,
      "learning_rate": 5.948717948717949e-06,
      "loss": 0.0035,
      "step": 2740
    },
    {
      "epoch": 35.02875399361022,
      "grad_norm": 1.7007302045822144,
      "learning_rate": 5.943589743589744e-06,
      "loss": 0.0162,
      "step": 2741
    },
    {
      "epoch": 35.04153354632588,
      "grad_norm": 0.7021458745002747,
      "learning_rate": 5.938461538461538e-06,
      "loss": 0.0129,
      "step": 2742
    },
    {
      "epoch": 35.054313099041536,
      "grad_norm": 3.1543846130371094,
      "learning_rate": 5.933333333333335e-06,
      "loss": 0.0175,
      "step": 2743
    },
    {
      "epoch": 35.067092651757186,
      "grad_norm": 10.487667083740234,
      "learning_rate": 5.928205128205129e-06,
      "loss": 0.0227,
      "step": 2744
    },
    {
      "epoch": 35.07987220447284,
      "grad_norm": 0.6180176734924316,
      "learning_rate": 5.923076923076924e-06,
      "loss": 0.0052,
      "step": 2745
    },
    {
      "epoch": 35.0926517571885,
      "grad_norm": 0.45216065645217896,
      "learning_rate": 5.9179487179487185e-06,
      "loss": 0.0049,
      "step": 2746
    },
    {
      "epoch": 35.105431309904155,
      "grad_norm": 0.40809720754623413,
      "learning_rate": 5.912820512820513e-06,
      "loss": 0.0043,
      "step": 2747
    },
    {
      "epoch": 35.11821086261981,
      "grad_norm": 1.078952670097351,
      "learning_rate": 5.907692307692308e-06,
      "loss": 0.0061,
      "step": 2748
    },
    {
      "epoch": 35.13099041533546,
      "grad_norm": 0.625017523765564,
      "learning_rate": 5.902564102564103e-06,
      "loss": 0.0045,
      "step": 2749
    },
    {
      "epoch": 35.14376996805112,
      "grad_norm": 0.6739218235015869,
      "learning_rate": 5.897435897435898e-06,
      "loss": 0.0057,
      "step": 2750
    },
    {
      "epoch": 35.156549520766774,
      "grad_norm": 0.4799802899360657,
      "learning_rate": 5.892307692307693e-06,
      "loss": 0.0027,
      "step": 2751
    },
    {
      "epoch": 35.16932907348243,
      "grad_norm": 3.5189943313598633,
      "learning_rate": 5.887179487179487e-06,
      "loss": 0.0168,
      "step": 2752
    },
    {
      "epoch": 35.18210862619808,
      "grad_norm": 3.1091294288635254,
      "learning_rate": 5.882051282051282e-06,
      "loss": 0.0276,
      "step": 2753
    },
    {
      "epoch": 35.194888178913736,
      "grad_norm": 3.039184331893921,
      "learning_rate": 5.876923076923078e-06,
      "loss": 0.0174,
      "step": 2754
    },
    {
      "epoch": 35.20766773162939,
      "grad_norm": 0.737232506275177,
      "learning_rate": 5.871794871794873e-06,
      "loss": 0.0067,
      "step": 2755
    },
    {
      "epoch": 35.22044728434505,
      "grad_norm": 1.1661545038223267,
      "learning_rate": 5.8666666666666675e-06,
      "loss": 0.005,
      "step": 2756
    },
    {
      "epoch": 35.233226837060705,
      "grad_norm": 2.1293647289276123,
      "learning_rate": 5.861538461538462e-06,
      "loss": 0.0107,
      "step": 2757
    },
    {
      "epoch": 35.246006389776355,
      "grad_norm": 0.275880366563797,
      "learning_rate": 5.8564102564102565e-06,
      "loss": 0.0026,
      "step": 2758
    },
    {
      "epoch": 35.25878594249201,
      "grad_norm": 0.5937846899032593,
      "learning_rate": 5.851282051282052e-06,
      "loss": 0.0045,
      "step": 2759
    },
    {
      "epoch": 35.27156549520767,
      "grad_norm": 0.37913593649864197,
      "learning_rate": 5.846153846153847e-06,
      "loss": 0.0039,
      "step": 2760
    },
    {
      "epoch": 35.284345047923324,
      "grad_norm": 0.5930600166320801,
      "learning_rate": 5.841025641025642e-06,
      "loss": 0.0052,
      "step": 2761
    },
    {
      "epoch": 35.29712460063898,
      "grad_norm": 0.6751174330711365,
      "learning_rate": 5.835897435897436e-06,
      "loss": 0.0048,
      "step": 2762
    },
    {
      "epoch": 35.30990415335463,
      "grad_norm": 0.31900155544281006,
      "learning_rate": 5.830769230769231e-06,
      "loss": 0.003,
      "step": 2763
    },
    {
      "epoch": 35.322683706070286,
      "grad_norm": 0.5694495439529419,
      "learning_rate": 5.825641025641026e-06,
      "loss": 0.0046,
      "step": 2764
    },
    {
      "epoch": 35.33546325878594,
      "grad_norm": 1.1516863107681274,
      "learning_rate": 5.820512820512822e-06,
      "loss": 0.0046,
      "step": 2765
    },
    {
      "epoch": 35.3482428115016,
      "grad_norm": 2.9431941509246826,
      "learning_rate": 5.815384615384616e-06,
      "loss": 0.0214,
      "step": 2766
    },
    {
      "epoch": 35.361022364217256,
      "grad_norm": 2.029663562774658,
      "learning_rate": 5.8102564102564106e-06,
      "loss": 0.0157,
      "step": 2767
    },
    {
      "epoch": 35.373801916932905,
      "grad_norm": 0.2581423223018646,
      "learning_rate": 5.8051282051282055e-06,
      "loss": 0.0022,
      "step": 2768
    },
    {
      "epoch": 35.38658146964856,
      "grad_norm": 0.3771861791610718,
      "learning_rate": 5.8e-06,
      "loss": 0.003,
      "step": 2769
    },
    {
      "epoch": 35.39936102236422,
      "grad_norm": 0.69022536277771,
      "learning_rate": 5.794871794871796e-06,
      "loss": 0.0048,
      "step": 2770
    },
    {
      "epoch": 35.412140575079874,
      "grad_norm": 2.675891637802124,
      "learning_rate": 5.78974358974359e-06,
      "loss": 0.0229,
      "step": 2771
    },
    {
      "epoch": 35.424920127795524,
      "grad_norm": 0.2880118489265442,
      "learning_rate": 5.784615384615385e-06,
      "loss": 0.003,
      "step": 2772
    },
    {
      "epoch": 35.43769968051118,
      "grad_norm": 0.38748475909233093,
      "learning_rate": 5.77948717948718e-06,
      "loss": 0.0043,
      "step": 2773
    },
    {
      "epoch": 35.45047923322684,
      "grad_norm": 2.8036811351776123,
      "learning_rate": 5.774358974358975e-06,
      "loss": 0.0183,
      "step": 2774
    },
    {
      "epoch": 35.46325878594249,
      "grad_norm": 1.2062289714813232,
      "learning_rate": 5.769230769230769e-06,
      "loss": 0.0097,
      "step": 2775
    },
    {
      "epoch": 35.47603833865815,
      "grad_norm": 1.0259376764297485,
      "learning_rate": 5.764102564102565e-06,
      "loss": 0.0075,
      "step": 2776
    },
    {
      "epoch": 35.4888178913738,
      "grad_norm": 0.43436387181282043,
      "learning_rate": 5.7589743589743596e-06,
      "loss": 0.005,
      "step": 2777
    },
    {
      "epoch": 35.501597444089455,
      "grad_norm": 0.4108300507068634,
      "learning_rate": 5.7538461538461545e-06,
      "loss": 0.0039,
      "step": 2778
    },
    {
      "epoch": 35.51437699680511,
      "grad_norm": 0.5943469405174255,
      "learning_rate": 5.748717948717949e-06,
      "loss": 0.0051,
      "step": 2779
    },
    {
      "epoch": 35.52715654952077,
      "grad_norm": 0.397146075963974,
      "learning_rate": 5.743589743589743e-06,
      "loss": 0.0039,
      "step": 2780
    },
    {
      "epoch": 35.539936102236425,
      "grad_norm": 2.4300386905670166,
      "learning_rate": 5.738461538461539e-06,
      "loss": 0.0132,
      "step": 2781
    },
    {
      "epoch": 35.552715654952074,
      "grad_norm": 0.35871580243110657,
      "learning_rate": 5.733333333333334e-06,
      "loss": 0.0039,
      "step": 2782
    },
    {
      "epoch": 35.56549520766773,
      "grad_norm": 1.9119770526885986,
      "learning_rate": 5.728205128205129e-06,
      "loss": 0.0137,
      "step": 2783
    },
    {
      "epoch": 35.57827476038339,
      "grad_norm": 0.35279327630996704,
      "learning_rate": 5.723076923076923e-06,
      "loss": 0.0035,
      "step": 2784
    },
    {
      "epoch": 35.59105431309904,
      "grad_norm": 3.315377712249756,
      "learning_rate": 5.717948717948718e-06,
      "loss": 0.0195,
      "step": 2785
    },
    {
      "epoch": 35.6038338658147,
      "grad_norm": 0.4838240444660187,
      "learning_rate": 5.712820512820513e-06,
      "loss": 0.004,
      "step": 2786
    },
    {
      "epoch": 35.61661341853035,
      "grad_norm": 0.40503624081611633,
      "learning_rate": 5.7076923076923086e-06,
      "loss": 0.0033,
      "step": 2787
    },
    {
      "epoch": 35.629392971246006,
      "grad_norm": 0.5054441690444946,
      "learning_rate": 5.7025641025641035e-06,
      "loss": 0.005,
      "step": 2788
    },
    {
      "epoch": 35.64217252396166,
      "grad_norm": 0.4412970542907715,
      "learning_rate": 5.6974358974358975e-06,
      "loss": 0.0033,
      "step": 2789
    },
    {
      "epoch": 35.65495207667732,
      "grad_norm": 0.36365410685539246,
      "learning_rate": 5.692307692307692e-06,
      "loss": 0.0036,
      "step": 2790
    },
    {
      "epoch": 35.66773162939297,
      "grad_norm": 0.5414665937423706,
      "learning_rate": 5.687179487179487e-06,
      "loss": 0.0055,
      "step": 2791
    },
    {
      "epoch": 35.680511182108624,
      "grad_norm": 2.9094033241271973,
      "learning_rate": 5.682051282051283e-06,
      "loss": 0.0174,
      "step": 2792
    },
    {
      "epoch": 35.69329073482428,
      "grad_norm": 0.43583613634109497,
      "learning_rate": 5.676923076923078e-06,
      "loss": 0.0038,
      "step": 2793
    },
    {
      "epoch": 35.70607028753994,
      "grad_norm": 0.3606933653354645,
      "learning_rate": 5.671794871794872e-06,
      "loss": 0.0035,
      "step": 2794
    },
    {
      "epoch": 35.718849840255594,
      "grad_norm": 1.6921635866165161,
      "learning_rate": 5.666666666666667e-06,
      "loss": 0.0112,
      "step": 2795
    },
    {
      "epoch": 35.73162939297124,
      "grad_norm": 0.2892257869243622,
      "learning_rate": 5.661538461538462e-06,
      "loss": 0.0037,
      "step": 2796
    },
    {
      "epoch": 35.7444089456869,
      "grad_norm": 0.5619503259658813,
      "learning_rate": 5.6564102564102575e-06,
      "loss": 0.0041,
      "step": 2797
    },
    {
      "epoch": 35.757188498402556,
      "grad_norm": 0.4655962884426117,
      "learning_rate": 5.6512820512820524e-06,
      "loss": 0.0047,
      "step": 2798
    },
    {
      "epoch": 35.76996805111821,
      "grad_norm": 0.2837389409542084,
      "learning_rate": 5.6461538461538465e-06,
      "loss": 0.0035,
      "step": 2799
    },
    {
      "epoch": 35.78274760383387,
      "grad_norm": 0.3791850805282593,
      "learning_rate": 5.641025641025641e-06,
      "loss": 0.0028,
      "step": 2800
    },
    {
      "epoch": 35.78274760383387,
      "eval_loss": 0.9578615427017212,
      "eval_runtime": 183.8282,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 2800
    },
    {
      "epoch": 35.79552715654952,
      "grad_norm": 1.7197462320327759,
      "learning_rate": 5.635897435897436e-06,
      "loss": 0.0053,
      "step": 2801
    },
    {
      "epoch": 35.808306709265175,
      "grad_norm": 0.36914217472076416,
      "learning_rate": 5.63076923076923e-06,
      "loss": 0.0036,
      "step": 2802
    },
    {
      "epoch": 35.82108626198083,
      "grad_norm": 0.648004412651062,
      "learning_rate": 5.625641025641027e-06,
      "loss": 0.0044,
      "step": 2803
    },
    {
      "epoch": 35.83386581469649,
      "grad_norm": 1.7772631645202637,
      "learning_rate": 5.620512820512821e-06,
      "loss": 0.0122,
      "step": 2804
    },
    {
      "epoch": 35.846645367412144,
      "grad_norm": 0.5996457934379578,
      "learning_rate": 5.615384615384616e-06,
      "loss": 0.0042,
      "step": 2805
    },
    {
      "epoch": 35.85942492012779,
      "grad_norm": 0.5983651280403137,
      "learning_rate": 5.610256410256411e-06,
      "loss": 0.0049,
      "step": 2806
    },
    {
      "epoch": 35.87220447284345,
      "grad_norm": 0.9759971499443054,
      "learning_rate": 5.605128205128205e-06,
      "loss": 0.0057,
      "step": 2807
    },
    {
      "epoch": 35.884984025559106,
      "grad_norm": 5.0989203453063965,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.032,
      "step": 2808
    },
    {
      "epoch": 35.89776357827476,
      "grad_norm": 3.5177414417266846,
      "learning_rate": 5.5948717948717955e-06,
      "loss": 0.016,
      "step": 2809
    },
    {
      "epoch": 35.91054313099041,
      "grad_norm": 0.5371911525726318,
      "learning_rate": 5.58974358974359e-06,
      "loss": 0.0057,
      "step": 2810
    },
    {
      "epoch": 35.92332268370607,
      "grad_norm": 0.5497738718986511,
      "learning_rate": 5.584615384615385e-06,
      "loss": 0.0045,
      "step": 2811
    },
    {
      "epoch": 35.936102236421725,
      "grad_norm": 1.1579020023345947,
      "learning_rate": 5.579487179487179e-06,
      "loss": 0.0111,
      "step": 2812
    },
    {
      "epoch": 35.94888178913738,
      "grad_norm": 0.5208929181098938,
      "learning_rate": 5.574358974358974e-06,
      "loss": 0.0041,
      "step": 2813
    },
    {
      "epoch": 35.96166134185304,
      "grad_norm": 0.4912267327308655,
      "learning_rate": 5.56923076923077e-06,
      "loss": 0.0042,
      "step": 2814
    },
    {
      "epoch": 35.97444089456869,
      "grad_norm": 0.3925940990447998,
      "learning_rate": 5.564102564102565e-06,
      "loss": 0.0034,
      "step": 2815
    },
    {
      "epoch": 35.98722044728434,
      "grad_norm": 0.5372114777565002,
      "learning_rate": 5.55897435897436e-06,
      "loss": 0.0035,
      "step": 2816
    },
    {
      "epoch": 36.0,
      "grad_norm": 0.3666168749332428,
      "learning_rate": 5.553846153846154e-06,
      "loss": 0.005,
      "step": 2817
    },
    {
      "epoch": 36.01277955271566,
      "grad_norm": 0.322618305683136,
      "learning_rate": 5.548717948717949e-06,
      "loss": 0.0034,
      "step": 2818
    },
    {
      "epoch": 36.02555910543131,
      "grad_norm": 0.4605313539505005,
      "learning_rate": 5.5435897435897445e-06,
      "loss": 0.0041,
      "step": 2819
    },
    {
      "epoch": 36.03833865814696,
      "grad_norm": 0.3681679666042328,
      "learning_rate": 5.538461538461539e-06,
      "loss": 0.003,
      "step": 2820
    },
    {
      "epoch": 36.05111821086262,
      "grad_norm": 10.06080436706543,
      "learning_rate": 5.533333333333334e-06,
      "loss": 0.0087,
      "step": 2821
    },
    {
      "epoch": 36.063897763578275,
      "grad_norm": 1.4203732013702393,
      "learning_rate": 5.528205128205128e-06,
      "loss": 0.006,
      "step": 2822
    },
    {
      "epoch": 36.07667731629393,
      "grad_norm": 0.5547159314155579,
      "learning_rate": 5.523076923076923e-06,
      "loss": 0.0038,
      "step": 2823
    },
    {
      "epoch": 36.08945686900959,
      "grad_norm": 0.4133658707141876,
      "learning_rate": 5.517948717948718e-06,
      "loss": 0.0034,
      "step": 2824
    },
    {
      "epoch": 36.10223642172524,
      "grad_norm": 0.4233897030353546,
      "learning_rate": 5.512820512820514e-06,
      "loss": 0.0036,
      "step": 2825
    },
    {
      "epoch": 36.115015974440894,
      "grad_norm": 2.718273639678955,
      "learning_rate": 5.507692307692308e-06,
      "loss": 0.0177,
      "step": 2826
    },
    {
      "epoch": 36.12779552715655,
      "grad_norm": 3.14428448677063,
      "learning_rate": 5.502564102564103e-06,
      "loss": 0.0135,
      "step": 2827
    },
    {
      "epoch": 36.14057507987221,
      "grad_norm": 0.31426680088043213,
      "learning_rate": 5.497435897435898e-06,
      "loss": 0.003,
      "step": 2828
    },
    {
      "epoch": 36.153354632587856,
      "grad_norm": 0.35171377658843994,
      "learning_rate": 5.492307692307693e-06,
      "loss": 0.0047,
      "step": 2829
    },
    {
      "epoch": 36.16613418530351,
      "grad_norm": 0.45544207096099854,
      "learning_rate": 5.487179487179488e-06,
      "loss": 0.0037,
      "step": 2830
    },
    {
      "epoch": 36.17891373801917,
      "grad_norm": 2.908658027648926,
      "learning_rate": 5.4820512820512824e-06,
      "loss": 0.0063,
      "step": 2831
    },
    {
      "epoch": 36.191693290734825,
      "grad_norm": 0.33017784357070923,
      "learning_rate": 5.476923076923077e-06,
      "loss": 0.0043,
      "step": 2832
    },
    {
      "epoch": 36.20447284345048,
      "grad_norm": 0.31544455885887146,
      "learning_rate": 5.471794871794872e-06,
      "loss": 0.0039,
      "step": 2833
    },
    {
      "epoch": 36.21725239616613,
      "grad_norm": 0.3912992477416992,
      "learning_rate": 5.466666666666667e-06,
      "loss": 0.004,
      "step": 2834
    },
    {
      "epoch": 36.23003194888179,
      "grad_norm": 1.5009045600891113,
      "learning_rate": 5.461538461538461e-06,
      "loss": 0.009,
      "step": 2835
    },
    {
      "epoch": 36.242811501597444,
      "grad_norm": 0.5826500654220581,
      "learning_rate": 5.456410256410257e-06,
      "loss": 0.0028,
      "step": 2836
    },
    {
      "epoch": 36.2555910543131,
      "grad_norm": 0.47322073578834534,
      "learning_rate": 5.451282051282052e-06,
      "loss": 0.0035,
      "step": 2837
    },
    {
      "epoch": 36.26837060702876,
      "grad_norm": 1.2564668655395508,
      "learning_rate": 5.446153846153847e-06,
      "loss": 0.0035,
      "step": 2838
    },
    {
      "epoch": 36.281150159744406,
      "grad_norm": 3.0140414237976074,
      "learning_rate": 5.441025641025642e-06,
      "loss": 0.0139,
      "step": 2839
    },
    {
      "epoch": 36.29392971246006,
      "grad_norm": 0.25751781463623047,
      "learning_rate": 5.435897435897436e-06,
      "loss": 0.0026,
      "step": 2840
    },
    {
      "epoch": 36.30670926517572,
      "grad_norm": 0.3114520311355591,
      "learning_rate": 5.430769230769231e-06,
      "loss": 0.0043,
      "step": 2841
    },
    {
      "epoch": 36.319488817891376,
      "grad_norm": 2.1022861003875732,
      "learning_rate": 5.425641025641026e-06,
      "loss": 0.0088,
      "step": 2842
    },
    {
      "epoch": 36.33226837060703,
      "grad_norm": 1.3880363702774048,
      "learning_rate": 5.420512820512821e-06,
      "loss": 0.0071,
      "step": 2843
    },
    {
      "epoch": 36.34504792332268,
      "grad_norm": 0.2697431743144989,
      "learning_rate": 5.415384615384615e-06,
      "loss": 0.0023,
      "step": 2844
    },
    {
      "epoch": 36.35782747603834,
      "grad_norm": 2.0496296882629395,
      "learning_rate": 5.41025641025641e-06,
      "loss": 0.0084,
      "step": 2845
    },
    {
      "epoch": 36.370607028753994,
      "grad_norm": 0.378528892993927,
      "learning_rate": 5.405128205128205e-06,
      "loss": 0.0042,
      "step": 2846
    },
    {
      "epoch": 36.38338658146965,
      "grad_norm": 0.2853531539440155,
      "learning_rate": 5.400000000000001e-06,
      "loss": 0.0035,
      "step": 2847
    },
    {
      "epoch": 36.3961661341853,
      "grad_norm": 0.3787241578102112,
      "learning_rate": 5.394871794871796e-06,
      "loss": 0.0046,
      "step": 2848
    },
    {
      "epoch": 36.40894568690096,
      "grad_norm": 2.358002185821533,
      "learning_rate": 5.38974358974359e-06,
      "loss": 0.0131,
      "step": 2849
    },
    {
      "epoch": 36.42172523961661,
      "grad_norm": 0.3563104271888733,
      "learning_rate": 5.384615384615385e-06,
      "loss": 0.0037,
      "step": 2850
    },
    {
      "epoch": 36.43450479233227,
      "grad_norm": 0.29914191365242004,
      "learning_rate": 5.3794871794871796e-06,
      "loss": 0.0026,
      "step": 2851
    },
    {
      "epoch": 36.447284345047926,
      "grad_norm": 0.7096664905548096,
      "learning_rate": 5.374358974358975e-06,
      "loss": 0.0036,
      "step": 2852
    },
    {
      "epoch": 36.460063897763575,
      "grad_norm": 0.6353800296783447,
      "learning_rate": 5.36923076923077e-06,
      "loss": 0.0043,
      "step": 2853
    },
    {
      "epoch": 36.47284345047923,
      "grad_norm": 0.3164236545562744,
      "learning_rate": 5.364102564102564e-06,
      "loss": 0.0025,
      "step": 2854
    },
    {
      "epoch": 36.48562300319489,
      "grad_norm": 2.6620049476623535,
      "learning_rate": 5.358974358974359e-06,
      "loss": 0.019,
      "step": 2855
    },
    {
      "epoch": 36.498402555910545,
      "grad_norm": 4.248519420623779,
      "learning_rate": 5.353846153846154e-06,
      "loss": 0.0128,
      "step": 2856
    },
    {
      "epoch": 36.5111821086262,
      "grad_norm": 0.45837897062301636,
      "learning_rate": 5.34871794871795e-06,
      "loss": 0.0043,
      "step": 2857
    },
    {
      "epoch": 36.52396166134185,
      "grad_norm": 0.44139301776885986,
      "learning_rate": 5.343589743589745e-06,
      "loss": 0.0036,
      "step": 2858
    },
    {
      "epoch": 36.53674121405751,
      "grad_norm": 0.7677149176597595,
      "learning_rate": 5.338461538461539e-06,
      "loss": 0.0068,
      "step": 2859
    },
    {
      "epoch": 36.54952076677316,
      "grad_norm": 0.5062174797058105,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0038,
      "step": 2860
    },
    {
      "epoch": 36.56230031948882,
      "grad_norm": 0.29120340943336487,
      "learning_rate": 5.3282051282051286e-06,
      "loss": 0.0034,
      "step": 2861
    },
    {
      "epoch": 36.575079872204476,
      "grad_norm": 0.27878037095069885,
      "learning_rate": 5.323076923076923e-06,
      "loss": 0.0027,
      "step": 2862
    },
    {
      "epoch": 36.587859424920126,
      "grad_norm": 3.050676107406616,
      "learning_rate": 5.317948717948719e-06,
      "loss": 0.0253,
      "step": 2863
    },
    {
      "epoch": 36.60063897763578,
      "grad_norm": 0.8370914459228516,
      "learning_rate": 5.312820512820513e-06,
      "loss": 0.0051,
      "step": 2864
    },
    {
      "epoch": 36.61341853035144,
      "grad_norm": 0.38411661982536316,
      "learning_rate": 5.307692307692308e-06,
      "loss": 0.0032,
      "step": 2865
    },
    {
      "epoch": 36.626198083067095,
      "grad_norm": 1.3328490257263184,
      "learning_rate": 5.302564102564103e-06,
      "loss": 0.0054,
      "step": 2866
    },
    {
      "epoch": 36.638977635782744,
      "grad_norm": 0.33674442768096924,
      "learning_rate": 5.297435897435897e-06,
      "loss": 0.0028,
      "step": 2867
    },
    {
      "epoch": 36.6517571884984,
      "grad_norm": 4.527010917663574,
      "learning_rate": 5.292307692307693e-06,
      "loss": 0.0184,
      "step": 2868
    },
    {
      "epoch": 36.66453674121406,
      "grad_norm": 0.3857632875442505,
      "learning_rate": 5.287179487179488e-06,
      "loss": 0.003,
      "step": 2869
    },
    {
      "epoch": 36.677316293929714,
      "grad_norm": 2.6783156394958496,
      "learning_rate": 5.282051282051283e-06,
      "loss": 0.0166,
      "step": 2870
    },
    {
      "epoch": 36.69009584664537,
      "grad_norm": 0.8151161670684814,
      "learning_rate": 5.2769230769230775e-06,
      "loss": 0.0036,
      "step": 2871
    },
    {
      "epoch": 36.70287539936102,
      "grad_norm": 2.524899959564209,
      "learning_rate": 5.271794871794872e-06,
      "loss": 0.023,
      "step": 2872
    },
    {
      "epoch": 36.715654952076676,
      "grad_norm": 1.1502883434295654,
      "learning_rate": 5.2666666666666665e-06,
      "loss": 0.0047,
      "step": 2873
    },
    {
      "epoch": 36.72843450479233,
      "grad_norm": 0.44396790862083435,
      "learning_rate": 5.261538461538462e-06,
      "loss": 0.0032,
      "step": 2874
    },
    {
      "epoch": 36.74121405750799,
      "grad_norm": 0.389405220746994,
      "learning_rate": 5.256410256410257e-06,
      "loss": 0.0027,
      "step": 2875
    },
    {
      "epoch": 36.753993610223645,
      "grad_norm": 0.5646959543228149,
      "learning_rate": 5.251282051282052e-06,
      "loss": 0.0048,
      "step": 2876
    },
    {
      "epoch": 36.766773162939295,
      "grad_norm": 1.245563268661499,
      "learning_rate": 5.246153846153846e-06,
      "loss": 0.0047,
      "step": 2877
    },
    {
      "epoch": 36.77955271565495,
      "grad_norm": 1.3920589685440063,
      "learning_rate": 5.241025641025641e-06,
      "loss": 0.0074,
      "step": 2878
    },
    {
      "epoch": 36.79233226837061,
      "grad_norm": 1.5733996629714966,
      "learning_rate": 5.235897435897437e-06,
      "loss": 0.0088,
      "step": 2879
    },
    {
      "epoch": 36.805111821086264,
      "grad_norm": 3.1889090538024902,
      "learning_rate": 5.230769230769232e-06,
      "loss": 0.0204,
      "step": 2880
    },
    {
      "epoch": 36.81789137380191,
      "grad_norm": 8.020994186401367,
      "learning_rate": 5.2256410256410265e-06,
      "loss": 0.0234,
      "step": 2881
    },
    {
      "epoch": 36.83067092651757,
      "grad_norm": 0.5541637539863586,
      "learning_rate": 5.220512820512821e-06,
      "loss": 0.0033,
      "step": 2882
    },
    {
      "epoch": 36.843450479233226,
      "grad_norm": 2.588081121444702,
      "learning_rate": 5.2153846153846155e-06,
      "loss": 0.016,
      "step": 2883
    },
    {
      "epoch": 36.85623003194888,
      "grad_norm": 4.180788040161133,
      "learning_rate": 5.21025641025641e-06,
      "loss": 0.0149,
      "step": 2884
    },
    {
      "epoch": 36.86900958466454,
      "grad_norm": 3.450303554534912,
      "learning_rate": 5.205128205128206e-06,
      "loss": 0.0214,
      "step": 2885
    },
    {
      "epoch": 36.88178913738019,
      "grad_norm": 3.1327109336853027,
      "learning_rate": 5.2e-06,
      "loss": 0.0166,
      "step": 2886
    },
    {
      "epoch": 36.894568690095845,
      "grad_norm": 0.4896284341812134,
      "learning_rate": 5.194871794871795e-06,
      "loss": 0.0033,
      "step": 2887
    },
    {
      "epoch": 36.9073482428115,
      "grad_norm": 0.5049822330474854,
      "learning_rate": 5.18974358974359e-06,
      "loss": 0.0041,
      "step": 2888
    },
    {
      "epoch": 36.92012779552716,
      "grad_norm": 2.475656747817993,
      "learning_rate": 5.184615384615385e-06,
      "loss": 0.0303,
      "step": 2889
    },
    {
      "epoch": 36.932907348242814,
      "grad_norm": 1.1910037994384766,
      "learning_rate": 5.179487179487181e-06,
      "loss": 0.0113,
      "step": 2890
    },
    {
      "epoch": 36.945686900958464,
      "grad_norm": 3.3851752281188965,
      "learning_rate": 5.174358974358975e-06,
      "loss": 0.0208,
      "step": 2891
    },
    {
      "epoch": 36.95846645367412,
      "grad_norm": 0.3671295940876007,
      "learning_rate": 5.16923076923077e-06,
      "loss": 0.0032,
      "step": 2892
    },
    {
      "epoch": 36.97124600638978,
      "grad_norm": 0.5597812533378601,
      "learning_rate": 5.1641025641025645e-06,
      "loss": 0.005,
      "step": 2893
    },
    {
      "epoch": 36.98402555910543,
      "grad_norm": 0.8427302837371826,
      "learning_rate": 5.158974358974359e-06,
      "loss": 0.0066,
      "step": 2894
    },
    {
      "epoch": 36.99680511182109,
      "grad_norm": 2.484133005142212,
      "learning_rate": 5.1538461538461534e-06,
      "loss": 0.018,
      "step": 2895
    },
    {
      "epoch": 37.00958466453674,
      "grad_norm": 0.325967401266098,
      "learning_rate": 5.148717948717949e-06,
      "loss": 0.0036,
      "step": 2896
    },
    {
      "epoch": 37.022364217252395,
      "grad_norm": 1.9392606019973755,
      "learning_rate": 5.143589743589744e-06,
      "loss": 0.0096,
      "step": 2897
    },
    {
      "epoch": 37.03514376996805,
      "grad_norm": 4.457958221435547,
      "learning_rate": 5.138461538461539e-06,
      "loss": 0.0136,
      "step": 2898
    },
    {
      "epoch": 37.04792332268371,
      "grad_norm": 0.3894760310649872,
      "learning_rate": 5.133333333333334e-06,
      "loss": 0.0033,
      "step": 2899
    },
    {
      "epoch": 37.06070287539936,
      "grad_norm": 0.23036618530750275,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.0022,
      "step": 2900
    },
    {
      "epoch": 37.06070287539936,
      "eval_loss": 0.9559409618377686,
      "eval_runtime": 183.503,
      "eval_samples_per_second": 0.856,
      "eval_steps_per_second": 0.109,
      "step": 2900
    },
    {
      "epoch": 37.073482428115014,
      "grad_norm": 0.328531950712204,
      "learning_rate": 5.123076923076924e-06,
      "loss": 0.0036,
      "step": 2901
    },
    {
      "epoch": 37.08626198083067,
      "grad_norm": 0.5235569477081299,
      "learning_rate": 5.1179487179487186e-06,
      "loss": 0.0055,
      "step": 2902
    },
    {
      "epoch": 37.09904153354633,
      "grad_norm": 0.3692654073238373,
      "learning_rate": 5.1128205128205135e-06,
      "loss": 0.0033,
      "step": 2903
    },
    {
      "epoch": 37.11182108626198,
      "grad_norm": 5.13992977142334,
      "learning_rate": 5.1076923076923075e-06,
      "loss": 0.0169,
      "step": 2904
    },
    {
      "epoch": 37.12460063897763,
      "grad_norm": 1.2046083211898804,
      "learning_rate": 5.1025641025641024e-06,
      "loss": 0.0066,
      "step": 2905
    },
    {
      "epoch": 37.13738019169329,
      "grad_norm": 1.247635006904602,
      "learning_rate": 5.097435897435898e-06,
      "loss": 0.0058,
      "step": 2906
    },
    {
      "epoch": 37.150159744408946,
      "grad_norm": 0.5680016875267029,
      "learning_rate": 5.092307692307693e-06,
      "loss": 0.0034,
      "step": 2907
    },
    {
      "epoch": 37.1629392971246,
      "grad_norm": 1.868544578552246,
      "learning_rate": 5.087179487179488e-06,
      "loss": 0.0071,
      "step": 2908
    },
    {
      "epoch": 37.17571884984026,
      "grad_norm": 0.21541500091552734,
      "learning_rate": 5.082051282051282e-06,
      "loss": 0.0021,
      "step": 2909
    },
    {
      "epoch": 37.18849840255591,
      "grad_norm": 0.38051384687423706,
      "learning_rate": 5.076923076923077e-06,
      "loss": 0.0032,
      "step": 2910
    },
    {
      "epoch": 37.201277955271564,
      "grad_norm": 1.883426547050476,
      "learning_rate": 5.071794871794872e-06,
      "loss": 0.0089,
      "step": 2911
    },
    {
      "epoch": 37.21405750798722,
      "grad_norm": 4.500365734100342,
      "learning_rate": 5.0666666666666676e-06,
      "loss": 0.0238,
      "step": 2912
    },
    {
      "epoch": 37.22683706070288,
      "grad_norm": 2.925161600112915,
      "learning_rate": 5.0615384615384625e-06,
      "loss": 0.0152,
      "step": 2913
    },
    {
      "epoch": 37.239616613418534,
      "grad_norm": 0.9527667760848999,
      "learning_rate": 5.0564102564102565e-06,
      "loss": 0.0047,
      "step": 2914
    },
    {
      "epoch": 37.25239616613418,
      "grad_norm": 0.31250178813934326,
      "learning_rate": 5.051282051282051e-06,
      "loss": 0.0028,
      "step": 2915
    },
    {
      "epoch": 37.26517571884984,
      "grad_norm": 0.2610889673233032,
      "learning_rate": 5.046153846153846e-06,
      "loss": 0.0022,
      "step": 2916
    },
    {
      "epoch": 37.277955271565496,
      "grad_norm": 0.3735126554965973,
      "learning_rate": 5.041025641025642e-06,
      "loss": 0.0035,
      "step": 2917
    },
    {
      "epoch": 37.29073482428115,
      "grad_norm": 0.30350491404533386,
      "learning_rate": 5.035897435897437e-06,
      "loss": 0.0028,
      "step": 2918
    },
    {
      "epoch": 37.3035143769968,
      "grad_norm": 3.4979629516601562,
      "learning_rate": 5.030769230769231e-06,
      "loss": 0.0192,
      "step": 2919
    },
    {
      "epoch": 37.31629392971246,
      "grad_norm": 0.3009140193462372,
      "learning_rate": 5.025641025641026e-06,
      "loss": 0.003,
      "step": 2920
    },
    {
      "epoch": 37.329073482428115,
      "grad_norm": 0.5664937496185303,
      "learning_rate": 5.020512820512821e-06,
      "loss": 0.0039,
      "step": 2921
    },
    {
      "epoch": 37.34185303514377,
      "grad_norm": 0.25941237807273865,
      "learning_rate": 5.015384615384616e-06,
      "loss": 0.003,
      "step": 2922
    },
    {
      "epoch": 37.35463258785943,
      "grad_norm": 0.35156363248825073,
      "learning_rate": 5.0102564102564115e-06,
      "loss": 0.0041,
      "step": 2923
    },
    {
      "epoch": 37.36741214057508,
      "grad_norm": 0.9193771481513977,
      "learning_rate": 5.0051282051282055e-06,
      "loss": 0.0053,
      "step": 2924
    },
    {
      "epoch": 37.38019169329073,
      "grad_norm": 0.4133692681789398,
      "learning_rate": 5e-06,
      "loss": 0.0048,
      "step": 2925
    },
    {
      "epoch": 37.39297124600639,
      "grad_norm": 4.615827560424805,
      "learning_rate": 4.994871794871795e-06,
      "loss": 0.0108,
      "step": 2926
    },
    {
      "epoch": 37.405750798722046,
      "grad_norm": 1.0765419006347656,
      "learning_rate": 4.98974358974359e-06,
      "loss": 0.0067,
      "step": 2927
    },
    {
      "epoch": 37.4185303514377,
      "grad_norm": 0.2866329252719879,
      "learning_rate": 4.984615384615385e-06,
      "loss": 0.0023,
      "step": 2928
    },
    {
      "epoch": 37.43130990415335,
      "grad_norm": 0.7527754902839661,
      "learning_rate": 4.97948717948718e-06,
      "loss": 0.003,
      "step": 2929
    },
    {
      "epoch": 37.44408945686901,
      "grad_norm": 2.4644179344177246,
      "learning_rate": 4.974358974358975e-06,
      "loss": 0.0119,
      "step": 2930
    },
    {
      "epoch": 37.456869009584665,
      "grad_norm": 0.17018601298332214,
      "learning_rate": 4.96923076923077e-06,
      "loss": 0.002,
      "step": 2931
    },
    {
      "epoch": 37.46964856230032,
      "grad_norm": 0.9139269590377808,
      "learning_rate": 4.964102564102565e-06,
      "loss": 0.006,
      "step": 2932
    },
    {
      "epoch": 37.48242811501598,
      "grad_norm": 0.37774431705474854,
      "learning_rate": 4.95897435897436e-06,
      "loss": 0.0036,
      "step": 2933
    },
    {
      "epoch": 37.49520766773163,
      "grad_norm": 0.31033891439437866,
      "learning_rate": 4.9538461538461545e-06,
      "loss": 0.0029,
      "step": 2934
    },
    {
      "epoch": 37.50798722044728,
      "grad_norm": 1.9837037324905396,
      "learning_rate": 4.948717948717949e-06,
      "loss": 0.023,
      "step": 2935
    },
    {
      "epoch": 37.52076677316294,
      "grad_norm": 0.4745427072048187,
      "learning_rate": 4.943589743589744e-06,
      "loss": 0.0037,
      "step": 2936
    },
    {
      "epoch": 37.533546325878596,
      "grad_norm": 0.38666072487831116,
      "learning_rate": 4.938461538461538e-06,
      "loss": 0.0038,
      "step": 2937
    },
    {
      "epoch": 37.546325878594246,
      "grad_norm": 0.3316282629966736,
      "learning_rate": 4.933333333333334e-06,
      "loss": 0.0044,
      "step": 2938
    },
    {
      "epoch": 37.5591054313099,
      "grad_norm": 0.5502758622169495,
      "learning_rate": 4.928205128205128e-06,
      "loss": 0.0035,
      "step": 2939
    },
    {
      "epoch": 37.57188498402556,
      "grad_norm": 3.197373867034912,
      "learning_rate": 4.923076923076924e-06,
      "loss": 0.0061,
      "step": 2940
    },
    {
      "epoch": 37.584664536741215,
      "grad_norm": 49.10783004760742,
      "learning_rate": 4.917948717948719e-06,
      "loss": 0.0113,
      "step": 2941
    },
    {
      "epoch": 37.59744408945687,
      "grad_norm": 3.1348283290863037,
      "learning_rate": 4.912820512820513e-06,
      "loss": 0.0161,
      "step": 2942
    },
    {
      "epoch": 37.61022364217252,
      "grad_norm": 0.2980424463748932,
      "learning_rate": 4.907692307692309e-06,
      "loss": 0.0026,
      "step": 2943
    },
    {
      "epoch": 37.62300319488818,
      "grad_norm": 0.37425318360328674,
      "learning_rate": 4.902564102564103e-06,
      "loss": 0.0029,
      "step": 2944
    },
    {
      "epoch": 37.635782747603834,
      "grad_norm": 0.34164369106292725,
      "learning_rate": 4.8974358974358975e-06,
      "loss": 0.0031,
      "step": 2945
    },
    {
      "epoch": 37.64856230031949,
      "grad_norm": 1.639210820198059,
      "learning_rate": 4.892307692307693e-06,
      "loss": 0.0159,
      "step": 2946
    },
    {
      "epoch": 37.66134185303515,
      "grad_norm": 0.34300586581230164,
      "learning_rate": 4.887179487179487e-06,
      "loss": 0.003,
      "step": 2947
    },
    {
      "epoch": 37.674121405750796,
      "grad_norm": 0.5739888548851013,
      "learning_rate": 4.882051282051282e-06,
      "loss": 0.0039,
      "step": 2948
    },
    {
      "epoch": 37.68690095846645,
      "grad_norm": 0.5022448301315308,
      "learning_rate": 4.876923076923077e-06,
      "loss": 0.0043,
      "step": 2949
    },
    {
      "epoch": 37.69968051118211,
      "grad_norm": 0.7767954468727112,
      "learning_rate": 4.871794871794872e-06,
      "loss": 0.0027,
      "step": 2950
    },
    {
      "epoch": 37.712460063897765,
      "grad_norm": 2.9445812702178955,
      "learning_rate": 4.866666666666667e-06,
      "loss": 0.0195,
      "step": 2951
    },
    {
      "epoch": 37.72523961661342,
      "grad_norm": 3.331125259399414,
      "learning_rate": 4.861538461538462e-06,
      "loss": 0.0362,
      "step": 2952
    },
    {
      "epoch": 37.73801916932907,
      "grad_norm": 0.38296833634376526,
      "learning_rate": 4.856410256410257e-06,
      "loss": 0.0027,
      "step": 2953
    },
    {
      "epoch": 37.75079872204473,
      "grad_norm": 0.27674436569213867,
      "learning_rate": 4.851282051282052e-06,
      "loss": 0.0033,
      "step": 2954
    },
    {
      "epoch": 37.763578274760384,
      "grad_norm": 0.3808777928352356,
      "learning_rate": 4.8461538461538465e-06,
      "loss": 0.0027,
      "step": 2955
    },
    {
      "epoch": 37.77635782747604,
      "grad_norm": 4.131234645843506,
      "learning_rate": 4.8410256410256414e-06,
      "loss": 0.04,
      "step": 2956
    },
    {
      "epoch": 37.78913738019169,
      "grad_norm": 0.3839050829410553,
      "learning_rate": 4.835897435897436e-06,
      "loss": 0.0037,
      "step": 2957
    },
    {
      "epoch": 37.801916932907346,
      "grad_norm": 0.4981125295162201,
      "learning_rate": 4.830769230769231e-06,
      "loss": 0.0054,
      "step": 2958
    },
    {
      "epoch": 37.814696485623,
      "grad_norm": 0.6987520456314087,
      "learning_rate": 4.825641025641026e-06,
      "loss": 0.0047,
      "step": 2959
    },
    {
      "epoch": 37.82747603833866,
      "grad_norm": 3.7170562744140625,
      "learning_rate": 4.820512820512821e-06,
      "loss": 0.0276,
      "step": 2960
    },
    {
      "epoch": 37.840255591054316,
      "grad_norm": 3.376850128173828,
      "learning_rate": 4.815384615384616e-06,
      "loss": 0.0226,
      "step": 2961
    },
    {
      "epoch": 37.853035143769965,
      "grad_norm": 0.2902922034263611,
      "learning_rate": 4.810256410256411e-06,
      "loss": 0.003,
      "step": 2962
    },
    {
      "epoch": 37.86581469648562,
      "grad_norm": 0.4471033215522766,
      "learning_rate": 4.805128205128206e-06,
      "loss": 0.0038,
      "step": 2963
    },
    {
      "epoch": 37.87859424920128,
      "grad_norm": 0.24476750195026398,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.0027,
      "step": 2964
    },
    {
      "epoch": 37.891373801916934,
      "grad_norm": 0.436432421207428,
      "learning_rate": 4.7948717948717955e-06,
      "loss": 0.0037,
      "step": 2965
    },
    {
      "epoch": 37.90415335463259,
      "grad_norm": 0.403041273355484,
      "learning_rate": 4.7897435897435904e-06,
      "loss": 0.004,
      "step": 2966
    },
    {
      "epoch": 37.91693290734824,
      "grad_norm": 0.39625176787376404,
      "learning_rate": 4.7846153846153845e-06,
      "loss": 0.0033,
      "step": 2967
    },
    {
      "epoch": 37.9297124600639,
      "grad_norm": 2.864386796951294,
      "learning_rate": 4.77948717948718e-06,
      "loss": 0.0088,
      "step": 2968
    },
    {
      "epoch": 37.94249201277955,
      "grad_norm": 0.7663731575012207,
      "learning_rate": 4.774358974358974e-06,
      "loss": 0.0086,
      "step": 2969
    },
    {
      "epoch": 37.95527156549521,
      "grad_norm": 0.352105051279068,
      "learning_rate": 4.76923076923077e-06,
      "loss": 0.0036,
      "step": 2970
    },
    {
      "epoch": 37.968051118210866,
      "grad_norm": 0.330779105424881,
      "learning_rate": 4.764102564102565e-06,
      "loss": 0.0028,
      "step": 2971
    },
    {
      "epoch": 37.980830670926515,
      "grad_norm": 0.3980065882205963,
      "learning_rate": 4.758974358974359e-06,
      "loss": 0.0035,
      "step": 2972
    },
    {
      "epoch": 37.99361022364217,
      "grad_norm": 0.41715294122695923,
      "learning_rate": 4.753846153846155e-06,
      "loss": 0.0033,
      "step": 2973
    },
    {
      "epoch": 38.00638977635783,
      "grad_norm": 0.305719256401062,
      "learning_rate": 4.748717948717949e-06,
      "loss": 0.004,
      "step": 2974
    },
    {
      "epoch": 38.019169329073485,
      "grad_norm": 0.3436070680618286,
      "learning_rate": 4.743589743589744e-06,
      "loss": 0.0035,
      "step": 2975
    },
    {
      "epoch": 38.031948881789134,
      "grad_norm": 0.7291021943092346,
      "learning_rate": 4.738461538461539e-06,
      "loss": 0.0035,
      "step": 2976
    },
    {
      "epoch": 38.04472843450479,
      "grad_norm": 0.37716665863990784,
      "learning_rate": 4.7333333333333335e-06,
      "loss": 0.0029,
      "step": 2977
    },
    {
      "epoch": 38.05750798722045,
      "grad_norm": 1.5769388675689697,
      "learning_rate": 4.728205128205128e-06,
      "loss": 0.0071,
      "step": 2978
    },
    {
      "epoch": 38.0702875399361,
      "grad_norm": 1.3593897819519043,
      "learning_rate": 4.723076923076923e-06,
      "loss": 0.0067,
      "step": 2979
    },
    {
      "epoch": 38.08306709265176,
      "grad_norm": 3.29752254486084,
      "learning_rate": 4.717948717948718e-06,
      "loss": 0.0193,
      "step": 2980
    },
    {
      "epoch": 38.09584664536741,
      "grad_norm": 0.26745957136154175,
      "learning_rate": 4.712820512820513e-06,
      "loss": 0.0032,
      "step": 2981
    },
    {
      "epoch": 38.108626198083066,
      "grad_norm": 0.2419317215681076,
      "learning_rate": 4.707692307692308e-06,
      "loss": 0.0027,
      "step": 2982
    },
    {
      "epoch": 38.12140575079872,
      "grad_norm": 0.45501840114593506,
      "learning_rate": 4.702564102564103e-06,
      "loss": 0.0035,
      "step": 2983
    },
    {
      "epoch": 38.13418530351438,
      "grad_norm": 0.28786441683769226,
      "learning_rate": 4.697435897435898e-06,
      "loss": 0.0023,
      "step": 2984
    },
    {
      "epoch": 38.146964856230035,
      "grad_norm": 0.9460015892982483,
      "learning_rate": 4.692307692307693e-06,
      "loss": 0.0033,
      "step": 2985
    },
    {
      "epoch": 38.159744408945684,
      "grad_norm": 0.3270238935947418,
      "learning_rate": 4.6871794871794876e-06,
      "loss": 0.0022,
      "step": 2986
    },
    {
      "epoch": 38.17252396166134,
      "grad_norm": 0.2666434645652771,
      "learning_rate": 4.6820512820512825e-06,
      "loss": 0.0017,
      "step": 2987
    },
    {
      "epoch": 38.185303514377,
      "grad_norm": 3.4922378063201904,
      "learning_rate": 4.676923076923077e-06,
      "loss": 0.0298,
      "step": 2988
    },
    {
      "epoch": 38.198083067092654,
      "grad_norm": 0.3040410280227661,
      "learning_rate": 4.671794871794872e-06,
      "loss": 0.0037,
      "step": 2989
    },
    {
      "epoch": 38.21086261980831,
      "grad_norm": 0.8676387667655945,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0045,
      "step": 2990
    },
    {
      "epoch": 38.22364217252396,
      "grad_norm": 3.9531896114349365,
      "learning_rate": 4.661538461538462e-06,
      "loss": 0.0048,
      "step": 2991
    },
    {
      "epoch": 38.236421725239616,
      "grad_norm": 1.1728514432907104,
      "learning_rate": 4.656410256410257e-06,
      "loss": 0.0063,
      "step": 2992
    },
    {
      "epoch": 38.24920127795527,
      "grad_norm": 0.9603011012077332,
      "learning_rate": 4.651282051282052e-06,
      "loss": 0.0042,
      "step": 2993
    },
    {
      "epoch": 38.26198083067093,
      "grad_norm": 0.2621918320655823,
      "learning_rate": 4.646153846153847e-06,
      "loss": 0.0023,
      "step": 2994
    },
    {
      "epoch": 38.27476038338658,
      "grad_norm": 2.27105712890625,
      "learning_rate": 4.641025641025642e-06,
      "loss": 0.0093,
      "step": 2995
    },
    {
      "epoch": 38.287539936102235,
      "grad_norm": 0.7400266528129578,
      "learning_rate": 4.6358974358974366e-06,
      "loss": 0.0036,
      "step": 2996
    },
    {
      "epoch": 38.30031948881789,
      "grad_norm": 0.25827065110206604,
      "learning_rate": 4.630769230769231e-06,
      "loss": 0.0028,
      "step": 2997
    },
    {
      "epoch": 38.31309904153355,
      "grad_norm": 0.3045773208141327,
      "learning_rate": 4.625641025641026e-06,
      "loss": 0.0022,
      "step": 2998
    },
    {
      "epoch": 38.325878594249204,
      "grad_norm": 0.4771479070186615,
      "learning_rate": 4.62051282051282e-06,
      "loss": 0.0044,
      "step": 2999
    },
    {
      "epoch": 38.33865814696485,
      "grad_norm": 5.8205246925354,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.0185,
      "step": 3000
    },
    {
      "epoch": 38.33865814696485,
      "eval_loss": 0.9848964214324951,
      "eval_runtime": 183.6684,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 3000
    },
    {
      "epoch": 38.35143769968051,
      "grad_norm": 0.44981399178504944,
      "learning_rate": 4.610256410256411e-06,
      "loss": 0.0034,
      "step": 3001
    },
    {
      "epoch": 38.364217252396166,
      "grad_norm": 0.2838537096977234,
      "learning_rate": 4.605128205128205e-06,
      "loss": 0.0024,
      "step": 3002
    },
    {
      "epoch": 38.37699680511182,
      "grad_norm": 3.2543389797210693,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.0138,
      "step": 3003
    },
    {
      "epoch": 38.38977635782748,
      "grad_norm": 0.35357922315597534,
      "learning_rate": 4.594871794871795e-06,
      "loss": 0.0028,
      "step": 3004
    },
    {
      "epoch": 38.40255591054313,
      "grad_norm": 0.2850092947483063,
      "learning_rate": 4.58974358974359e-06,
      "loss": 0.0025,
      "step": 3005
    },
    {
      "epoch": 38.415335463258785,
      "grad_norm": 3.4489376544952393,
      "learning_rate": 4.5846153846153855e-06,
      "loss": 0.0251,
      "step": 3006
    },
    {
      "epoch": 38.42811501597444,
      "grad_norm": 0.3378681242465973,
      "learning_rate": 4.57948717948718e-06,
      "loss": 0.0025,
      "step": 3007
    },
    {
      "epoch": 38.4408945686901,
      "grad_norm": 1.5556156635284424,
      "learning_rate": 4.5743589743589745e-06,
      "loss": 0.0109,
      "step": 3008
    },
    {
      "epoch": 38.453674121405754,
      "grad_norm": 0.47560226917266846,
      "learning_rate": 4.569230769230769e-06,
      "loss": 0.0034,
      "step": 3009
    },
    {
      "epoch": 38.466453674121404,
      "grad_norm": 0.6198320984840393,
      "learning_rate": 4.564102564102564e-06,
      "loss": 0.0122,
      "step": 3010
    },
    {
      "epoch": 38.47923322683706,
      "grad_norm": 0.3300226628780365,
      "learning_rate": 4.558974358974359e-06,
      "loss": 0.0034,
      "step": 3011
    },
    {
      "epoch": 38.49201277955272,
      "grad_norm": 0.2798616290092468,
      "learning_rate": 4.553846153846154e-06,
      "loss": 0.003,
      "step": 3012
    },
    {
      "epoch": 38.50479233226837,
      "grad_norm": 0.7014890909194946,
      "learning_rate": 4.548717948717949e-06,
      "loss": 0.0048,
      "step": 3013
    },
    {
      "epoch": 38.51757188498402,
      "grad_norm": 0.28109487891197205,
      "learning_rate": 4.543589743589744e-06,
      "loss": 0.0027,
      "step": 3014
    },
    {
      "epoch": 38.53035143769968,
      "grad_norm": 0.45775285363197327,
      "learning_rate": 4.538461538461539e-06,
      "loss": 0.0025,
      "step": 3015
    },
    {
      "epoch": 38.543130990415335,
      "grad_norm": 0.34922537207603455,
      "learning_rate": 4.533333333333334e-06,
      "loss": 0.0036,
      "step": 3016
    },
    {
      "epoch": 38.55591054313099,
      "grad_norm": 2.250239133834839,
      "learning_rate": 4.528205128205129e-06,
      "loss": 0.0098,
      "step": 3017
    },
    {
      "epoch": 38.56869009584665,
      "grad_norm": 0.5357843041419983,
      "learning_rate": 4.5230769230769235e-06,
      "loss": 0.0048,
      "step": 3018
    },
    {
      "epoch": 38.5814696485623,
      "grad_norm": 1.595799207687378,
      "learning_rate": 4.517948717948718e-06,
      "loss": 0.0158,
      "step": 3019
    },
    {
      "epoch": 38.594249201277954,
      "grad_norm": 1.6008353233337402,
      "learning_rate": 4.512820512820513e-06,
      "loss": 0.0074,
      "step": 3020
    },
    {
      "epoch": 38.60702875399361,
      "grad_norm": 1.9269359111785889,
      "learning_rate": 4.507692307692308e-06,
      "loss": 0.0155,
      "step": 3021
    },
    {
      "epoch": 38.61980830670927,
      "grad_norm": 0.2042221873998642,
      "learning_rate": 4.502564102564103e-06,
      "loss": 0.002,
      "step": 3022
    },
    {
      "epoch": 38.63258785942492,
      "grad_norm": 0.35417866706848145,
      "learning_rate": 4.497435897435898e-06,
      "loss": 0.0025,
      "step": 3023
    },
    {
      "epoch": 38.64536741214057,
      "grad_norm": 3.2746737003326416,
      "learning_rate": 4.492307692307693e-06,
      "loss": 0.0119,
      "step": 3024
    },
    {
      "epoch": 38.65814696485623,
      "grad_norm": 0.31555283069610596,
      "learning_rate": 4.487179487179488e-06,
      "loss": 0.0029,
      "step": 3025
    },
    {
      "epoch": 38.670926517571885,
      "grad_norm": 0.38801953196525574,
      "learning_rate": 4.482051282051283e-06,
      "loss": 0.0026,
      "step": 3026
    },
    {
      "epoch": 38.68370607028754,
      "grad_norm": 1.4678237438201904,
      "learning_rate": 4.476923076923077e-06,
      "loss": 0.0065,
      "step": 3027
    },
    {
      "epoch": 38.6964856230032,
      "grad_norm": 0.3929405212402344,
      "learning_rate": 4.4717948717948725e-06,
      "loss": 0.0028,
      "step": 3028
    },
    {
      "epoch": 38.70926517571885,
      "grad_norm": 0.3595089912414551,
      "learning_rate": 4.4666666666666665e-06,
      "loss": 0.0028,
      "step": 3029
    },
    {
      "epoch": 38.722044728434504,
      "grad_norm": 0.2732454240322113,
      "learning_rate": 4.461538461538462e-06,
      "loss": 0.0024,
      "step": 3030
    },
    {
      "epoch": 38.73482428115016,
      "grad_norm": 2.7754180431365967,
      "learning_rate": 4.456410256410257e-06,
      "loss": 0.0156,
      "step": 3031
    },
    {
      "epoch": 38.74760383386582,
      "grad_norm": 0.30526700615882874,
      "learning_rate": 4.451282051282051e-06,
      "loss": 0.0034,
      "step": 3032
    },
    {
      "epoch": 38.760383386581466,
      "grad_norm": 0.27021709084510803,
      "learning_rate": 4.446153846153847e-06,
      "loss": 0.0023,
      "step": 3033
    },
    {
      "epoch": 38.77316293929712,
      "grad_norm": 0.3636169731616974,
      "learning_rate": 4.441025641025641e-06,
      "loss": 0.0033,
      "step": 3034
    },
    {
      "epoch": 38.78594249201278,
      "grad_norm": 0.2715635895729065,
      "learning_rate": 4.435897435897436e-06,
      "loss": 0.0031,
      "step": 3035
    },
    {
      "epoch": 38.798722044728436,
      "grad_norm": 0.4538557529449463,
      "learning_rate": 4.430769230769232e-06,
      "loss": 0.003,
      "step": 3036
    },
    {
      "epoch": 38.81150159744409,
      "grad_norm": 0.35114040970802307,
      "learning_rate": 4.425641025641026e-06,
      "loss": 0.0026,
      "step": 3037
    },
    {
      "epoch": 38.82428115015974,
      "grad_norm": 2.235126495361328,
      "learning_rate": 4.420512820512821e-06,
      "loss": 0.0113,
      "step": 3038
    },
    {
      "epoch": 38.8370607028754,
      "grad_norm": 1.4098985195159912,
      "learning_rate": 4.4153846153846155e-06,
      "loss": 0.0102,
      "step": 3039
    },
    {
      "epoch": 38.849840255591054,
      "grad_norm": 0.2800547182559967,
      "learning_rate": 4.4102564102564104e-06,
      "loss": 0.0033,
      "step": 3040
    },
    {
      "epoch": 38.86261980830671,
      "grad_norm": 0.9228421449661255,
      "learning_rate": 4.405128205128205e-06,
      "loss": 0.0071,
      "step": 3041
    },
    {
      "epoch": 38.87539936102237,
      "grad_norm": 0.4252464473247528,
      "learning_rate": 4.4e-06,
      "loss": 0.0034,
      "step": 3042
    },
    {
      "epoch": 38.88817891373802,
      "grad_norm": 1.1408131122589111,
      "learning_rate": 4.394871794871795e-06,
      "loss": 0.003,
      "step": 3043
    },
    {
      "epoch": 38.90095846645367,
      "grad_norm": 3.826991081237793,
      "learning_rate": 4.38974358974359e-06,
      "loss": 0.0224,
      "step": 3044
    },
    {
      "epoch": 38.91373801916933,
      "grad_norm": 0.3651482164859772,
      "learning_rate": 4.384615384615385e-06,
      "loss": 0.0026,
      "step": 3045
    },
    {
      "epoch": 38.926517571884986,
      "grad_norm": 3.332813024520874,
      "learning_rate": 4.37948717948718e-06,
      "loss": 0.0297,
      "step": 3046
    },
    {
      "epoch": 38.93929712460064,
      "grad_norm": 0.4605509042739868,
      "learning_rate": 4.374358974358975e-06,
      "loss": 0.0044,
      "step": 3047
    },
    {
      "epoch": 38.95207667731629,
      "grad_norm": 0.5724936723709106,
      "learning_rate": 4.36923076923077e-06,
      "loss": 0.0052,
      "step": 3048
    },
    {
      "epoch": 38.96485623003195,
      "grad_norm": 0.2828877568244934,
      "learning_rate": 4.3641025641025645e-06,
      "loss": 0.003,
      "step": 3049
    },
    {
      "epoch": 38.977635782747605,
      "grad_norm": 2.8186399936676025,
      "learning_rate": 4.358974358974359e-06,
      "loss": 0.0181,
      "step": 3050
    },
    {
      "epoch": 38.99041533546326,
      "grad_norm": 7.178375244140625,
      "learning_rate": 4.353846153846154e-06,
      "loss": 0.0181,
      "step": 3051
    },
    {
      "epoch": 39.00319488817891,
      "grad_norm": 0.39194220304489136,
      "learning_rate": 4.348717948717949e-06,
      "loss": 0.0034,
      "step": 3052
    },
    {
      "epoch": 39.01597444089457,
      "grad_norm": 0.32534223794937134,
      "learning_rate": 4.343589743589744e-06,
      "loss": 0.003,
      "step": 3053
    },
    {
      "epoch": 39.02875399361022,
      "grad_norm": 4.849320411682129,
      "learning_rate": 4.338461538461539e-06,
      "loss": 0.0042,
      "step": 3054
    },
    {
      "epoch": 39.04153354632588,
      "grad_norm": 0.8111512064933777,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.0019,
      "step": 3055
    },
    {
      "epoch": 39.054313099041536,
      "grad_norm": 0.2802063226699829,
      "learning_rate": 4.328205128205129e-06,
      "loss": 0.0024,
      "step": 3056
    },
    {
      "epoch": 39.067092651757186,
      "grad_norm": 2.6525161266326904,
      "learning_rate": 4.323076923076923e-06,
      "loss": 0.0137,
      "step": 3057
    },
    {
      "epoch": 39.07987220447284,
      "grad_norm": 0.2541932761669159,
      "learning_rate": 4.317948717948719e-06,
      "loss": 0.003,
      "step": 3058
    },
    {
      "epoch": 39.0926517571885,
      "grad_norm": 0.43173283338546753,
      "learning_rate": 4.312820512820513e-06,
      "loss": 0.0037,
      "step": 3059
    },
    {
      "epoch": 39.105431309904155,
      "grad_norm": 1.6924418210983276,
      "learning_rate": 4.307692307692308e-06,
      "loss": 0.0097,
      "step": 3060
    },
    {
      "epoch": 39.11821086261981,
      "grad_norm": 0.2868981659412384,
      "learning_rate": 4.302564102564103e-06,
      "loss": 0.0025,
      "step": 3061
    },
    {
      "epoch": 39.13099041533546,
      "grad_norm": 0.24275152385234833,
      "learning_rate": 4.297435897435897e-06,
      "loss": 0.0021,
      "step": 3062
    },
    {
      "epoch": 39.14376996805112,
      "grad_norm": 1.1402617692947388,
      "learning_rate": 4.292307692307693e-06,
      "loss": 0.0057,
      "step": 3063
    },
    {
      "epoch": 39.156549520766774,
      "grad_norm": 0.29625436663627625,
      "learning_rate": 4.287179487179487e-06,
      "loss": 0.0028,
      "step": 3064
    },
    {
      "epoch": 39.16932907348243,
      "grad_norm": 0.9124428629875183,
      "learning_rate": 4.282051282051282e-06,
      "loss": 0.0033,
      "step": 3065
    },
    {
      "epoch": 39.18210862619808,
      "grad_norm": 0.31289470195770264,
      "learning_rate": 4.276923076923078e-06,
      "loss": 0.0028,
      "step": 3066
    },
    {
      "epoch": 39.194888178913736,
      "grad_norm": 1.9680798053741455,
      "learning_rate": 4.271794871794872e-06,
      "loss": 0.0081,
      "step": 3067
    },
    {
      "epoch": 39.20766773162939,
      "grad_norm": 0.2573452293872833,
      "learning_rate": 4.266666666666668e-06,
      "loss": 0.0029,
      "step": 3068
    },
    {
      "epoch": 39.22044728434505,
      "grad_norm": 2.266730308532715,
      "learning_rate": 4.261538461538462e-06,
      "loss": 0.0179,
      "step": 3069
    },
    {
      "epoch": 39.233226837060705,
      "grad_norm": 0.40569251775741577,
      "learning_rate": 4.2564102564102566e-06,
      "loss": 0.0044,
      "step": 3070
    },
    {
      "epoch": 39.246006389776355,
      "grad_norm": 0.24747313559055328,
      "learning_rate": 4.2512820512820515e-06,
      "loss": 0.0029,
      "step": 3071
    },
    {
      "epoch": 39.25878594249201,
      "grad_norm": 0.6296175122261047,
      "learning_rate": 4.246153846153846e-06,
      "loss": 0.0039,
      "step": 3072
    },
    {
      "epoch": 39.27156549520767,
      "grad_norm": 0.3082866966724396,
      "learning_rate": 4.241025641025641e-06,
      "loss": 0.0022,
      "step": 3073
    },
    {
      "epoch": 39.284345047923324,
      "grad_norm": 2.4070003032684326,
      "learning_rate": 4.235897435897436e-06,
      "loss": 0.0071,
      "step": 3074
    },
    {
      "epoch": 39.29712460063898,
      "grad_norm": 0.5413769483566284,
      "learning_rate": 4.230769230769231e-06,
      "loss": 0.0095,
      "step": 3075
    },
    {
      "epoch": 39.30990415335463,
      "grad_norm": 0.3211362659931183,
      "learning_rate": 4.225641025641026e-06,
      "loss": 0.002,
      "step": 3076
    },
    {
      "epoch": 39.322683706070286,
      "grad_norm": 1.298190951347351,
      "learning_rate": 4.220512820512821e-06,
      "loss": 0.0081,
      "step": 3077
    },
    {
      "epoch": 39.33546325878594,
      "grad_norm": 0.22302180528640747,
      "learning_rate": 4.215384615384616e-06,
      "loss": 0.0018,
      "step": 3078
    },
    {
      "epoch": 39.3482428115016,
      "grad_norm": 0.25386327505111694,
      "learning_rate": 4.210256410256411e-06,
      "loss": 0.0023,
      "step": 3079
    },
    {
      "epoch": 39.361022364217256,
      "grad_norm": 0.2036847472190857,
      "learning_rate": 4.2051282051282055e-06,
      "loss": 0.002,
      "step": 3080
    },
    {
      "epoch": 39.373801916932905,
      "grad_norm": 0.314125120639801,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.003,
      "step": 3081
    },
    {
      "epoch": 39.38658146964856,
      "grad_norm": 0.35973840951919556,
      "learning_rate": 4.194871794871795e-06,
      "loss": 0.0033,
      "step": 3082
    },
    {
      "epoch": 39.39936102236422,
      "grad_norm": 0.20182707905769348,
      "learning_rate": 4.18974358974359e-06,
      "loss": 0.0023,
      "step": 3083
    },
    {
      "epoch": 39.412140575079874,
      "grad_norm": 4.579025745391846,
      "learning_rate": 4.184615384615385e-06,
      "loss": 0.0253,
      "step": 3084
    },
    {
      "epoch": 39.424920127795524,
      "grad_norm": 0.4407001733779907,
      "learning_rate": 4.17948717948718e-06,
      "loss": 0.004,
      "step": 3085
    },
    {
      "epoch": 39.43769968051118,
      "grad_norm": 2.039889097213745,
      "learning_rate": 4.174358974358975e-06,
      "loss": 0.0088,
      "step": 3086
    },
    {
      "epoch": 39.45047923322684,
      "grad_norm": 0.9066301584243774,
      "learning_rate": 4.169230769230769e-06,
      "loss": 0.0076,
      "step": 3087
    },
    {
      "epoch": 39.46325878594249,
      "grad_norm": 10.292778015136719,
      "learning_rate": 4.164102564102565e-06,
      "loss": 0.0347,
      "step": 3088
    },
    {
      "epoch": 39.47603833865815,
      "grad_norm": 0.22446775436401367,
      "learning_rate": 4.158974358974359e-06,
      "loss": 0.0024,
      "step": 3089
    },
    {
      "epoch": 39.4888178913738,
      "grad_norm": 2.612048387527466,
      "learning_rate": 4.1538461538461545e-06,
      "loss": 0.0233,
      "step": 3090
    },
    {
      "epoch": 39.501597444089455,
      "grad_norm": 1.2295528650283813,
      "learning_rate": 4.1487179487179494e-06,
      "loss": 0.0043,
      "step": 3091
    },
    {
      "epoch": 39.51437699680511,
      "grad_norm": 0.2916811406612396,
      "learning_rate": 4.1435897435897435e-06,
      "loss": 0.0024,
      "step": 3092
    },
    {
      "epoch": 39.52715654952077,
      "grad_norm": 0.45092812180519104,
      "learning_rate": 4.138461538461539e-06,
      "loss": 0.0033,
      "step": 3093
    },
    {
      "epoch": 39.539936102236425,
      "grad_norm": 0.4554513990879059,
      "learning_rate": 4.133333333333333e-06,
      "loss": 0.0034,
      "step": 3094
    },
    {
      "epoch": 39.552715654952074,
      "grad_norm": 0.3552434742450714,
      "learning_rate": 4.128205128205128e-06,
      "loss": 0.0024,
      "step": 3095
    },
    {
      "epoch": 39.56549520766773,
      "grad_norm": 2.480602264404297,
      "learning_rate": 4.123076923076924e-06,
      "loss": 0.0075,
      "step": 3096
    },
    {
      "epoch": 39.57827476038339,
      "grad_norm": 0.3173219561576843,
      "learning_rate": 4.117948717948718e-06,
      "loss": 0.0022,
      "step": 3097
    },
    {
      "epoch": 39.59105431309904,
      "grad_norm": 0.46305596828460693,
      "learning_rate": 4.112820512820514e-06,
      "loss": 0.004,
      "step": 3098
    },
    {
      "epoch": 39.6038338658147,
      "grad_norm": 0.5275067090988159,
      "learning_rate": 4.107692307692308e-06,
      "loss": 0.0039,
      "step": 3099
    },
    {
      "epoch": 39.61661341853035,
      "grad_norm": 2.0127456188201904,
      "learning_rate": 4.102564102564103e-06,
      "loss": 0.0135,
      "step": 3100
    },
    {
      "epoch": 39.61661341853035,
      "eval_loss": 0.9879663586616516,
      "eval_runtime": 183.9322,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 3100
    },
    {
      "epoch": 39.629392971246006,
      "grad_norm": 0.3211822211742401,
      "learning_rate": 4.097435897435898e-06,
      "loss": 0.0026,
      "step": 3101
    },
    {
      "epoch": 39.64217252396166,
      "grad_norm": 1.2732728719711304,
      "learning_rate": 4.0923076923076925e-06,
      "loss": 0.0035,
      "step": 3102
    },
    {
      "epoch": 39.65495207667732,
      "grad_norm": 1.2496635913848877,
      "learning_rate": 4.087179487179487e-06,
      "loss": 0.0037,
      "step": 3103
    },
    {
      "epoch": 39.66773162939297,
      "grad_norm": 3.28835391998291,
      "learning_rate": 4.082051282051282e-06,
      "loss": 0.0081,
      "step": 3104
    },
    {
      "epoch": 39.680511182108624,
      "grad_norm": 0.25976139307022095,
      "learning_rate": 4.076923076923077e-06,
      "loss": 0.0025,
      "step": 3105
    },
    {
      "epoch": 39.69329073482428,
      "grad_norm": 0.3309383690357208,
      "learning_rate": 4.071794871794872e-06,
      "loss": 0.0036,
      "step": 3106
    },
    {
      "epoch": 39.70607028753994,
      "grad_norm": 1.8688527345657349,
      "learning_rate": 4.066666666666667e-06,
      "loss": 0.0079,
      "step": 3107
    },
    {
      "epoch": 39.718849840255594,
      "grad_norm": 1.6854923963546753,
      "learning_rate": 4.061538461538462e-06,
      "loss": 0.0113,
      "step": 3108
    },
    {
      "epoch": 39.73162939297124,
      "grad_norm": 1.842885136604309,
      "learning_rate": 4.056410256410257e-06,
      "loss": 0.0077,
      "step": 3109
    },
    {
      "epoch": 39.7444089456869,
      "grad_norm": 3.535550594329834,
      "learning_rate": 4.051282051282052e-06,
      "loss": 0.0205,
      "step": 3110
    },
    {
      "epoch": 39.757188498402556,
      "grad_norm": 0.29745614528656006,
      "learning_rate": 4.0461538461538466e-06,
      "loss": 0.0032,
      "step": 3111
    },
    {
      "epoch": 39.76996805111821,
      "grad_norm": 3.15605092048645,
      "learning_rate": 4.0410256410256415e-06,
      "loss": 0.0232,
      "step": 3112
    },
    {
      "epoch": 39.78274760383387,
      "grad_norm": 0.2782306969165802,
      "learning_rate": 4.035897435897436e-06,
      "loss": 0.0027,
      "step": 3113
    },
    {
      "epoch": 39.79552715654952,
      "grad_norm": 0.4322669804096222,
      "learning_rate": 4.030769230769231e-06,
      "loss": 0.004,
      "step": 3114
    },
    {
      "epoch": 39.808306709265175,
      "grad_norm": 1.0812246799468994,
      "learning_rate": 4.025641025641026e-06,
      "loss": 0.0057,
      "step": 3115
    },
    {
      "epoch": 39.82108626198083,
      "grad_norm": 3.448333978652954,
      "learning_rate": 4.020512820512821e-06,
      "loss": 0.0227,
      "step": 3116
    },
    {
      "epoch": 39.83386581469649,
      "grad_norm": 2.9860076904296875,
      "learning_rate": 4.015384615384615e-06,
      "loss": 0.0124,
      "step": 3117
    },
    {
      "epoch": 39.846645367412144,
      "grad_norm": 2.200941562652588,
      "learning_rate": 4.010256410256411e-06,
      "loss": 0.0127,
      "step": 3118
    },
    {
      "epoch": 39.85942492012779,
      "grad_norm": 0.4125717580318451,
      "learning_rate": 4.005128205128205e-06,
      "loss": 0.0026,
      "step": 3119
    },
    {
      "epoch": 39.87220447284345,
      "grad_norm": 2.127324342727661,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.015,
      "step": 3120
    },
    {
      "epoch": 39.884984025559106,
      "grad_norm": 0.2309257984161377,
      "learning_rate": 3.9948717948717956e-06,
      "loss": 0.0025,
      "step": 3121
    },
    {
      "epoch": 39.89776357827476,
      "grad_norm": 0.2240867018699646,
      "learning_rate": 3.98974358974359e-06,
      "loss": 0.0018,
      "step": 3122
    },
    {
      "epoch": 39.91054313099041,
      "grad_norm": 0.3975813686847687,
      "learning_rate": 3.984615384615385e-06,
      "loss": 0.0031,
      "step": 3123
    },
    {
      "epoch": 39.92332268370607,
      "grad_norm": 2.780297040939331,
      "learning_rate": 3.979487179487179e-06,
      "loss": 0.0178,
      "step": 3124
    },
    {
      "epoch": 39.936102236421725,
      "grad_norm": 0.2866804003715515,
      "learning_rate": 3.974358974358974e-06,
      "loss": 0.0024,
      "step": 3125
    },
    {
      "epoch": 39.94888178913738,
      "grad_norm": 0.3142434358596802,
      "learning_rate": 3.96923076923077e-06,
      "loss": 0.0028,
      "step": 3126
    },
    {
      "epoch": 39.96166134185304,
      "grad_norm": 0.3115275204181671,
      "learning_rate": 3.964102564102564e-06,
      "loss": 0.0028,
      "step": 3127
    },
    {
      "epoch": 39.97444089456869,
      "grad_norm": 2.3010635375976562,
      "learning_rate": 3.95897435897436e-06,
      "loss": 0.0112,
      "step": 3128
    },
    {
      "epoch": 39.98722044728434,
      "grad_norm": 0.289372980594635,
      "learning_rate": 3.953846153846154e-06,
      "loss": 0.0024,
      "step": 3129
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.49848124384880066,
      "learning_rate": 3.948717948717949e-06,
      "loss": 0.0032,
      "step": 3130
    },
    {
      "epoch": 40.01277955271566,
      "grad_norm": 0.19198329746723175,
      "learning_rate": 3.943589743589744e-06,
      "loss": 0.0018,
      "step": 3131
    },
    {
      "epoch": 40.02555910543131,
      "grad_norm": 0.3196083903312683,
      "learning_rate": 3.938461538461539e-06,
      "loss": 0.0036,
      "step": 3132
    },
    {
      "epoch": 40.03833865814696,
      "grad_norm": 0.593464195728302,
      "learning_rate": 3.9333333333333335e-06,
      "loss": 0.0028,
      "step": 3133
    },
    {
      "epoch": 40.05111821086262,
      "grad_norm": 0.2691340148448944,
      "learning_rate": 3.928205128205128e-06,
      "loss": 0.0032,
      "step": 3134
    },
    {
      "epoch": 40.063897763578275,
      "grad_norm": 0.20105499029159546,
      "learning_rate": 3.923076923076923e-06,
      "loss": 0.0026,
      "step": 3135
    },
    {
      "epoch": 40.07667731629393,
      "grad_norm": 0.30826905369758606,
      "learning_rate": 3.917948717948718e-06,
      "loss": 0.0035,
      "step": 3136
    },
    {
      "epoch": 40.08945686900959,
      "grad_norm": 0.37188711762428284,
      "learning_rate": 3.912820512820513e-06,
      "loss": 0.0034,
      "step": 3137
    },
    {
      "epoch": 40.10223642172524,
      "grad_norm": 0.3785606026649475,
      "learning_rate": 3.907692307692308e-06,
      "loss": 0.0025,
      "step": 3138
    },
    {
      "epoch": 40.115015974440894,
      "grad_norm": 1.697972297668457,
      "learning_rate": 3.902564102564103e-06,
      "loss": 0.0042,
      "step": 3139
    },
    {
      "epoch": 40.12779552715655,
      "grad_norm": 1.2668052911758423,
      "learning_rate": 3.897435897435898e-06,
      "loss": 0.0052,
      "step": 3140
    },
    {
      "epoch": 40.14057507987221,
      "grad_norm": 2.1785550117492676,
      "learning_rate": 3.892307692307693e-06,
      "loss": 0.0086,
      "step": 3141
    },
    {
      "epoch": 40.153354632587856,
      "grad_norm": 0.5524407029151917,
      "learning_rate": 3.887179487179488e-06,
      "loss": 0.0029,
      "step": 3142
    },
    {
      "epoch": 40.16613418530351,
      "grad_norm": 0.754425585269928,
      "learning_rate": 3.8820512820512825e-06,
      "loss": 0.0049,
      "step": 3143
    },
    {
      "epoch": 40.17891373801917,
      "grad_norm": 0.3464694917201996,
      "learning_rate": 3.876923076923077e-06,
      "loss": 0.0045,
      "step": 3144
    },
    {
      "epoch": 40.191693290734825,
      "grad_norm": 0.23320907354354858,
      "learning_rate": 3.871794871794872e-06,
      "loss": 0.0022,
      "step": 3145
    },
    {
      "epoch": 40.20447284345048,
      "grad_norm": 0.5744196772575378,
      "learning_rate": 3.866666666666667e-06,
      "loss": 0.0043,
      "step": 3146
    },
    {
      "epoch": 40.21725239616613,
      "grad_norm": 0.42899665236473083,
      "learning_rate": 3.861538461538462e-06,
      "loss": 0.0031,
      "step": 3147
    },
    {
      "epoch": 40.23003194888179,
      "grad_norm": 0.516017735004425,
      "learning_rate": 3.856410256410257e-06,
      "loss": 0.0079,
      "step": 3148
    },
    {
      "epoch": 40.242811501597444,
      "grad_norm": 0.4321247637271881,
      "learning_rate": 3.851282051282051e-06,
      "loss": 0.0029,
      "step": 3149
    },
    {
      "epoch": 40.2555910543131,
      "grad_norm": 0.26432710886001587,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.0032,
      "step": 3150
    },
    {
      "epoch": 40.26837060702876,
      "grad_norm": 0.9468638300895691,
      "learning_rate": 3.841025641025642e-06,
      "loss": 0.0048,
      "step": 3151
    },
    {
      "epoch": 40.281150159744406,
      "grad_norm": 0.1992044448852539,
      "learning_rate": 3.835897435897436e-06,
      "loss": 0.0024,
      "step": 3152
    },
    {
      "epoch": 40.29392971246006,
      "grad_norm": 3.1869659423828125,
      "learning_rate": 3.8307692307692315e-06,
      "loss": 0.013,
      "step": 3153
    },
    {
      "epoch": 40.30670926517572,
      "grad_norm": 0.7191022038459778,
      "learning_rate": 3.8256410256410255e-06,
      "loss": 0.0042,
      "step": 3154
    },
    {
      "epoch": 40.319488817891376,
      "grad_norm": 0.34319254755973816,
      "learning_rate": 3.8205128205128204e-06,
      "loss": 0.0037,
      "step": 3155
    },
    {
      "epoch": 40.33226837060703,
      "grad_norm": 0.37301310896873474,
      "learning_rate": 3.815384615384616e-06,
      "loss": 0.0029,
      "step": 3156
    },
    {
      "epoch": 40.34504792332268,
      "grad_norm": 2.212172031402588,
      "learning_rate": 3.8102564102564107e-06,
      "loss": 0.0188,
      "step": 3157
    },
    {
      "epoch": 40.35782747603834,
      "grad_norm": 0.9722811579704285,
      "learning_rate": 3.8051282051282056e-06,
      "loss": 0.0025,
      "step": 3158
    },
    {
      "epoch": 40.370607028753994,
      "grad_norm": 2.930816173553467,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.0175,
      "step": 3159
    },
    {
      "epoch": 40.38338658146965,
      "grad_norm": 3.5118746757507324,
      "learning_rate": 3.794871794871795e-06,
      "loss": 0.0169,
      "step": 3160
    },
    {
      "epoch": 40.3961661341853,
      "grad_norm": 0.24613343179225922,
      "learning_rate": 3.7897435897435903e-06,
      "loss": 0.0025,
      "step": 3161
    },
    {
      "epoch": 40.40894568690096,
      "grad_norm": 0.21622180938720703,
      "learning_rate": 3.7846153846153847e-06,
      "loss": 0.0024,
      "step": 3162
    },
    {
      "epoch": 40.42172523961661,
      "grad_norm": 1.3650420904159546,
      "learning_rate": 3.7794871794871796e-06,
      "loss": 0.0067,
      "step": 3163
    },
    {
      "epoch": 40.43450479233227,
      "grad_norm": 1.9687597751617432,
      "learning_rate": 3.774358974358975e-06,
      "loss": 0.0151,
      "step": 3164
    },
    {
      "epoch": 40.447284345047926,
      "grad_norm": 0.23424799740314484,
      "learning_rate": 3.7692307692307694e-06,
      "loss": 0.0018,
      "step": 3165
    },
    {
      "epoch": 40.460063897763575,
      "grad_norm": 0.3177472651004791,
      "learning_rate": 3.7641025641025643e-06,
      "loss": 0.0025,
      "step": 3166
    },
    {
      "epoch": 40.47284345047923,
      "grad_norm": 0.22956353425979614,
      "learning_rate": 3.7589743589743592e-06,
      "loss": 0.0027,
      "step": 3167
    },
    {
      "epoch": 40.48562300319489,
      "grad_norm": 4.320156574249268,
      "learning_rate": 3.753846153846154e-06,
      "loss": 0.0166,
      "step": 3168
    },
    {
      "epoch": 40.498402555910545,
      "grad_norm": 0.34956955909729004,
      "learning_rate": 3.7487179487179495e-06,
      "loss": 0.003,
      "step": 3169
    },
    {
      "epoch": 40.5111821086262,
      "grad_norm": 0.33838948607444763,
      "learning_rate": 3.743589743589744e-06,
      "loss": 0.0031,
      "step": 3170
    },
    {
      "epoch": 40.52396166134185,
      "grad_norm": 0.42558255791664124,
      "learning_rate": 3.7384615384615384e-06,
      "loss": 0.0028,
      "step": 3171
    },
    {
      "epoch": 40.53674121405751,
      "grad_norm": 0.2204248309135437,
      "learning_rate": 3.7333333333333337e-06,
      "loss": 0.0019,
      "step": 3172
    },
    {
      "epoch": 40.54952076677316,
      "grad_norm": 0.4001370370388031,
      "learning_rate": 3.7282051282051286e-06,
      "loss": 0.0026,
      "step": 3173
    },
    {
      "epoch": 40.56230031948882,
      "grad_norm": 0.21183380484580994,
      "learning_rate": 3.723076923076923e-06,
      "loss": 0.0024,
      "step": 3174
    },
    {
      "epoch": 40.575079872204476,
      "grad_norm": 1.0086758136749268,
      "learning_rate": 3.7179487179487184e-06,
      "loss": 0.0052,
      "step": 3175
    },
    {
      "epoch": 40.587859424920126,
      "grad_norm": 2.1529321670532227,
      "learning_rate": 3.712820512820513e-06,
      "loss": 0.015,
      "step": 3176
    },
    {
      "epoch": 40.60063897763578,
      "grad_norm": 2.298342227935791,
      "learning_rate": 3.7076923076923082e-06,
      "loss": 0.0077,
      "step": 3177
    },
    {
      "epoch": 40.61341853035144,
      "grad_norm": 0.2786836326122284,
      "learning_rate": 3.702564102564103e-06,
      "loss": 0.0025,
      "step": 3178
    },
    {
      "epoch": 40.626198083067095,
      "grad_norm": 0.22754481434822083,
      "learning_rate": 3.6974358974358976e-06,
      "loss": 0.0022,
      "step": 3179
    },
    {
      "epoch": 40.638977635782744,
      "grad_norm": 2.1523077487945557,
      "learning_rate": 3.692307692307693e-06,
      "loss": 0.0126,
      "step": 3180
    },
    {
      "epoch": 40.6517571884984,
      "grad_norm": 0.2736090123653412,
      "learning_rate": 3.6871794871794874e-06,
      "loss": 0.0023,
      "step": 3181
    },
    {
      "epoch": 40.66453674121406,
      "grad_norm": 0.23136316239833832,
      "learning_rate": 3.6820512820512823e-06,
      "loss": 0.0026,
      "step": 3182
    },
    {
      "epoch": 40.677316293929714,
      "grad_norm": 0.3643839657306671,
      "learning_rate": 3.676923076923077e-06,
      "loss": 0.0034,
      "step": 3183
    },
    {
      "epoch": 40.69009584664537,
      "grad_norm": 0.3675735890865326,
      "learning_rate": 3.671794871794872e-06,
      "loss": 0.0022,
      "step": 3184
    },
    {
      "epoch": 40.70287539936102,
      "grad_norm": 3.977429151535034,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.0171,
      "step": 3185
    },
    {
      "epoch": 40.715654952076676,
      "grad_norm": 0.21517342329025269,
      "learning_rate": 3.661538461538462e-06,
      "loss": 0.0022,
      "step": 3186
    },
    {
      "epoch": 40.72843450479233,
      "grad_norm": 0.25581324100494385,
      "learning_rate": 3.656410256410257e-06,
      "loss": 0.0025,
      "step": 3187
    },
    {
      "epoch": 40.74121405750799,
      "grad_norm": 0.5155754089355469,
      "learning_rate": 3.6512820512820517e-06,
      "loss": 0.0031,
      "step": 3188
    },
    {
      "epoch": 40.753993610223645,
      "grad_norm": 0.24920091032981873,
      "learning_rate": 3.6461538461538466e-06,
      "loss": 0.0022,
      "step": 3189
    },
    {
      "epoch": 40.766773162939295,
      "grad_norm": 0.22267504036426544,
      "learning_rate": 3.641025641025641e-06,
      "loss": 0.0024,
      "step": 3190
    },
    {
      "epoch": 40.77955271565495,
      "grad_norm": 0.2679823040962219,
      "learning_rate": 3.6358974358974364e-06,
      "loss": 0.0029,
      "step": 3191
    },
    {
      "epoch": 40.79233226837061,
      "grad_norm": 1.3183385133743286,
      "learning_rate": 3.630769230769231e-06,
      "loss": 0.0065,
      "step": 3192
    },
    {
      "epoch": 40.805111821086264,
      "grad_norm": 0.2925851345062256,
      "learning_rate": 3.6256410256410258e-06,
      "loss": 0.0022,
      "step": 3193
    },
    {
      "epoch": 40.81789137380191,
      "grad_norm": 2.8205389976501465,
      "learning_rate": 3.620512820512821e-06,
      "loss": 0.0182,
      "step": 3194
    },
    {
      "epoch": 40.83067092651757,
      "grad_norm": 0.31817200779914856,
      "learning_rate": 3.6153846153846156e-06,
      "loss": 0.0025,
      "step": 3195
    },
    {
      "epoch": 40.843450479233226,
      "grad_norm": 0.19292470812797546,
      "learning_rate": 3.610256410256411e-06,
      "loss": 0.0023,
      "step": 3196
    },
    {
      "epoch": 40.85623003194888,
      "grad_norm": 0.25802552700042725,
      "learning_rate": 3.6051282051282054e-06,
      "loss": 0.0025,
      "step": 3197
    },
    {
      "epoch": 40.86900958466454,
      "grad_norm": 0.3933732807636261,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.0023,
      "step": 3198
    },
    {
      "epoch": 40.88178913738019,
      "grad_norm": 2.052565097808838,
      "learning_rate": 3.5948717948717956e-06,
      "loss": 0.0154,
      "step": 3199
    },
    {
      "epoch": 40.894568690095845,
      "grad_norm": 0.23817439377307892,
      "learning_rate": 3.58974358974359e-06,
      "loss": 0.0021,
      "step": 3200
    },
    {
      "epoch": 40.894568690095845,
      "eval_loss": 0.9943722486495972,
      "eval_runtime": 183.9313,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 3200
    },
    {
      "epoch": 40.9073482428115,
      "grad_norm": 0.36221644282341003,
      "learning_rate": 3.5846153846153845e-06,
      "loss": 0.0024,
      "step": 3201
    },
    {
      "epoch": 40.92012779552716,
      "grad_norm": 0.21248416602611542,
      "learning_rate": 3.57948717948718e-06,
      "loss": 0.0024,
      "step": 3202
    },
    {
      "epoch": 40.932907348242814,
      "grad_norm": 0.24060960114002228,
      "learning_rate": 3.5743589743589748e-06,
      "loss": 0.0021,
      "step": 3203
    },
    {
      "epoch": 40.945686900958464,
      "grad_norm": 2.167067289352417,
      "learning_rate": 3.5692307692307692e-06,
      "loss": 0.0132,
      "step": 3204
    },
    {
      "epoch": 40.95846645367412,
      "grad_norm": 0.2268885225057602,
      "learning_rate": 3.5641025641025646e-06,
      "loss": 0.0022,
      "step": 3205
    },
    {
      "epoch": 40.97124600638978,
      "grad_norm": 0.19051924347877502,
      "learning_rate": 3.558974358974359e-06,
      "loss": 0.0015,
      "step": 3206
    },
    {
      "epoch": 40.98402555910543,
      "grad_norm": 0.848506510257721,
      "learning_rate": 3.5538461538461544e-06,
      "loss": 0.0022,
      "step": 3207
    },
    {
      "epoch": 40.99680511182109,
      "grad_norm": 1.9270353317260742,
      "learning_rate": 3.5487179487179493e-06,
      "loss": 0.0075,
      "step": 3208
    },
    {
      "epoch": 41.00958466453674,
      "grad_norm": 2.1617424488067627,
      "learning_rate": 3.5435897435897437e-06,
      "loss": 0.0091,
      "step": 3209
    },
    {
      "epoch": 41.022364217252395,
      "grad_norm": 1.1429330110549927,
      "learning_rate": 3.538461538461539e-06,
      "loss": 0.0054,
      "step": 3210
    },
    {
      "epoch": 41.03514376996805,
      "grad_norm": 0.2642124593257904,
      "learning_rate": 3.5333333333333335e-06,
      "loss": 0.0032,
      "step": 3211
    },
    {
      "epoch": 41.04792332268371,
      "grad_norm": 2.530320167541504,
      "learning_rate": 3.5282051282051284e-06,
      "loss": 0.0096,
      "step": 3212
    },
    {
      "epoch": 41.06070287539936,
      "grad_norm": 2.0875837802886963,
      "learning_rate": 3.5230769230769233e-06,
      "loss": 0.0108,
      "step": 3213
    },
    {
      "epoch": 41.073482428115014,
      "grad_norm": 0.20744475722312927,
      "learning_rate": 3.5179487179487182e-06,
      "loss": 0.0016,
      "step": 3214
    },
    {
      "epoch": 41.08626198083067,
      "grad_norm": 0.17290177941322327,
      "learning_rate": 3.5128205128205127e-06,
      "loss": 0.0021,
      "step": 3215
    },
    {
      "epoch": 41.09904153354633,
      "grad_norm": 0.23610471189022064,
      "learning_rate": 3.507692307692308e-06,
      "loss": 0.002,
      "step": 3216
    },
    {
      "epoch": 41.11182108626198,
      "grad_norm": 0.978726863861084,
      "learning_rate": 3.502564102564103e-06,
      "loss": 0.0044,
      "step": 3217
    },
    {
      "epoch": 41.12460063897763,
      "grad_norm": 1.1839767694473267,
      "learning_rate": 3.497435897435898e-06,
      "loss": 0.0039,
      "step": 3218
    },
    {
      "epoch": 41.13738019169329,
      "grad_norm": 1.5846623182296753,
      "learning_rate": 3.4923076923076927e-06,
      "loss": 0.0127,
      "step": 3219
    },
    {
      "epoch": 41.150159744408946,
      "grad_norm": 0.3181246519088745,
      "learning_rate": 3.487179487179487e-06,
      "loss": 0.0026,
      "step": 3220
    },
    {
      "epoch": 41.1629392971246,
      "grad_norm": 0.21884626150131226,
      "learning_rate": 3.4820512820512825e-06,
      "loss": 0.0023,
      "step": 3221
    },
    {
      "epoch": 41.17571884984026,
      "grad_norm": 0.8826262354850769,
      "learning_rate": 3.476923076923077e-06,
      "loss": 0.0029,
      "step": 3222
    },
    {
      "epoch": 41.18849840255591,
      "grad_norm": 0.27310654520988464,
      "learning_rate": 3.471794871794872e-06,
      "loss": 0.0026,
      "step": 3223
    },
    {
      "epoch": 41.201277955271564,
      "grad_norm": 1.3389889001846313,
      "learning_rate": 3.4666666666666672e-06,
      "loss": 0.0049,
      "step": 3224
    },
    {
      "epoch": 41.21405750798722,
      "grad_norm": 2.1886544227600098,
      "learning_rate": 3.4615384615384617e-06,
      "loss": 0.0089,
      "step": 3225
    },
    {
      "epoch": 41.22683706070288,
      "grad_norm": 0.17433784902095795,
      "learning_rate": 3.456410256410257e-06,
      "loss": 0.0022,
      "step": 3226
    },
    {
      "epoch": 41.239616613418534,
      "grad_norm": 0.37125158309936523,
      "learning_rate": 3.4512820512820515e-06,
      "loss": 0.0019,
      "step": 3227
    },
    {
      "epoch": 41.25239616613418,
      "grad_norm": 0.21987852454185486,
      "learning_rate": 3.4461538461538464e-06,
      "loss": 0.0022,
      "step": 3228
    },
    {
      "epoch": 41.26517571884984,
      "grad_norm": 0.2506076395511627,
      "learning_rate": 3.4410256410256417e-06,
      "loss": 0.0032,
      "step": 3229
    },
    {
      "epoch": 41.277955271565496,
      "grad_norm": 0.24976013600826263,
      "learning_rate": 3.435897435897436e-06,
      "loss": 0.002,
      "step": 3230
    },
    {
      "epoch": 41.29073482428115,
      "grad_norm": 0.21577253937721252,
      "learning_rate": 3.4307692307692307e-06,
      "loss": 0.0024,
      "step": 3231
    },
    {
      "epoch": 41.3035143769968,
      "grad_norm": 0.2852231562137604,
      "learning_rate": 3.425641025641026e-06,
      "loss": 0.0027,
      "step": 3232
    },
    {
      "epoch": 41.31629392971246,
      "grad_norm": 0.240945965051651,
      "learning_rate": 3.420512820512821e-06,
      "loss": 0.0019,
      "step": 3233
    },
    {
      "epoch": 41.329073482428115,
      "grad_norm": 1.3859814405441284,
      "learning_rate": 3.4153846153846154e-06,
      "loss": 0.0069,
      "step": 3234
    },
    {
      "epoch": 41.34185303514377,
      "grad_norm": 0.23556867241859436,
      "learning_rate": 3.4102564102564107e-06,
      "loss": 0.0032,
      "step": 3235
    },
    {
      "epoch": 41.35463258785943,
      "grad_norm": 0.2779128849506378,
      "learning_rate": 3.405128205128205e-06,
      "loss": 0.0025,
      "step": 3236
    },
    {
      "epoch": 41.36741214057508,
      "grad_norm": 0.38275742530822754,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.0029,
      "step": 3237
    },
    {
      "epoch": 41.38019169329073,
      "grad_norm": 0.2573952078819275,
      "learning_rate": 3.3948717948717954e-06,
      "loss": 0.0023,
      "step": 3238
    },
    {
      "epoch": 41.39297124600639,
      "grad_norm": 0.5133835673332214,
      "learning_rate": 3.38974358974359e-06,
      "loss": 0.0077,
      "step": 3239
    },
    {
      "epoch": 41.405750798722046,
      "grad_norm": 0.2631109952926636,
      "learning_rate": 3.384615384615385e-06,
      "loss": 0.0023,
      "step": 3240
    },
    {
      "epoch": 41.4185303514377,
      "grad_norm": 0.1972748339176178,
      "learning_rate": 3.3794871794871797e-06,
      "loss": 0.0019,
      "step": 3241
    },
    {
      "epoch": 41.43130990415335,
      "grad_norm": 1.8707705736160278,
      "learning_rate": 3.3743589743589746e-06,
      "loss": 0.007,
      "step": 3242
    },
    {
      "epoch": 41.44408945686901,
      "grad_norm": 0.32073575258255005,
      "learning_rate": 3.3692307692307695e-06,
      "loss": 0.0026,
      "step": 3243
    },
    {
      "epoch": 41.456869009584665,
      "grad_norm": 1.5978692770004272,
      "learning_rate": 3.3641025641025644e-06,
      "loss": 0.0071,
      "step": 3244
    },
    {
      "epoch": 41.46964856230032,
      "grad_norm": 3.0393266677856445,
      "learning_rate": 3.358974358974359e-06,
      "loss": 0.0208,
      "step": 3245
    },
    {
      "epoch": 41.48242811501598,
      "grad_norm": 0.16218027472496033,
      "learning_rate": 3.353846153846154e-06,
      "loss": 0.0021,
      "step": 3246
    },
    {
      "epoch": 41.49520766773163,
      "grad_norm": 0.19654710590839386,
      "learning_rate": 3.348717948717949e-06,
      "loss": 0.0023,
      "step": 3247
    },
    {
      "epoch": 41.50798722044728,
      "grad_norm": 3.003314971923828,
      "learning_rate": 3.343589743589744e-06,
      "loss": 0.0086,
      "step": 3248
    },
    {
      "epoch": 41.52076677316294,
      "grad_norm": 0.26050227880477905,
      "learning_rate": 3.338461538461539e-06,
      "loss": 0.0026,
      "step": 3249
    },
    {
      "epoch": 41.533546325878596,
      "grad_norm": 1.9690760374069214,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0086,
      "step": 3250
    },
    {
      "epoch": 41.546325878594246,
      "grad_norm": 0.18360324203968048,
      "learning_rate": 3.3282051282051286e-06,
      "loss": 0.0021,
      "step": 3251
    },
    {
      "epoch": 41.5591054313099,
      "grad_norm": 0.41593506932258606,
      "learning_rate": 3.323076923076923e-06,
      "loss": 0.0037,
      "step": 3252
    },
    {
      "epoch": 41.57188498402556,
      "grad_norm": 0.26756003499031067,
      "learning_rate": 3.317948717948718e-06,
      "loss": 0.0026,
      "step": 3253
    },
    {
      "epoch": 41.584664536741215,
      "grad_norm": 2.4918761253356934,
      "learning_rate": 3.3128205128205133e-06,
      "loss": 0.005,
      "step": 3254
    },
    {
      "epoch": 41.59744408945687,
      "grad_norm": 0.1668577343225479,
      "learning_rate": 3.307692307692308e-06,
      "loss": 0.0013,
      "step": 3255
    },
    {
      "epoch": 41.61022364217252,
      "grad_norm": 0.19826513528823853,
      "learning_rate": 3.302564102564103e-06,
      "loss": 0.002,
      "step": 3256
    },
    {
      "epoch": 41.62300319488818,
      "grad_norm": 10.235542297363281,
      "learning_rate": 3.2974358974358976e-06,
      "loss": 0.035,
      "step": 3257
    },
    {
      "epoch": 41.635782747603834,
      "grad_norm": 0.4807860553264618,
      "learning_rate": 3.2923076923076925e-06,
      "loss": 0.0027,
      "step": 3258
    },
    {
      "epoch": 41.64856230031949,
      "grad_norm": 0.39432165026664734,
      "learning_rate": 3.287179487179488e-06,
      "loss": 0.0034,
      "step": 3259
    },
    {
      "epoch": 41.66134185303515,
      "grad_norm": 3.0132107734680176,
      "learning_rate": 3.2820512820512823e-06,
      "loss": 0.0077,
      "step": 3260
    },
    {
      "epoch": 41.674121405750796,
      "grad_norm": 0.17874084413051605,
      "learning_rate": 3.276923076923077e-06,
      "loss": 0.0012,
      "step": 3261
    },
    {
      "epoch": 41.68690095846645,
      "grad_norm": 1.4562616348266602,
      "learning_rate": 3.271794871794872e-06,
      "loss": 0.0105,
      "step": 3262
    },
    {
      "epoch": 41.69968051118211,
      "grad_norm": 0.5152537822723389,
      "learning_rate": 3.266666666666667e-06,
      "loss": 0.0028,
      "step": 3263
    },
    {
      "epoch": 41.712460063897765,
      "grad_norm": 0.23496030271053314,
      "learning_rate": 3.2615384615384615e-06,
      "loss": 0.0025,
      "step": 3264
    },
    {
      "epoch": 41.72523961661342,
      "grad_norm": 1.5751100778579712,
      "learning_rate": 3.256410256410257e-06,
      "loss": 0.0078,
      "step": 3265
    },
    {
      "epoch": 41.73801916932907,
      "grad_norm": 1.7696713209152222,
      "learning_rate": 3.2512820512820513e-06,
      "loss": 0.0057,
      "step": 3266
    },
    {
      "epoch": 41.75079872204473,
      "grad_norm": 2.3506906032562256,
      "learning_rate": 3.2461538461538466e-06,
      "loss": 0.0149,
      "step": 3267
    },
    {
      "epoch": 41.763578274760384,
      "grad_norm": 0.7308444380760193,
      "learning_rate": 3.2410256410256415e-06,
      "loss": 0.0053,
      "step": 3268
    },
    {
      "epoch": 41.77635782747604,
      "grad_norm": 0.8668710589408875,
      "learning_rate": 3.235897435897436e-06,
      "loss": 0.006,
      "step": 3269
    },
    {
      "epoch": 41.78913738019169,
      "grad_norm": 0.2136469930410385,
      "learning_rate": 3.2307692307692313e-06,
      "loss": 0.0016,
      "step": 3270
    },
    {
      "epoch": 41.801916932907346,
      "grad_norm": 1.6871494054794312,
      "learning_rate": 3.2256410256410258e-06,
      "loss": 0.0068,
      "step": 3271
    },
    {
      "epoch": 41.814696485623,
      "grad_norm": 0.23570038378238678,
      "learning_rate": 3.2205128205128207e-06,
      "loss": 0.003,
      "step": 3272
    },
    {
      "epoch": 41.82747603833866,
      "grad_norm": 4.999492168426514,
      "learning_rate": 3.2153846153846156e-06,
      "loss": 0.0216,
      "step": 3273
    },
    {
      "epoch": 41.840255591054316,
      "grad_norm": 0.1835344433784485,
      "learning_rate": 3.2102564102564105e-06,
      "loss": 0.0017,
      "step": 3274
    },
    {
      "epoch": 41.853035143769965,
      "grad_norm": 1.509971261024475,
      "learning_rate": 3.205128205128206e-06,
      "loss": 0.008,
      "step": 3275
    },
    {
      "epoch": 41.86581469648562,
      "grad_norm": 0.3104620575904846,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.003,
      "step": 3276
    },
    {
      "epoch": 41.87859424920128,
      "grad_norm": 0.21908356249332428,
      "learning_rate": 3.194871794871795e-06,
      "loss": 0.0024,
      "step": 3277
    },
    {
      "epoch": 41.891373801916934,
      "grad_norm": 0.40520310401916504,
      "learning_rate": 3.18974358974359e-06,
      "loss": 0.0027,
      "step": 3278
    },
    {
      "epoch": 41.90415335463259,
      "grad_norm": 0.20701780915260315,
      "learning_rate": 3.184615384615385e-06,
      "loss": 0.002,
      "step": 3279
    },
    {
      "epoch": 41.91693290734824,
      "grad_norm": 0.21897658705711365,
      "learning_rate": 3.1794871794871795e-06,
      "loss": 0.0023,
      "step": 3280
    },
    {
      "epoch": 41.9297124600639,
      "grad_norm": 0.24286028742790222,
      "learning_rate": 3.1743589743589748e-06,
      "loss": 0.0025,
      "step": 3281
    },
    {
      "epoch": 41.94249201277955,
      "grad_norm": 0.25564444065093994,
      "learning_rate": 3.1692307692307693e-06,
      "loss": 0.002,
      "step": 3282
    },
    {
      "epoch": 41.95527156549521,
      "grad_norm": 0.2859918773174286,
      "learning_rate": 3.164102564102564e-06,
      "loss": 0.0021,
      "step": 3283
    },
    {
      "epoch": 41.968051118210866,
      "grad_norm": 0.4360431134700775,
      "learning_rate": 3.1589743589743595e-06,
      "loss": 0.0029,
      "step": 3284
    },
    {
      "epoch": 41.980830670926515,
      "grad_norm": 0.1724979281425476,
      "learning_rate": 3.153846153846154e-06,
      "loss": 0.0017,
      "step": 3285
    },
    {
      "epoch": 41.99361022364217,
      "grad_norm": 3.4681003093719482,
      "learning_rate": 3.1487179487179493e-06,
      "loss": 0.0175,
      "step": 3286
    },
    {
      "epoch": 42.00638977635783,
      "grad_norm": 0.25220033526420593,
      "learning_rate": 3.1435897435897437e-06,
      "loss": 0.0023,
      "step": 3287
    },
    {
      "epoch": 42.019169329073485,
      "grad_norm": 0.44308167695999146,
      "learning_rate": 3.1384615384615386e-06,
      "loss": 0.003,
      "step": 3288
    },
    {
      "epoch": 42.031948881789134,
      "grad_norm": 0.23785103857517242,
      "learning_rate": 3.133333333333334e-06,
      "loss": 0.0025,
      "step": 3289
    },
    {
      "epoch": 42.04472843450479,
      "grad_norm": 0.21543189883232117,
      "learning_rate": 3.1282051282051284e-06,
      "loss": 0.0022,
      "step": 3290
    },
    {
      "epoch": 42.05750798722045,
      "grad_norm": 0.1751144528388977,
      "learning_rate": 3.123076923076923e-06,
      "loss": 0.002,
      "step": 3291
    },
    {
      "epoch": 42.0702875399361,
      "grad_norm": 1.3925294876098633,
      "learning_rate": 3.1179487179487182e-06,
      "loss": 0.0082,
      "step": 3292
    },
    {
      "epoch": 42.08306709265176,
      "grad_norm": 0.1362583190202713,
      "learning_rate": 3.112820512820513e-06,
      "loss": 0.0015,
      "step": 3293
    },
    {
      "epoch": 42.09584664536741,
      "grad_norm": 0.21337972581386566,
      "learning_rate": 3.1076923076923076e-06,
      "loss": 0.0024,
      "step": 3294
    },
    {
      "epoch": 42.108626198083066,
      "grad_norm": 1.0860240459442139,
      "learning_rate": 3.102564102564103e-06,
      "loss": 0.0059,
      "step": 3295
    },
    {
      "epoch": 42.12140575079872,
      "grad_norm": 0.9178386926651001,
      "learning_rate": 3.0974358974358974e-06,
      "loss": 0.0031,
      "step": 3296
    },
    {
      "epoch": 42.13418530351438,
      "grad_norm": 0.21546006202697754,
      "learning_rate": 3.0923076923076927e-06,
      "loss": 0.0024,
      "step": 3297
    },
    {
      "epoch": 42.146964856230035,
      "grad_norm": 1.8423643112182617,
      "learning_rate": 3.0871794871794876e-06,
      "loss": 0.0079,
      "step": 3298
    },
    {
      "epoch": 42.159744408945684,
      "grad_norm": 0.3426179587841034,
      "learning_rate": 3.082051282051282e-06,
      "loss": 0.0028,
      "step": 3299
    },
    {
      "epoch": 42.17252396166134,
      "grad_norm": 0.2042442262172699,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.002,
      "step": 3300
    },
    {
      "epoch": 42.17252396166134,
      "eval_loss": 0.9923915863037109,
      "eval_runtime": 183.7945,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 3300
    },
    {
      "epoch": 42.185303514377,
      "grad_norm": 0.3152225911617279,
      "learning_rate": 3.071794871794872e-06,
      "loss": 0.0022,
      "step": 3301
    },
    {
      "epoch": 42.198083067092654,
      "grad_norm": 0.25565582513809204,
      "learning_rate": 3.066666666666667e-06,
      "loss": 0.0024,
      "step": 3302
    },
    {
      "epoch": 42.21086261980831,
      "grad_norm": 3.9883928298950195,
      "learning_rate": 3.0615384615384617e-06,
      "loss": 0.0269,
      "step": 3303
    },
    {
      "epoch": 42.22364217252396,
      "grad_norm": 0.37186405062675476,
      "learning_rate": 3.0564102564102566e-06,
      "loss": 0.0027,
      "step": 3304
    },
    {
      "epoch": 42.236421725239616,
      "grad_norm": 0.6624550819396973,
      "learning_rate": 3.051282051282052e-06,
      "loss": 0.0031,
      "step": 3305
    },
    {
      "epoch": 42.24920127795527,
      "grad_norm": 0.5737404227256775,
      "learning_rate": 3.0461538461538464e-06,
      "loss": 0.0017,
      "step": 3306
    },
    {
      "epoch": 42.26198083067093,
      "grad_norm": 0.8662362098693848,
      "learning_rate": 3.0410256410256413e-06,
      "loss": 0.0042,
      "step": 3307
    },
    {
      "epoch": 42.27476038338658,
      "grad_norm": 0.6699799299240112,
      "learning_rate": 3.035897435897436e-06,
      "loss": 0.0092,
      "step": 3308
    },
    {
      "epoch": 42.287539936102235,
      "grad_norm": 0.6369069218635559,
      "learning_rate": 3.030769230769231e-06,
      "loss": 0.0044,
      "step": 3309
    },
    {
      "epoch": 42.30031948881789,
      "grad_norm": 1.9807689189910889,
      "learning_rate": 3.0256410256410256e-06,
      "loss": 0.0099,
      "step": 3310
    },
    {
      "epoch": 42.31309904153355,
      "grad_norm": 0.16607657074928284,
      "learning_rate": 3.020512820512821e-06,
      "loss": 0.0017,
      "step": 3311
    },
    {
      "epoch": 42.325878594249204,
      "grad_norm": 0.35444679856300354,
      "learning_rate": 3.0153846153846154e-06,
      "loss": 0.003,
      "step": 3312
    },
    {
      "epoch": 42.33865814696485,
      "grad_norm": 0.22463780641555786,
      "learning_rate": 3.0102564102564103e-06,
      "loss": 0.0021,
      "step": 3313
    },
    {
      "epoch": 42.35143769968051,
      "grad_norm": 0.20150989294052124,
      "learning_rate": 3.0051282051282056e-06,
      "loss": 0.0023,
      "step": 3314
    },
    {
      "epoch": 42.364217252396166,
      "grad_norm": 0.24138055741786957,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 3315
    },
    {
      "epoch": 42.37699680511182,
      "grad_norm": 0.2151312679052353,
      "learning_rate": 2.9948717948717954e-06,
      "loss": 0.002,
      "step": 3316
    },
    {
      "epoch": 42.38977635782748,
      "grad_norm": 0.24440868198871613,
      "learning_rate": 2.98974358974359e-06,
      "loss": 0.0019,
      "step": 3317
    },
    {
      "epoch": 42.40255591054313,
      "grad_norm": 0.18149599432945251,
      "learning_rate": 2.9846153846153848e-06,
      "loss": 0.0018,
      "step": 3318
    },
    {
      "epoch": 42.415335463258785,
      "grad_norm": 0.2243649810552597,
      "learning_rate": 2.97948717948718e-06,
      "loss": 0.0016,
      "step": 3319
    },
    {
      "epoch": 42.42811501597444,
      "grad_norm": 1.1802103519439697,
      "learning_rate": 2.9743589743589746e-06,
      "loss": 0.0032,
      "step": 3320
    },
    {
      "epoch": 42.4408945686901,
      "grad_norm": 3.019127607345581,
      "learning_rate": 2.969230769230769e-06,
      "loss": 0.0113,
      "step": 3321
    },
    {
      "epoch": 42.453674121405754,
      "grad_norm": 2.25211763381958,
      "learning_rate": 2.9641025641025644e-06,
      "loss": 0.0111,
      "step": 3322
    },
    {
      "epoch": 42.466453674121404,
      "grad_norm": 0.3121747374534607,
      "learning_rate": 2.9589743589743593e-06,
      "loss": 0.0029,
      "step": 3323
    },
    {
      "epoch": 42.47923322683706,
      "grad_norm": 2.901637315750122,
      "learning_rate": 2.953846153846154e-06,
      "loss": 0.0125,
      "step": 3324
    },
    {
      "epoch": 42.49201277955272,
      "grad_norm": 0.2974953353404999,
      "learning_rate": 2.948717948717949e-06,
      "loss": 0.003,
      "step": 3325
    },
    {
      "epoch": 42.50479233226837,
      "grad_norm": 0.270323783159256,
      "learning_rate": 2.9435897435897435e-06,
      "loss": 0.0023,
      "step": 3326
    },
    {
      "epoch": 42.51757188498402,
      "grad_norm": 0.18607768416404724,
      "learning_rate": 2.938461538461539e-06,
      "loss": 0.0016,
      "step": 3327
    },
    {
      "epoch": 42.53035143769968,
      "grad_norm": 0.16889779269695282,
      "learning_rate": 2.9333333333333338e-06,
      "loss": 0.0015,
      "step": 3328
    },
    {
      "epoch": 42.543130990415335,
      "grad_norm": 0.2744269073009491,
      "learning_rate": 2.9282051282051282e-06,
      "loss": 0.0026,
      "step": 3329
    },
    {
      "epoch": 42.55591054313099,
      "grad_norm": 0.29199185967445374,
      "learning_rate": 2.9230769230769236e-06,
      "loss": 0.0022,
      "step": 3330
    },
    {
      "epoch": 42.56869009584665,
      "grad_norm": 3.15661358833313,
      "learning_rate": 2.917948717948718e-06,
      "loss": 0.0132,
      "step": 3331
    },
    {
      "epoch": 42.5814696485623,
      "grad_norm": 0.2181830108165741,
      "learning_rate": 2.912820512820513e-06,
      "loss": 0.0026,
      "step": 3332
    },
    {
      "epoch": 42.594249201277954,
      "grad_norm": 0.17178262770175934,
      "learning_rate": 2.907692307692308e-06,
      "loss": 0.0021,
      "step": 3333
    },
    {
      "epoch": 42.60702875399361,
      "grad_norm": 0.21246619522571564,
      "learning_rate": 2.9025641025641027e-06,
      "loss": 0.0017,
      "step": 3334
    },
    {
      "epoch": 42.61980830670927,
      "grad_norm": 0.23296082019805908,
      "learning_rate": 2.897435897435898e-06,
      "loss": 0.002,
      "step": 3335
    },
    {
      "epoch": 42.63258785942492,
      "grad_norm": 2.3438727855682373,
      "learning_rate": 2.8923076923076925e-06,
      "loss": 0.0161,
      "step": 3336
    },
    {
      "epoch": 42.64536741214057,
      "grad_norm": 0.23535417020320892,
      "learning_rate": 2.8871794871794874e-06,
      "loss": 0.0023,
      "step": 3337
    },
    {
      "epoch": 42.65814696485623,
      "grad_norm": 0.19845890998840332,
      "learning_rate": 2.8820512820512823e-06,
      "loss": 0.0023,
      "step": 3338
    },
    {
      "epoch": 42.670926517571885,
      "grad_norm": 0.23049385845661163,
      "learning_rate": 2.8769230769230772e-06,
      "loss": 0.0018,
      "step": 3339
    },
    {
      "epoch": 42.68370607028754,
      "grad_norm": 1.5415147542953491,
      "learning_rate": 2.8717948717948717e-06,
      "loss": 0.0076,
      "step": 3340
    },
    {
      "epoch": 42.6964856230032,
      "grad_norm": 1.1236743927001953,
      "learning_rate": 2.866666666666667e-06,
      "loss": 0.0049,
      "step": 3341
    },
    {
      "epoch": 42.70926517571885,
      "grad_norm": 3.2558176517486572,
      "learning_rate": 2.8615384615384615e-06,
      "loss": 0.0125,
      "step": 3342
    },
    {
      "epoch": 42.722044728434504,
      "grad_norm": 0.40388306975364685,
      "learning_rate": 2.8564102564102564e-06,
      "loss": 0.0027,
      "step": 3343
    },
    {
      "epoch": 42.73482428115016,
      "grad_norm": 0.5524995923042297,
      "learning_rate": 2.8512820512820517e-06,
      "loss": 0.0028,
      "step": 3344
    },
    {
      "epoch": 42.74760383386582,
      "grad_norm": 0.34669360518455505,
      "learning_rate": 2.846153846153846e-06,
      "loss": 0.003,
      "step": 3345
    },
    {
      "epoch": 42.760383386581466,
      "grad_norm": 1.4820116758346558,
      "learning_rate": 2.8410256410256415e-06,
      "loss": 0.0072,
      "step": 3346
    },
    {
      "epoch": 42.77316293929712,
      "grad_norm": 0.27356067299842834,
      "learning_rate": 2.835897435897436e-06,
      "loss": 0.0027,
      "step": 3347
    },
    {
      "epoch": 42.78594249201278,
      "grad_norm": 2.8262336254119873,
      "learning_rate": 2.830769230769231e-06,
      "loss": 0.009,
      "step": 3348
    },
    {
      "epoch": 42.798722044728436,
      "grad_norm": 0.6990281939506531,
      "learning_rate": 2.8256410256410262e-06,
      "loss": 0.005,
      "step": 3349
    },
    {
      "epoch": 42.81150159744409,
      "grad_norm": 0.2962161898612976,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 0.0023,
      "step": 3350
    },
    {
      "epoch": 42.82428115015974,
      "grad_norm": 1.1499954462051392,
      "learning_rate": 2.815384615384615e-06,
      "loss": 0.0036,
      "step": 3351
    },
    {
      "epoch": 42.8370607028754,
      "grad_norm": 2.363464117050171,
      "learning_rate": 2.8102564102564105e-06,
      "loss": 0.0114,
      "step": 3352
    },
    {
      "epoch": 42.849840255591054,
      "grad_norm": 2.9876015186309814,
      "learning_rate": 2.8051282051282054e-06,
      "loss": 0.0117,
      "step": 3353
    },
    {
      "epoch": 42.86261980830671,
      "grad_norm": 1.9356184005737305,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.0053,
      "step": 3354
    },
    {
      "epoch": 42.87539936102237,
      "grad_norm": 0.16695638000965118,
      "learning_rate": 2.794871794871795e-06,
      "loss": 0.0018,
      "step": 3355
    },
    {
      "epoch": 42.88817891373802,
      "grad_norm": 1.636541724205017,
      "learning_rate": 2.7897435897435897e-06,
      "loss": 0.0038,
      "step": 3356
    },
    {
      "epoch": 42.90095846645367,
      "grad_norm": 0.3252117335796356,
      "learning_rate": 2.784615384615385e-06,
      "loss": 0.0028,
      "step": 3357
    },
    {
      "epoch": 42.91373801916933,
      "grad_norm": 3.5567755699157715,
      "learning_rate": 2.77948717948718e-06,
      "loss": 0.0273,
      "step": 3358
    },
    {
      "epoch": 42.926517571884986,
      "grad_norm": 0.3256838321685791,
      "learning_rate": 2.7743589743589744e-06,
      "loss": 0.0027,
      "step": 3359
    },
    {
      "epoch": 42.93929712460064,
      "grad_norm": 0.17103208601474762,
      "learning_rate": 2.7692307692307697e-06,
      "loss": 0.0016,
      "step": 3360
    },
    {
      "epoch": 42.95207667731629,
      "grad_norm": 1.3459094762802124,
      "learning_rate": 2.764102564102564e-06,
      "loss": 0.0058,
      "step": 3361
    },
    {
      "epoch": 42.96485623003195,
      "grad_norm": 0.1608177274465561,
      "learning_rate": 2.758974358974359e-06,
      "loss": 0.0019,
      "step": 3362
    },
    {
      "epoch": 42.977635782747605,
      "grad_norm": 0.2738124430179596,
      "learning_rate": 2.753846153846154e-06,
      "loss": 0.0026,
      "step": 3363
    },
    {
      "epoch": 42.99041533546326,
      "grad_norm": 0.4253382980823517,
      "learning_rate": 2.748717948717949e-06,
      "loss": 0.0035,
      "step": 3364
    },
    {
      "epoch": 43.00319488817891,
      "grad_norm": 0.41924622654914856,
      "learning_rate": 2.743589743589744e-06,
      "loss": 0.0036,
      "step": 3365
    },
    {
      "epoch": 43.01597444089457,
      "grad_norm": 0.18903520703315735,
      "learning_rate": 2.7384615384615387e-06,
      "loss": 0.0019,
      "step": 3366
    },
    {
      "epoch": 43.02875399361022,
      "grad_norm": 0.20113691687583923,
      "learning_rate": 2.7333333333333336e-06,
      "loss": 0.002,
      "step": 3367
    },
    {
      "epoch": 43.04153354632588,
      "grad_norm": 0.8627514839172363,
      "learning_rate": 2.7282051282051285e-06,
      "loss": 0.0041,
      "step": 3368
    },
    {
      "epoch": 43.054313099041536,
      "grad_norm": 1.7973607778549194,
      "learning_rate": 2.7230769230769234e-06,
      "loss": 0.0083,
      "step": 3369
    },
    {
      "epoch": 43.067092651757186,
      "grad_norm": 0.3085075318813324,
      "learning_rate": 2.717948717948718e-06,
      "loss": 0.0022,
      "step": 3370
    },
    {
      "epoch": 43.07987220447284,
      "grad_norm": 0.19819965958595276,
      "learning_rate": 2.712820512820513e-06,
      "loss": 0.002,
      "step": 3371
    },
    {
      "epoch": 43.0926517571885,
      "grad_norm": 0.23764221370220184,
      "learning_rate": 2.7076923076923076e-06,
      "loss": 0.0018,
      "step": 3372
    },
    {
      "epoch": 43.105431309904155,
      "grad_norm": 0.20486418902873993,
      "learning_rate": 2.7025641025641025e-06,
      "loss": 0.002,
      "step": 3373
    },
    {
      "epoch": 43.11821086261981,
      "grad_norm": 0.6274747252464294,
      "learning_rate": 2.697435897435898e-06,
      "loss": 0.0034,
      "step": 3374
    },
    {
      "epoch": 43.13099041533546,
      "grad_norm": 0.25211164355278015,
      "learning_rate": 2.6923076923076923e-06,
      "loss": 0.0024,
      "step": 3375
    },
    {
      "epoch": 43.14376996805112,
      "grad_norm": 2.6378355026245117,
      "learning_rate": 2.6871794871794877e-06,
      "loss": 0.0107,
      "step": 3376
    },
    {
      "epoch": 43.156549520766774,
      "grad_norm": 1.5214442014694214,
      "learning_rate": 2.682051282051282e-06,
      "loss": 0.0045,
      "step": 3377
    },
    {
      "epoch": 43.16932907348243,
      "grad_norm": 2.6448490619659424,
      "learning_rate": 2.676923076923077e-06,
      "loss": 0.0141,
      "step": 3378
    },
    {
      "epoch": 43.18210862619808,
      "grad_norm": 1.736473560333252,
      "learning_rate": 2.6717948717948724e-06,
      "loss": 0.0044,
      "step": 3379
    },
    {
      "epoch": 43.194888178913736,
      "grad_norm": 6.5054121017456055,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0107,
      "step": 3380
    },
    {
      "epoch": 43.20766773162939,
      "grad_norm": 2.9443705081939697,
      "learning_rate": 2.6615384615384613e-06,
      "loss": 0.0145,
      "step": 3381
    },
    {
      "epoch": 43.22044728434505,
      "grad_norm": 0.2461695671081543,
      "learning_rate": 2.6564102564102566e-06,
      "loss": 0.002,
      "step": 3382
    },
    {
      "epoch": 43.233226837060705,
      "grad_norm": 0.21339628100395203,
      "learning_rate": 2.6512820512820515e-06,
      "loss": 0.0015,
      "step": 3383
    },
    {
      "epoch": 43.246006389776355,
      "grad_norm": 0.12452192604541779,
      "learning_rate": 2.6461538461538464e-06,
      "loss": 0.0013,
      "step": 3384
    },
    {
      "epoch": 43.25878594249201,
      "grad_norm": 0.18087056279182434,
      "learning_rate": 2.6410256410256413e-06,
      "loss": 0.0017,
      "step": 3385
    },
    {
      "epoch": 43.27156549520767,
      "grad_norm": 2.1488447189331055,
      "learning_rate": 2.635897435897436e-06,
      "loss": 0.0067,
      "step": 3386
    },
    {
      "epoch": 43.284345047923324,
      "grad_norm": 0.18616983294487,
      "learning_rate": 2.630769230769231e-06,
      "loss": 0.0017,
      "step": 3387
    },
    {
      "epoch": 43.29712460063898,
      "grad_norm": 0.15523196756839752,
      "learning_rate": 2.625641025641026e-06,
      "loss": 0.0016,
      "step": 3388
    },
    {
      "epoch": 43.30990415335463,
      "grad_norm": 0.16610245406627655,
      "learning_rate": 2.6205128205128205e-06,
      "loss": 0.0019,
      "step": 3389
    },
    {
      "epoch": 43.322683706070286,
      "grad_norm": 0.2482658475637436,
      "learning_rate": 2.615384615384616e-06,
      "loss": 0.0026,
      "step": 3390
    },
    {
      "epoch": 43.33546325878594,
      "grad_norm": 0.20092348754405975,
      "learning_rate": 2.6102564102564103e-06,
      "loss": 0.0021,
      "step": 3391
    },
    {
      "epoch": 43.3482428115016,
      "grad_norm": 0.30072319507598877,
      "learning_rate": 2.605128205128205e-06,
      "loss": 0.0022,
      "step": 3392
    },
    {
      "epoch": 43.361022364217256,
      "grad_norm": 1.7357025146484375,
      "learning_rate": 2.6e-06,
      "loss": 0.01,
      "step": 3393
    },
    {
      "epoch": 43.373801916932905,
      "grad_norm": 0.37861010432243347,
      "learning_rate": 2.594871794871795e-06,
      "loss": 0.0042,
      "step": 3394
    },
    {
      "epoch": 43.38658146964856,
      "grad_norm": 0.26171085238456726,
      "learning_rate": 2.5897435897435903e-06,
      "loss": 0.0024,
      "step": 3395
    },
    {
      "epoch": 43.39936102236422,
      "grad_norm": 0.1893923431634903,
      "learning_rate": 2.584615384615385e-06,
      "loss": 0.0023,
      "step": 3396
    },
    {
      "epoch": 43.412140575079874,
      "grad_norm": 0.19825433194637299,
      "learning_rate": 2.5794871794871797e-06,
      "loss": 0.0015,
      "step": 3397
    },
    {
      "epoch": 43.424920127795524,
      "grad_norm": 0.17092125117778778,
      "learning_rate": 2.5743589743589746e-06,
      "loss": 0.0021,
      "step": 3398
    },
    {
      "epoch": 43.43769968051118,
      "grad_norm": 0.857297420501709,
      "learning_rate": 2.5692307692307695e-06,
      "loss": 0.0052,
      "step": 3399
    },
    {
      "epoch": 43.45047923322684,
      "grad_norm": 1.8602384328842163,
      "learning_rate": 2.564102564102564e-06,
      "loss": 0.0079,
      "step": 3400
    },
    {
      "epoch": 43.45047923322684,
      "eval_loss": 1.001037836074829,
      "eval_runtime": 183.7653,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 3400
    },
    {
      "epoch": 43.46325878594249,
      "grad_norm": 2.684396743774414,
      "learning_rate": 2.5589743589743593e-06,
      "loss": 0.0137,
      "step": 3401
    },
    {
      "epoch": 43.47603833865815,
      "grad_norm": 0.7329052090644836,
      "learning_rate": 2.5538461538461538e-06,
      "loss": 0.0028,
      "step": 3402
    },
    {
      "epoch": 43.4888178913738,
      "grad_norm": 3.4295055866241455,
      "learning_rate": 2.548717948717949e-06,
      "loss": 0.0225,
      "step": 3403
    },
    {
      "epoch": 43.501597444089455,
      "grad_norm": 0.20623815059661865,
      "learning_rate": 2.543589743589744e-06,
      "loss": 0.0024,
      "step": 3404
    },
    {
      "epoch": 43.51437699680511,
      "grad_norm": 0.20949018001556396,
      "learning_rate": 2.5384615384615385e-06,
      "loss": 0.0023,
      "step": 3405
    },
    {
      "epoch": 43.52715654952077,
      "grad_norm": 0.29363885521888733,
      "learning_rate": 2.5333333333333338e-06,
      "loss": 0.003,
      "step": 3406
    },
    {
      "epoch": 43.539936102236425,
      "grad_norm": 0.44644200801849365,
      "learning_rate": 2.5282051282051283e-06,
      "loss": 0.0023,
      "step": 3407
    },
    {
      "epoch": 43.552715654952074,
      "grad_norm": 10.232358932495117,
      "learning_rate": 2.523076923076923e-06,
      "loss": 0.0425,
      "step": 3408
    },
    {
      "epoch": 43.56549520766773,
      "grad_norm": 0.23224446177482605,
      "learning_rate": 2.5179487179487185e-06,
      "loss": 0.0024,
      "step": 3409
    },
    {
      "epoch": 43.57827476038339,
      "grad_norm": 0.3602463901042938,
      "learning_rate": 2.512820512820513e-06,
      "loss": 0.0029,
      "step": 3410
    },
    {
      "epoch": 43.59105431309904,
      "grad_norm": 0.2921648323535919,
      "learning_rate": 2.507692307692308e-06,
      "loss": 0.0021,
      "step": 3411
    },
    {
      "epoch": 43.6038338658147,
      "grad_norm": 0.21882840991020203,
      "learning_rate": 2.5025641025641028e-06,
      "loss": 0.002,
      "step": 3412
    },
    {
      "epoch": 43.61661341853035,
      "grad_norm": 0.22644449770450592,
      "learning_rate": 2.4974358974358977e-06,
      "loss": 0.0018,
      "step": 3413
    },
    {
      "epoch": 43.629392971246006,
      "grad_norm": 3.2386162281036377,
      "learning_rate": 2.4923076923076926e-06,
      "loss": 0.0214,
      "step": 3414
    },
    {
      "epoch": 43.64217252396166,
      "grad_norm": 0.6819486618041992,
      "learning_rate": 2.4871794871794875e-06,
      "loss": 0.0049,
      "step": 3415
    },
    {
      "epoch": 43.65495207667732,
      "grad_norm": 2.0926530361175537,
      "learning_rate": 2.4820512820512824e-06,
      "loss": 0.0069,
      "step": 3416
    },
    {
      "epoch": 43.66773162939297,
      "grad_norm": 4.31627082824707,
      "learning_rate": 2.4769230769230773e-06,
      "loss": 0.0065,
      "step": 3417
    },
    {
      "epoch": 43.680511182108624,
      "grad_norm": 2.114065647125244,
      "learning_rate": 2.471794871794872e-06,
      "loss": 0.0089,
      "step": 3418
    },
    {
      "epoch": 43.69329073482428,
      "grad_norm": 0.32869184017181396,
      "learning_rate": 2.466666666666667e-06,
      "loss": 0.0028,
      "step": 3419
    },
    {
      "epoch": 43.70607028753994,
      "grad_norm": 0.21543429791927338,
      "learning_rate": 2.461538461538462e-06,
      "loss": 0.0023,
      "step": 3420
    },
    {
      "epoch": 43.718849840255594,
      "grad_norm": 0.31127381324768066,
      "learning_rate": 2.4564102564102564e-06,
      "loss": 0.0033,
      "step": 3421
    },
    {
      "epoch": 43.73162939297124,
      "grad_norm": 1.919940710067749,
      "learning_rate": 2.4512820512820513e-06,
      "loss": 0.006,
      "step": 3422
    },
    {
      "epoch": 43.7444089456869,
      "grad_norm": 0.20285959541797638,
      "learning_rate": 2.4461538461538466e-06,
      "loss": 0.0019,
      "step": 3423
    },
    {
      "epoch": 43.757188498402556,
      "grad_norm": 0.2318740338087082,
      "learning_rate": 2.441025641025641e-06,
      "loss": 0.0019,
      "step": 3424
    },
    {
      "epoch": 43.76996805111821,
      "grad_norm": 0.30630600452423096,
      "learning_rate": 2.435897435897436e-06,
      "loss": 0.0026,
      "step": 3425
    },
    {
      "epoch": 43.78274760383387,
      "grad_norm": 0.23179428279399872,
      "learning_rate": 2.430769230769231e-06,
      "loss": 0.002,
      "step": 3426
    },
    {
      "epoch": 43.79552715654952,
      "grad_norm": 1.9203321933746338,
      "learning_rate": 2.425641025641026e-06,
      "loss": 0.0049,
      "step": 3427
    },
    {
      "epoch": 43.808306709265175,
      "grad_norm": 0.611625611782074,
      "learning_rate": 2.4205128205128207e-06,
      "loss": 0.0064,
      "step": 3428
    },
    {
      "epoch": 43.82108626198083,
      "grad_norm": 0.4640527069568634,
      "learning_rate": 2.4153846153846156e-06,
      "loss": 0.0035,
      "step": 3429
    },
    {
      "epoch": 43.83386581469649,
      "grad_norm": 0.23337207734584808,
      "learning_rate": 2.4102564102564105e-06,
      "loss": 0.0023,
      "step": 3430
    },
    {
      "epoch": 43.846645367412144,
      "grad_norm": 0.2928684651851654,
      "learning_rate": 2.4051282051282054e-06,
      "loss": 0.0027,
      "step": 3431
    },
    {
      "epoch": 43.85942492012779,
      "grad_norm": 2.824985980987549,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.0083,
      "step": 3432
    },
    {
      "epoch": 43.87220447284345,
      "grad_norm": 0.2678843140602112,
      "learning_rate": 2.3948717948717952e-06,
      "loss": 0.0028,
      "step": 3433
    },
    {
      "epoch": 43.884984025559106,
      "grad_norm": 5.614882469177246,
      "learning_rate": 2.38974358974359e-06,
      "loss": 0.0041,
      "step": 3434
    },
    {
      "epoch": 43.89776357827476,
      "grad_norm": 0.19786269962787628,
      "learning_rate": 2.384615384615385e-06,
      "loss": 0.002,
      "step": 3435
    },
    {
      "epoch": 43.91054313099041,
      "grad_norm": 0.18499012291431427,
      "learning_rate": 2.3794871794871795e-06,
      "loss": 0.0019,
      "step": 3436
    },
    {
      "epoch": 43.92332268370607,
      "grad_norm": 0.18536154925823212,
      "learning_rate": 2.3743589743589744e-06,
      "loss": 0.0025,
      "step": 3437
    },
    {
      "epoch": 43.936102236421725,
      "grad_norm": 0.2182464748620987,
      "learning_rate": 2.3692307692307697e-06,
      "loss": 0.0023,
      "step": 3438
    },
    {
      "epoch": 43.94888178913738,
      "grad_norm": 0.19127224385738373,
      "learning_rate": 2.364102564102564e-06,
      "loss": 0.0018,
      "step": 3439
    },
    {
      "epoch": 43.96166134185304,
      "grad_norm": 0.5205743312835693,
      "learning_rate": 2.358974358974359e-06,
      "loss": 0.0024,
      "step": 3440
    },
    {
      "epoch": 43.97444089456869,
      "grad_norm": 3.6712076663970947,
      "learning_rate": 2.353846153846154e-06,
      "loss": 0.0217,
      "step": 3441
    },
    {
      "epoch": 43.98722044728434,
      "grad_norm": 0.6396458745002747,
      "learning_rate": 2.348717948717949e-06,
      "loss": 0.0019,
      "step": 3442
    },
    {
      "epoch": 44.0,
      "grad_norm": 2.4400033950805664,
      "learning_rate": 2.3435897435897438e-06,
      "loss": 0.0133,
      "step": 3443
    },
    {
      "epoch": 44.01277955271566,
      "grad_norm": 0.16040480136871338,
      "learning_rate": 2.3384615384615387e-06,
      "loss": 0.0015,
      "step": 3444
    },
    {
      "epoch": 44.02555910543131,
      "grad_norm": 0.2779988646507263,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.0019,
      "step": 3445
    },
    {
      "epoch": 44.03833865814696,
      "grad_norm": 1.6264729499816895,
      "learning_rate": 2.3282051282051285e-06,
      "loss": 0.006,
      "step": 3446
    },
    {
      "epoch": 44.05111821086262,
      "grad_norm": 0.21936100721359253,
      "learning_rate": 2.3230769230769234e-06,
      "loss": 0.0019,
      "step": 3447
    },
    {
      "epoch": 44.063897763578275,
      "grad_norm": 1.0929688215255737,
      "learning_rate": 2.3179487179487183e-06,
      "loss": 0.0057,
      "step": 3448
    },
    {
      "epoch": 44.07667731629393,
      "grad_norm": 0.2932341694831848,
      "learning_rate": 2.312820512820513e-06,
      "loss": 0.0027,
      "step": 3449
    },
    {
      "epoch": 44.08945686900959,
      "grad_norm": 0.8114387392997742,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.0033,
      "step": 3450
    },
    {
      "epoch": 44.10223642172524,
      "grad_norm": 0.17941796779632568,
      "learning_rate": 2.3025641025641026e-06,
      "loss": 0.0021,
      "step": 3451
    },
    {
      "epoch": 44.115015974440894,
      "grad_norm": 0.16614599525928497,
      "learning_rate": 2.2974358974358975e-06,
      "loss": 0.0016,
      "step": 3452
    },
    {
      "epoch": 44.12779552715655,
      "grad_norm": 0.47889798879623413,
      "learning_rate": 2.2923076923076928e-06,
      "loss": 0.0022,
      "step": 3453
    },
    {
      "epoch": 44.14057507987221,
      "grad_norm": 0.2067495584487915,
      "learning_rate": 2.2871794871794872e-06,
      "loss": 0.0025,
      "step": 3454
    },
    {
      "epoch": 44.153354632587856,
      "grad_norm": 1.3304522037506104,
      "learning_rate": 2.282051282051282e-06,
      "loss": 0.0038,
      "step": 3455
    },
    {
      "epoch": 44.16613418530351,
      "grad_norm": 0.1890897899866104,
      "learning_rate": 2.276923076923077e-06,
      "loss": 0.0019,
      "step": 3456
    },
    {
      "epoch": 44.17891373801917,
      "grad_norm": 1.378034234046936,
      "learning_rate": 2.271794871794872e-06,
      "loss": 0.0063,
      "step": 3457
    },
    {
      "epoch": 44.191693290734825,
      "grad_norm": 0.1573825478553772,
      "learning_rate": 2.266666666666667e-06,
      "loss": 0.0018,
      "step": 3458
    },
    {
      "epoch": 44.20447284345048,
      "grad_norm": 0.2968226373195648,
      "learning_rate": 2.2615384615384617e-06,
      "loss": 0.0026,
      "step": 3459
    },
    {
      "epoch": 44.21725239616613,
      "grad_norm": 0.4137711226940155,
      "learning_rate": 2.2564102564102566e-06,
      "loss": 0.0018,
      "step": 3460
    },
    {
      "epoch": 44.23003194888179,
      "grad_norm": 0.19083532691001892,
      "learning_rate": 2.2512820512820515e-06,
      "loss": 0.0016,
      "step": 3461
    },
    {
      "epoch": 44.242811501597444,
      "grad_norm": 0.14727646112442017,
      "learning_rate": 2.2461538461538464e-06,
      "loss": 0.0015,
      "step": 3462
    },
    {
      "epoch": 44.2555910543131,
      "grad_norm": 0.2691571116447449,
      "learning_rate": 2.2410256410256413e-06,
      "loss": 0.0019,
      "step": 3463
    },
    {
      "epoch": 44.26837060702876,
      "grad_norm": 0.24418750405311584,
      "learning_rate": 2.2358974358974362e-06,
      "loss": 0.0018,
      "step": 3464
    },
    {
      "epoch": 44.281150159744406,
      "grad_norm": 0.8175804615020752,
      "learning_rate": 2.230769230769231e-06,
      "loss": 0.0037,
      "step": 3465
    },
    {
      "epoch": 44.29392971246006,
      "grad_norm": 0.14546926319599152,
      "learning_rate": 2.2256410256410256e-06,
      "loss": 0.0017,
      "step": 3466
    },
    {
      "epoch": 44.30670926517572,
      "grad_norm": 0.267969012260437,
      "learning_rate": 2.2205128205128205e-06,
      "loss": 0.0027,
      "step": 3467
    },
    {
      "epoch": 44.319488817891376,
      "grad_norm": 0.17838163673877716,
      "learning_rate": 2.215384615384616e-06,
      "loss": 0.0018,
      "step": 3468
    },
    {
      "epoch": 44.33226837060703,
      "grad_norm": 0.18786239624023438,
      "learning_rate": 2.2102564102564103e-06,
      "loss": 0.0016,
      "step": 3469
    },
    {
      "epoch": 44.34504792332268,
      "grad_norm": 0.22198481857776642,
      "learning_rate": 2.2051282051282052e-06,
      "loss": 0.0027,
      "step": 3470
    },
    {
      "epoch": 44.35782747603834,
      "grad_norm": 0.2121235728263855,
      "learning_rate": 2.2e-06,
      "loss": 0.0021,
      "step": 3471
    },
    {
      "epoch": 44.370607028753994,
      "grad_norm": 0.7737230062484741,
      "learning_rate": 2.194871794871795e-06,
      "loss": 0.0033,
      "step": 3472
    },
    {
      "epoch": 44.38338658146965,
      "grad_norm": 0.29892033338546753,
      "learning_rate": 2.18974358974359e-06,
      "loss": 0.0017,
      "step": 3473
    },
    {
      "epoch": 44.3961661341853,
      "grad_norm": 0.7860128879547119,
      "learning_rate": 2.184615384615385e-06,
      "loss": 0.0028,
      "step": 3474
    },
    {
      "epoch": 44.40894568690096,
      "grad_norm": 0.4038985073566437,
      "learning_rate": 2.1794871794871797e-06,
      "loss": 0.0022,
      "step": 3475
    },
    {
      "epoch": 44.42172523961661,
      "grad_norm": 0.23415660858154297,
      "learning_rate": 2.1743589743589746e-06,
      "loss": 0.0023,
      "step": 3476
    },
    {
      "epoch": 44.43450479233227,
      "grad_norm": 0.1980719119310379,
      "learning_rate": 2.1692307692307695e-06,
      "loss": 0.0022,
      "step": 3477
    },
    {
      "epoch": 44.447284345047926,
      "grad_norm": 3.2138073444366455,
      "learning_rate": 2.1641025641025644e-06,
      "loss": 0.0176,
      "step": 3478
    },
    {
      "epoch": 44.460063897763575,
      "grad_norm": 0.22042608261108398,
      "learning_rate": 2.1589743589743593e-06,
      "loss": 0.0023,
      "step": 3479
    },
    {
      "epoch": 44.47284345047923,
      "grad_norm": 0.21012216806411743,
      "learning_rate": 2.153846153846154e-06,
      "loss": 0.0021,
      "step": 3480
    },
    {
      "epoch": 44.48562300319489,
      "grad_norm": 0.2902188301086426,
      "learning_rate": 2.1487179487179487e-06,
      "loss": 0.0027,
      "step": 3481
    },
    {
      "epoch": 44.498402555910545,
      "grad_norm": 3.315211296081543,
      "learning_rate": 2.1435897435897436e-06,
      "loss": 0.0171,
      "step": 3482
    },
    {
      "epoch": 44.5111821086262,
      "grad_norm": 0.19962161779403687,
      "learning_rate": 2.138461538461539e-06,
      "loss": 0.0019,
      "step": 3483
    },
    {
      "epoch": 44.52396166134185,
      "grad_norm": 0.23143111169338226,
      "learning_rate": 2.133333333333334e-06,
      "loss": 0.0022,
      "step": 3484
    },
    {
      "epoch": 44.53674121405751,
      "grad_norm": 0.20259682834148407,
      "learning_rate": 2.1282051282051283e-06,
      "loss": 0.0021,
      "step": 3485
    },
    {
      "epoch": 44.54952076677316,
      "grad_norm": 1.61074697971344,
      "learning_rate": 2.123076923076923e-06,
      "loss": 0.0072,
      "step": 3486
    },
    {
      "epoch": 44.56230031948882,
      "grad_norm": 0.18312349915504456,
      "learning_rate": 2.117948717948718e-06,
      "loss": 0.0019,
      "step": 3487
    },
    {
      "epoch": 44.575079872204476,
      "grad_norm": 0.7772158980369568,
      "learning_rate": 2.112820512820513e-06,
      "loss": 0.0033,
      "step": 3488
    },
    {
      "epoch": 44.587859424920126,
      "grad_norm": 0.530954897403717,
      "learning_rate": 2.107692307692308e-06,
      "loss": 0.0073,
      "step": 3489
    },
    {
      "epoch": 44.60063897763578,
      "grad_norm": 0.31667378544807434,
      "learning_rate": 2.1025641025641028e-06,
      "loss": 0.0022,
      "step": 3490
    },
    {
      "epoch": 44.61341853035144,
      "grad_norm": 1.6839007139205933,
      "learning_rate": 2.0974358974358977e-06,
      "loss": 0.0101,
      "step": 3491
    },
    {
      "epoch": 44.626198083067095,
      "grad_norm": 2.2546749114990234,
      "learning_rate": 2.0923076923076926e-06,
      "loss": 0.0112,
      "step": 3492
    },
    {
      "epoch": 44.638977635782744,
      "grad_norm": 0.6103613972663879,
      "learning_rate": 2.0871794871794875e-06,
      "loss": 0.0037,
      "step": 3493
    },
    {
      "epoch": 44.6517571884984,
      "grad_norm": 0.21684005856513977,
      "learning_rate": 2.0820512820512824e-06,
      "loss": 0.002,
      "step": 3494
    },
    {
      "epoch": 44.66453674121406,
      "grad_norm": 0.17034326493740082,
      "learning_rate": 2.0769230769230773e-06,
      "loss": 0.0021,
      "step": 3495
    },
    {
      "epoch": 44.677316293929714,
      "grad_norm": 0.22330626845359802,
      "learning_rate": 2.0717948717948717e-06,
      "loss": 0.0021,
      "step": 3496
    },
    {
      "epoch": 44.69009584664537,
      "grad_norm": 3.0296339988708496,
      "learning_rate": 2.0666666666666666e-06,
      "loss": 0.0089,
      "step": 3497
    },
    {
      "epoch": 44.70287539936102,
      "grad_norm": 0.19906385242938995,
      "learning_rate": 2.061538461538462e-06,
      "loss": 0.0023,
      "step": 3498
    },
    {
      "epoch": 44.715654952076676,
      "grad_norm": 0.1592789739370346,
      "learning_rate": 2.056410256410257e-06,
      "loss": 0.0016,
      "step": 3499
    },
    {
      "epoch": 44.72843450479233,
      "grad_norm": 0.1241416409611702,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 0.0012,
      "step": 3500
    },
    {
      "epoch": 44.72843450479233,
      "eval_loss": 1.0085582733154297,
      "eval_runtime": 183.8592,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 3500
    },
    {
      "epoch": 44.74121405750799,
      "grad_norm": 0.216583251953125,
      "learning_rate": 2.0461538461538462e-06,
      "loss": 0.0028,
      "step": 3501
    },
    {
      "epoch": 44.753993610223645,
      "grad_norm": 1.870769739151001,
      "learning_rate": 2.041025641025641e-06,
      "loss": 0.0104,
      "step": 3502
    },
    {
      "epoch": 44.766773162939295,
      "grad_norm": 1.918493390083313,
      "learning_rate": 2.035897435897436e-06,
      "loss": 0.01,
      "step": 3503
    },
    {
      "epoch": 44.77955271565495,
      "grad_norm": 4.812563896179199,
      "learning_rate": 2.030769230769231e-06,
      "loss": 0.0039,
      "step": 3504
    },
    {
      "epoch": 44.79233226837061,
      "grad_norm": 4.635837554931641,
      "learning_rate": 2.025641025641026e-06,
      "loss": 0.0121,
      "step": 3505
    },
    {
      "epoch": 44.805111821086264,
      "grad_norm": 0.2971982955932617,
      "learning_rate": 2.0205128205128207e-06,
      "loss": 0.0026,
      "step": 3506
    },
    {
      "epoch": 44.81789137380191,
      "grad_norm": 0.3084527254104614,
      "learning_rate": 2.0153846153846156e-06,
      "loss": 0.0034,
      "step": 3507
    },
    {
      "epoch": 44.83067092651757,
      "grad_norm": 0.1466815024614334,
      "learning_rate": 2.0102564102564105e-06,
      "loss": 0.0017,
      "step": 3508
    },
    {
      "epoch": 44.843450479233226,
      "grad_norm": 2.4110586643218994,
      "learning_rate": 2.0051282051282054e-06,
      "loss": 0.0115,
      "step": 3509
    },
    {
      "epoch": 44.85623003194888,
      "grad_norm": 5.122982978820801,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0182,
      "step": 3510
    },
    {
      "epoch": 44.86900958466454,
      "grad_norm": 0.19120436906814575,
      "learning_rate": 1.994871794871795e-06,
      "loss": 0.0021,
      "step": 3511
    },
    {
      "epoch": 44.88178913738019,
      "grad_norm": 1.6846766471862793,
      "learning_rate": 1.9897435897435897e-06,
      "loss": 0.0079,
      "step": 3512
    },
    {
      "epoch": 44.894568690095845,
      "grad_norm": 0.2311975210905075,
      "learning_rate": 1.984615384615385e-06,
      "loss": 0.0023,
      "step": 3513
    },
    {
      "epoch": 44.9073482428115,
      "grad_norm": 0.26933738589286804,
      "learning_rate": 1.97948717948718e-06,
      "loss": 0.0018,
      "step": 3514
    },
    {
      "epoch": 44.92012779552716,
      "grad_norm": 0.19096258282661438,
      "learning_rate": 1.9743589743589744e-06,
      "loss": 0.0023,
      "step": 3515
    },
    {
      "epoch": 44.932907348242814,
      "grad_norm": 3.0288987159729004,
      "learning_rate": 1.9692307692307693e-06,
      "loss": 0.0144,
      "step": 3516
    },
    {
      "epoch": 44.945686900958464,
      "grad_norm": 0.8595489263534546,
      "learning_rate": 1.964102564102564e-06,
      "loss": 0.0043,
      "step": 3517
    },
    {
      "epoch": 44.95846645367412,
      "grad_norm": 0.15664945542812347,
      "learning_rate": 1.958974358974359e-06,
      "loss": 0.0015,
      "step": 3518
    },
    {
      "epoch": 44.97124600638978,
      "grad_norm": 0.17332378029823303,
      "learning_rate": 1.953846153846154e-06,
      "loss": 0.0016,
      "step": 3519
    },
    {
      "epoch": 44.98402555910543,
      "grad_norm": 0.198051318526268,
      "learning_rate": 1.948717948717949e-06,
      "loss": 0.002,
      "step": 3520
    },
    {
      "epoch": 44.99680511182109,
      "grad_norm": 0.1939736008644104,
      "learning_rate": 1.943589743589744e-06,
      "loss": 0.002,
      "step": 3521
    },
    {
      "epoch": 45.00958466453674,
      "grad_norm": 0.20715303719043732,
      "learning_rate": 1.9384615384615387e-06,
      "loss": 0.0021,
      "step": 3522
    },
    {
      "epoch": 45.022364217252395,
      "grad_norm": 0.1790590137243271,
      "learning_rate": 1.9333333333333336e-06,
      "loss": 0.0019,
      "step": 3523
    },
    {
      "epoch": 45.03514376996805,
      "grad_norm": 0.18158192932605743,
      "learning_rate": 1.9282051282051285e-06,
      "loss": 0.0017,
      "step": 3524
    },
    {
      "epoch": 45.04792332268371,
      "grad_norm": 1.3653088808059692,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 0.0052,
      "step": 3525
    },
    {
      "epoch": 45.06070287539936,
      "grad_norm": 1.2879325151443481,
      "learning_rate": 1.917948717948718e-06,
      "loss": 0.0044,
      "step": 3526
    },
    {
      "epoch": 45.073482428115014,
      "grad_norm": 1.6927472352981567,
      "learning_rate": 1.9128205128205128e-06,
      "loss": 0.0065,
      "step": 3527
    },
    {
      "epoch": 45.08626198083067,
      "grad_norm": 2.4802870750427246,
      "learning_rate": 1.907692307692308e-06,
      "loss": 0.0145,
      "step": 3528
    },
    {
      "epoch": 45.09904153354633,
      "grad_norm": 0.4064551293849945,
      "learning_rate": 1.9025641025641028e-06,
      "loss": 0.0028,
      "step": 3529
    },
    {
      "epoch": 45.11182108626198,
      "grad_norm": 1.8474481105804443,
      "learning_rate": 1.8974358974358975e-06,
      "loss": 0.0055,
      "step": 3530
    },
    {
      "epoch": 45.12460063897763,
      "grad_norm": 0.17202718555927277,
      "learning_rate": 1.8923076923076924e-06,
      "loss": 0.0013,
      "step": 3531
    },
    {
      "epoch": 45.13738019169329,
      "grad_norm": 0.3263075351715088,
      "learning_rate": 1.8871794871794875e-06,
      "loss": 0.003,
      "step": 3532
    },
    {
      "epoch": 45.150159744408946,
      "grad_norm": 0.195256307721138,
      "learning_rate": 1.8820512820512822e-06,
      "loss": 0.0019,
      "step": 3533
    },
    {
      "epoch": 45.1629392971246,
      "grad_norm": 3.3216702938079834,
      "learning_rate": 1.876923076923077e-06,
      "loss": 0.011,
      "step": 3534
    },
    {
      "epoch": 45.17571884984026,
      "grad_norm": 0.5705974102020264,
      "learning_rate": 1.871794871794872e-06,
      "loss": 0.003,
      "step": 3535
    },
    {
      "epoch": 45.18849840255591,
      "grad_norm": 0.15692636370658875,
      "learning_rate": 1.8666666666666669e-06,
      "loss": 0.0017,
      "step": 3536
    },
    {
      "epoch": 45.201277955271564,
      "grad_norm": 0.28287068009376526,
      "learning_rate": 1.8615384615384616e-06,
      "loss": 0.0024,
      "step": 3537
    },
    {
      "epoch": 45.21405750798722,
      "grad_norm": 0.14750763773918152,
      "learning_rate": 1.8564102564102565e-06,
      "loss": 0.0014,
      "step": 3538
    },
    {
      "epoch": 45.22683706070288,
      "grad_norm": 0.15956076979637146,
      "learning_rate": 1.8512820512820516e-06,
      "loss": 0.0014,
      "step": 3539
    },
    {
      "epoch": 45.239616613418534,
      "grad_norm": 0.17729100584983826,
      "learning_rate": 1.8461538461538465e-06,
      "loss": 0.0018,
      "step": 3540
    },
    {
      "epoch": 45.25239616613418,
      "grad_norm": 0.13930313289165497,
      "learning_rate": 1.8410256410256411e-06,
      "loss": 0.0013,
      "step": 3541
    },
    {
      "epoch": 45.26517571884984,
      "grad_norm": 0.2076169103384018,
      "learning_rate": 1.835897435897436e-06,
      "loss": 0.0016,
      "step": 3542
    },
    {
      "epoch": 45.277955271565496,
      "grad_norm": 0.3562934100627899,
      "learning_rate": 1.830769230769231e-06,
      "loss": 0.0026,
      "step": 3543
    },
    {
      "epoch": 45.29073482428115,
      "grad_norm": 0.706603467464447,
      "learning_rate": 1.8256410256410258e-06,
      "loss": 0.0026,
      "step": 3544
    },
    {
      "epoch": 45.3035143769968,
      "grad_norm": 0.1821099817752838,
      "learning_rate": 1.8205128205128205e-06,
      "loss": 0.0021,
      "step": 3545
    },
    {
      "epoch": 45.31629392971246,
      "grad_norm": 0.12550444900989532,
      "learning_rate": 1.8153846153846154e-06,
      "loss": 0.0016,
      "step": 3546
    },
    {
      "epoch": 45.329073482428115,
      "grad_norm": 0.25906607508659363,
      "learning_rate": 1.8102564102564105e-06,
      "loss": 0.0022,
      "step": 3547
    },
    {
      "epoch": 45.34185303514377,
      "grad_norm": 0.20704346895217896,
      "learning_rate": 1.8051282051282054e-06,
      "loss": 0.0019,
      "step": 3548
    },
    {
      "epoch": 45.35463258785943,
      "grad_norm": 0.17126823961734772,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.0022,
      "step": 3549
    },
    {
      "epoch": 45.36741214057508,
      "grad_norm": 3.5319011211395264,
      "learning_rate": 1.794871794871795e-06,
      "loss": 0.0223,
      "step": 3550
    },
    {
      "epoch": 45.38019169329073,
      "grad_norm": 3.9966418743133545,
      "learning_rate": 1.78974358974359e-06,
      "loss": 0.0119,
      "step": 3551
    },
    {
      "epoch": 45.39297124600639,
      "grad_norm": 0.8652167320251465,
      "learning_rate": 1.7846153846153846e-06,
      "loss": 0.0034,
      "step": 3552
    },
    {
      "epoch": 45.405750798722046,
      "grad_norm": 0.15536391735076904,
      "learning_rate": 1.7794871794871795e-06,
      "loss": 0.0017,
      "step": 3553
    },
    {
      "epoch": 45.4185303514377,
      "grad_norm": 0.26495033502578735,
      "learning_rate": 1.7743589743589746e-06,
      "loss": 0.002,
      "step": 3554
    },
    {
      "epoch": 45.43130990415335,
      "grad_norm": 0.209041565656662,
      "learning_rate": 1.7692307692307695e-06,
      "loss": 0.0018,
      "step": 3555
    },
    {
      "epoch": 45.44408945686901,
      "grad_norm": 0.2236473560333252,
      "learning_rate": 1.7641025641025642e-06,
      "loss": 0.0023,
      "step": 3556
    },
    {
      "epoch": 45.456869009584665,
      "grad_norm": 1.227617859840393,
      "learning_rate": 1.7589743589743591e-06,
      "loss": 0.0062,
      "step": 3557
    },
    {
      "epoch": 45.46964856230032,
      "grad_norm": 0.4426514208316803,
      "learning_rate": 1.753846153846154e-06,
      "loss": 0.0029,
      "step": 3558
    },
    {
      "epoch": 45.48242811501598,
      "grad_norm": 0.14357681572437286,
      "learning_rate": 1.748717948717949e-06,
      "loss": 0.0015,
      "step": 3559
    },
    {
      "epoch": 45.49520766773163,
      "grad_norm": 0.1984434574842453,
      "learning_rate": 1.7435897435897436e-06,
      "loss": 0.0019,
      "step": 3560
    },
    {
      "epoch": 45.50798722044728,
      "grad_norm": 0.4589730203151703,
      "learning_rate": 1.7384615384615385e-06,
      "loss": 0.0022,
      "step": 3561
    },
    {
      "epoch": 45.52076677316294,
      "grad_norm": 0.14460931718349457,
      "learning_rate": 1.7333333333333336e-06,
      "loss": 0.0015,
      "step": 3562
    },
    {
      "epoch": 45.533546325878596,
      "grad_norm": 0.18210144340991974,
      "learning_rate": 1.7282051282051285e-06,
      "loss": 0.0021,
      "step": 3563
    },
    {
      "epoch": 45.546325878594246,
      "grad_norm": 0.23533381521701813,
      "learning_rate": 1.7230769230769232e-06,
      "loss": 0.0016,
      "step": 3564
    },
    {
      "epoch": 45.5591054313099,
      "grad_norm": 0.17796681821346283,
      "learning_rate": 1.717948717948718e-06,
      "loss": 0.0018,
      "step": 3565
    },
    {
      "epoch": 45.57188498402556,
      "grad_norm": 0.29740259051322937,
      "learning_rate": 1.712820512820513e-06,
      "loss": 0.0019,
      "step": 3566
    },
    {
      "epoch": 45.584664536741215,
      "grad_norm": 0.11241334676742554,
      "learning_rate": 1.7076923076923077e-06,
      "loss": 0.0012,
      "step": 3567
    },
    {
      "epoch": 45.59744408945687,
      "grad_norm": 0.17197149991989136,
      "learning_rate": 1.7025641025641026e-06,
      "loss": 0.0018,
      "step": 3568
    },
    {
      "epoch": 45.61022364217252,
      "grad_norm": 0.207344189286232,
      "learning_rate": 1.6974358974358977e-06,
      "loss": 0.0021,
      "step": 3569
    },
    {
      "epoch": 45.62300319488818,
      "grad_norm": 0.1945933699607849,
      "learning_rate": 1.6923076923076926e-06,
      "loss": 0.0025,
      "step": 3570
    },
    {
      "epoch": 45.635782747603834,
      "grad_norm": 2.508556842803955,
      "learning_rate": 1.6871794871794873e-06,
      "loss": 0.0074,
      "step": 3571
    },
    {
      "epoch": 45.64856230031949,
      "grad_norm": 0.8266756534576416,
      "learning_rate": 1.6820512820512822e-06,
      "loss": 0.0024,
      "step": 3572
    },
    {
      "epoch": 45.66134185303515,
      "grad_norm": 1.3747810125350952,
      "learning_rate": 1.676923076923077e-06,
      "loss": 0.0058,
      "step": 3573
    },
    {
      "epoch": 45.674121405750796,
      "grad_norm": 1.8754050731658936,
      "learning_rate": 1.671794871794872e-06,
      "loss": 0.0096,
      "step": 3574
    },
    {
      "epoch": 45.68690095846645,
      "grad_norm": 0.19026875495910645,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.0026,
      "step": 3575
    },
    {
      "epoch": 45.69968051118211,
      "grad_norm": 0.8473201394081116,
      "learning_rate": 1.6615384615384616e-06,
      "loss": 0.0042,
      "step": 3576
    },
    {
      "epoch": 45.712460063897765,
      "grad_norm": 0.47852179408073425,
      "learning_rate": 1.6564102564102567e-06,
      "loss": 0.0021,
      "step": 3577
    },
    {
      "epoch": 45.72523961661342,
      "grad_norm": 23.40967559814453,
      "learning_rate": 1.6512820512820516e-06,
      "loss": 0.0304,
      "step": 3578
    },
    {
      "epoch": 45.73801916932907,
      "grad_norm": 1.9984266757965088,
      "learning_rate": 1.6461538461538463e-06,
      "loss": 0.0097,
      "step": 3579
    },
    {
      "epoch": 45.75079872204473,
      "grad_norm": 0.2590010166168213,
      "learning_rate": 1.6410256410256412e-06,
      "loss": 0.0016,
      "step": 3580
    },
    {
      "epoch": 45.763578274760384,
      "grad_norm": 0.16944079101085663,
      "learning_rate": 1.635897435897436e-06,
      "loss": 0.0022,
      "step": 3581
    },
    {
      "epoch": 45.77635782747604,
      "grad_norm": 2.3917136192321777,
      "learning_rate": 1.6307692307692307e-06,
      "loss": 0.0101,
      "step": 3582
    },
    {
      "epoch": 45.78913738019169,
      "grad_norm": 0.2407619208097458,
      "learning_rate": 1.6256410256410256e-06,
      "loss": 0.0021,
      "step": 3583
    },
    {
      "epoch": 45.801916932907346,
      "grad_norm": 0.23970063030719757,
      "learning_rate": 1.6205128205128208e-06,
      "loss": 0.0026,
      "step": 3584
    },
    {
      "epoch": 45.814696485623,
      "grad_norm": 0.1650267094373703,
      "learning_rate": 1.6153846153846157e-06,
      "loss": 0.0015,
      "step": 3585
    },
    {
      "epoch": 45.82747603833866,
      "grad_norm": 0.2412625551223755,
      "learning_rate": 1.6102564102564103e-06,
      "loss": 0.0026,
      "step": 3586
    },
    {
      "epoch": 45.840255591054316,
      "grad_norm": 0.14923495054244995,
      "learning_rate": 1.6051282051282052e-06,
      "loss": 0.0017,
      "step": 3587
    },
    {
      "epoch": 45.853035143769965,
      "grad_norm": 5.012885570526123,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.0028,
      "step": 3588
    },
    {
      "epoch": 45.86581469648562,
      "grad_norm": 0.2059669941663742,
      "learning_rate": 1.594871794871795e-06,
      "loss": 0.0018,
      "step": 3589
    },
    {
      "epoch": 45.87859424920128,
      "grad_norm": 0.44068247079849243,
      "learning_rate": 1.5897435897435897e-06,
      "loss": 0.0017,
      "step": 3590
    },
    {
      "epoch": 45.891373801916934,
      "grad_norm": 1.4092161655426025,
      "learning_rate": 1.5846153846153846e-06,
      "loss": 0.0074,
      "step": 3591
    },
    {
      "epoch": 45.90415335463259,
      "grad_norm": 0.23094308376312256,
      "learning_rate": 1.5794871794871797e-06,
      "loss": 0.0022,
      "step": 3592
    },
    {
      "epoch": 45.91693290734824,
      "grad_norm": 0.2290555089712143,
      "learning_rate": 1.5743589743589746e-06,
      "loss": 0.0023,
      "step": 3593
    },
    {
      "epoch": 45.9297124600639,
      "grad_norm": 0.1883171647787094,
      "learning_rate": 1.5692307692307693e-06,
      "loss": 0.0017,
      "step": 3594
    },
    {
      "epoch": 45.94249201277955,
      "grad_norm": 0.36363181471824646,
      "learning_rate": 1.5641025641025642e-06,
      "loss": 0.0019,
      "step": 3595
    },
    {
      "epoch": 45.95527156549521,
      "grad_norm": 0.18326343595981598,
      "learning_rate": 1.5589743589743591e-06,
      "loss": 0.0017,
      "step": 3596
    },
    {
      "epoch": 45.968051118210866,
      "grad_norm": 2.4599180221557617,
      "learning_rate": 1.5538461538461538e-06,
      "loss": 0.0098,
      "step": 3597
    },
    {
      "epoch": 45.980830670926515,
      "grad_norm": 0.5097063779830933,
      "learning_rate": 1.5487179487179487e-06,
      "loss": 0.0064,
      "step": 3598
    },
    {
      "epoch": 45.99361022364217,
      "grad_norm": 1.839652180671692,
      "learning_rate": 1.5435897435897438e-06,
      "loss": 0.01,
      "step": 3599
    },
    {
      "epoch": 46.00638977635783,
      "grad_norm": 1.2611196041107178,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.008,
      "step": 3600
    },
    {
      "epoch": 46.00638977635783,
      "eval_loss": 1.0099000930786133,
      "eval_runtime": 183.9179,
      "eval_samples_per_second": 0.854,
      "eval_steps_per_second": 0.109,
      "step": 3600
    },
    {
      "epoch": 46.019169329073485,
      "grad_norm": 0.23989491164684296,
      "learning_rate": 1.5333333333333334e-06,
      "loss": 0.0017,
      "step": 3601
    },
    {
      "epoch": 46.031948881789134,
      "grad_norm": 1.742085576057434,
      "learning_rate": 1.5282051282051283e-06,
      "loss": 0.0039,
      "step": 3602
    },
    {
      "epoch": 46.04472843450479,
      "grad_norm": 0.4358189105987549,
      "learning_rate": 1.5230769230769232e-06,
      "loss": 0.0019,
      "step": 3603
    },
    {
      "epoch": 46.05750798722045,
      "grad_norm": 2.8469250202178955,
      "learning_rate": 1.517948717948718e-06,
      "loss": 0.0082,
      "step": 3604
    },
    {
      "epoch": 46.0702875399361,
      "grad_norm": 0.18406391143798828,
      "learning_rate": 1.5128205128205128e-06,
      "loss": 0.002,
      "step": 3605
    },
    {
      "epoch": 46.08306709265176,
      "grad_norm": 1.457541584968567,
      "learning_rate": 1.5076923076923077e-06,
      "loss": 0.0062,
      "step": 3606
    },
    {
      "epoch": 46.09584664536741,
      "grad_norm": 0.17311814427375793,
      "learning_rate": 1.5025641025641028e-06,
      "loss": 0.002,
      "step": 3607
    },
    {
      "epoch": 46.108626198083066,
      "grad_norm": 1.3626841306686401,
      "learning_rate": 1.4974358974358977e-06,
      "loss": 0.0047,
      "step": 3608
    },
    {
      "epoch": 46.12140575079872,
      "grad_norm": 0.17242155969142914,
      "learning_rate": 1.4923076923076924e-06,
      "loss": 0.0018,
      "step": 3609
    },
    {
      "epoch": 46.13418530351438,
      "grad_norm": 0.1784520000219345,
      "learning_rate": 1.4871794871794873e-06,
      "loss": 0.0014,
      "step": 3610
    },
    {
      "epoch": 46.146964856230035,
      "grad_norm": 2.5860233306884766,
      "learning_rate": 1.4820512820512822e-06,
      "loss": 0.0118,
      "step": 3611
    },
    {
      "epoch": 46.159744408945684,
      "grad_norm": 0.16604258120059967,
      "learning_rate": 1.476923076923077e-06,
      "loss": 0.002,
      "step": 3612
    },
    {
      "epoch": 46.17252396166134,
      "grad_norm": 0.20272715389728546,
      "learning_rate": 1.4717948717948718e-06,
      "loss": 0.0018,
      "step": 3613
    },
    {
      "epoch": 46.185303514377,
      "grad_norm": 0.21260668337345123,
      "learning_rate": 1.4666666666666669e-06,
      "loss": 0.0022,
      "step": 3614
    },
    {
      "epoch": 46.198083067092654,
      "grad_norm": 0.20811006426811218,
      "learning_rate": 1.4615384615384618e-06,
      "loss": 0.0022,
      "step": 3615
    },
    {
      "epoch": 46.21086261980831,
      "grad_norm": 0.15792424976825714,
      "learning_rate": 1.4564102564102565e-06,
      "loss": 0.0019,
      "step": 3616
    },
    {
      "epoch": 46.22364217252396,
      "grad_norm": 0.2316557615995407,
      "learning_rate": 1.4512820512820514e-06,
      "loss": 0.0029,
      "step": 3617
    },
    {
      "epoch": 46.236421725239616,
      "grad_norm": 0.17002829909324646,
      "learning_rate": 1.4461538461538463e-06,
      "loss": 0.0016,
      "step": 3618
    },
    {
      "epoch": 46.24920127795527,
      "grad_norm": 0.16938528418540955,
      "learning_rate": 1.4410256410256412e-06,
      "loss": 0.0017,
      "step": 3619
    },
    {
      "epoch": 46.26198083067093,
      "grad_norm": 0.1708260029554367,
      "learning_rate": 1.4358974358974359e-06,
      "loss": 0.0016,
      "step": 3620
    },
    {
      "epoch": 46.27476038338658,
      "grad_norm": 0.1572258621454239,
      "learning_rate": 1.4307692307692308e-06,
      "loss": 0.0014,
      "step": 3621
    },
    {
      "epoch": 46.287539936102235,
      "grad_norm": 0.2195149064064026,
      "learning_rate": 1.4256410256410259e-06,
      "loss": 0.0016,
      "step": 3622
    },
    {
      "epoch": 46.30031948881789,
      "grad_norm": 0.19424758851528168,
      "learning_rate": 1.4205128205128208e-06,
      "loss": 0.0015,
      "step": 3623
    },
    {
      "epoch": 46.31309904153355,
      "grad_norm": 3.137338161468506,
      "learning_rate": 1.4153846153846155e-06,
      "loss": 0.0129,
      "step": 3624
    },
    {
      "epoch": 46.325878594249204,
      "grad_norm": 0.14023944735527039,
      "learning_rate": 1.4102564102564104e-06,
      "loss": 0.0013,
      "step": 3625
    },
    {
      "epoch": 46.33865814696485,
      "grad_norm": 0.17934924364089966,
      "learning_rate": 1.4051282051282052e-06,
      "loss": 0.0015,
      "step": 3626
    },
    {
      "epoch": 46.35143769968051,
      "grad_norm": 2.365750312805176,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.006,
      "step": 3627
    },
    {
      "epoch": 46.364217252396166,
      "grad_norm": 0.4310351014137268,
      "learning_rate": 1.3948717948717948e-06,
      "loss": 0.0039,
      "step": 3628
    },
    {
      "epoch": 46.37699680511182,
      "grad_norm": 0.18593768775463104,
      "learning_rate": 1.38974358974359e-06,
      "loss": 0.0022,
      "step": 3629
    },
    {
      "epoch": 46.38977635782748,
      "grad_norm": 0.15974895656108856,
      "learning_rate": 1.3846153846153848e-06,
      "loss": 0.0017,
      "step": 3630
    },
    {
      "epoch": 46.40255591054313,
      "grad_norm": 0.4098556935787201,
      "learning_rate": 1.3794871794871795e-06,
      "loss": 0.0055,
      "step": 3631
    },
    {
      "epoch": 46.415335463258785,
      "grad_norm": 0.2783844470977783,
      "learning_rate": 1.3743589743589744e-06,
      "loss": 0.0025,
      "step": 3632
    },
    {
      "epoch": 46.42811501597444,
      "grad_norm": 1.6561388969421387,
      "learning_rate": 1.3692307692307693e-06,
      "loss": 0.006,
      "step": 3633
    },
    {
      "epoch": 46.4408945686901,
      "grad_norm": 0.14973992109298706,
      "learning_rate": 1.3641025641025642e-06,
      "loss": 0.0015,
      "step": 3634
    },
    {
      "epoch": 46.453674121405754,
      "grad_norm": 0.16307753324508667,
      "learning_rate": 1.358974358974359e-06,
      "loss": 0.0018,
      "step": 3635
    },
    {
      "epoch": 46.466453674121404,
      "grad_norm": 3.942716598510742,
      "learning_rate": 1.3538461538461538e-06,
      "loss": 0.0208,
      "step": 3636
    },
    {
      "epoch": 46.47923322683706,
      "grad_norm": 0.33902183175086975,
      "learning_rate": 1.348717948717949e-06,
      "loss": 0.0023,
      "step": 3637
    },
    {
      "epoch": 46.49201277955272,
      "grad_norm": 0.12845934927463531,
      "learning_rate": 1.3435897435897438e-06,
      "loss": 0.0014,
      "step": 3638
    },
    {
      "epoch": 46.50479233226837,
      "grad_norm": 1.4974967241287231,
      "learning_rate": 1.3384615384615385e-06,
      "loss": 0.005,
      "step": 3639
    },
    {
      "epoch": 46.51757188498402,
      "grad_norm": 0.1921854466199875,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0025,
      "step": 3640
    },
    {
      "epoch": 46.53035143769968,
      "grad_norm": 0.2795115113258362,
      "learning_rate": 1.3282051282051283e-06,
      "loss": 0.002,
      "step": 3641
    },
    {
      "epoch": 46.543130990415335,
      "grad_norm": 0.17906340956687927,
      "learning_rate": 1.3230769230769232e-06,
      "loss": 0.0023,
      "step": 3642
    },
    {
      "epoch": 46.55591054313099,
      "grad_norm": 0.16304001212120056,
      "learning_rate": 1.317948717948718e-06,
      "loss": 0.0019,
      "step": 3643
    },
    {
      "epoch": 46.56869009584665,
      "grad_norm": 2.5445923805236816,
      "learning_rate": 1.312820512820513e-06,
      "loss": 0.0069,
      "step": 3644
    },
    {
      "epoch": 46.5814696485623,
      "grad_norm": 0.23029464483261108,
      "learning_rate": 1.307692307692308e-06,
      "loss": 0.0024,
      "step": 3645
    },
    {
      "epoch": 46.594249201277954,
      "grad_norm": 0.10605199635028839,
      "learning_rate": 1.3025641025641026e-06,
      "loss": 0.0011,
      "step": 3646
    },
    {
      "epoch": 46.60702875399361,
      "grad_norm": 2.268832206726074,
      "learning_rate": 1.2974358974358975e-06,
      "loss": 0.0107,
      "step": 3647
    },
    {
      "epoch": 46.61980830670927,
      "grad_norm": 1.6812770366668701,
      "learning_rate": 1.2923076923076924e-06,
      "loss": 0.0085,
      "step": 3648
    },
    {
      "epoch": 46.63258785942492,
      "grad_norm": 0.11973489075899124,
      "learning_rate": 1.2871794871794873e-06,
      "loss": 0.0013,
      "step": 3649
    },
    {
      "epoch": 46.64536741214057,
      "grad_norm": 0.11298532783985138,
      "learning_rate": 1.282051282051282e-06,
      "loss": 0.0012,
      "step": 3650
    },
    {
      "epoch": 46.65814696485623,
      "grad_norm": 0.19396765530109406,
      "learning_rate": 1.2769230769230769e-06,
      "loss": 0.0021,
      "step": 3651
    },
    {
      "epoch": 46.670926517571885,
      "grad_norm": 0.265076220035553,
      "learning_rate": 1.271794871794872e-06,
      "loss": 0.0027,
      "step": 3652
    },
    {
      "epoch": 46.68370607028754,
      "grad_norm": 0.14036832749843597,
      "learning_rate": 1.2666666666666669e-06,
      "loss": 0.0011,
      "step": 3653
    },
    {
      "epoch": 46.6964856230032,
      "grad_norm": 0.18519698083400726,
      "learning_rate": 1.2615384615384616e-06,
      "loss": 0.0024,
      "step": 3654
    },
    {
      "epoch": 46.70926517571885,
      "grad_norm": 0.8766704201698303,
      "learning_rate": 1.2564102564102565e-06,
      "loss": 0.0097,
      "step": 3655
    },
    {
      "epoch": 46.722044728434504,
      "grad_norm": 0.16572509706020355,
      "learning_rate": 1.2512820512820514e-06,
      "loss": 0.0016,
      "step": 3656
    },
    {
      "epoch": 46.73482428115016,
      "grad_norm": 0.2027369737625122,
      "learning_rate": 1.2461538461538463e-06,
      "loss": 0.0023,
      "step": 3657
    },
    {
      "epoch": 46.74760383386582,
      "grad_norm": 1.8083834648132324,
      "learning_rate": 1.2410256410256412e-06,
      "loss": 0.0049,
      "step": 3658
    },
    {
      "epoch": 46.760383386581466,
      "grad_norm": 0.19146762788295746,
      "learning_rate": 1.235897435897436e-06,
      "loss": 0.002,
      "step": 3659
    },
    {
      "epoch": 46.77316293929712,
      "grad_norm": 1.8865890502929688,
      "learning_rate": 1.230769230769231e-06,
      "loss": 0.0048,
      "step": 3660
    },
    {
      "epoch": 46.78594249201278,
      "grad_norm": 0.19453971087932587,
      "learning_rate": 1.2256410256410257e-06,
      "loss": 0.002,
      "step": 3661
    },
    {
      "epoch": 46.798722044728436,
      "grad_norm": 2.282036304473877,
      "learning_rate": 1.2205128205128206e-06,
      "loss": 0.0095,
      "step": 3662
    },
    {
      "epoch": 46.81150159744409,
      "grad_norm": 1.6410950422286987,
      "learning_rate": 1.2153846153846155e-06,
      "loss": 0.0076,
      "step": 3663
    },
    {
      "epoch": 46.82428115015974,
      "grad_norm": 0.8596600890159607,
      "learning_rate": 1.2102564102564104e-06,
      "loss": 0.0039,
      "step": 3664
    },
    {
      "epoch": 46.8370607028754,
      "grad_norm": 0.16360537707805634,
      "learning_rate": 1.2051282051282053e-06,
      "loss": 0.0014,
      "step": 3665
    },
    {
      "epoch": 46.849840255591054,
      "grad_norm": 0.18147307634353638,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.0018,
      "step": 3666
    },
    {
      "epoch": 46.86261980830671,
      "grad_norm": 0.16816723346710205,
      "learning_rate": 1.194871794871795e-06,
      "loss": 0.0015,
      "step": 3667
    },
    {
      "epoch": 46.87539936102237,
      "grad_norm": 1.8942514657974243,
      "learning_rate": 1.1897435897435897e-06,
      "loss": 0.0084,
      "step": 3668
    },
    {
      "epoch": 46.88817891373802,
      "grad_norm": 0.14104534685611725,
      "learning_rate": 1.1846153846153849e-06,
      "loss": 0.0014,
      "step": 3669
    },
    {
      "epoch": 46.90095846645367,
      "grad_norm": 0.18655520677566528,
      "learning_rate": 1.1794871794871795e-06,
      "loss": 0.0017,
      "step": 3670
    },
    {
      "epoch": 46.91373801916933,
      "grad_norm": 0.3814006447792053,
      "learning_rate": 1.1743589743589744e-06,
      "loss": 0.0027,
      "step": 3671
    },
    {
      "epoch": 46.926517571884986,
      "grad_norm": 0.2369621992111206,
      "learning_rate": 1.1692307692307693e-06,
      "loss": 0.002,
      "step": 3672
    },
    {
      "epoch": 46.93929712460064,
      "grad_norm": 2.7895922660827637,
      "learning_rate": 1.1641025641025642e-06,
      "loss": 0.008,
      "step": 3673
    },
    {
      "epoch": 46.95207667731629,
      "grad_norm": 0.2802821695804596,
      "learning_rate": 1.1589743589743591e-06,
      "loss": 0.0028,
      "step": 3674
    },
    {
      "epoch": 46.96485623003195,
      "grad_norm": 0.1200646385550499,
      "learning_rate": 1.153846153846154e-06,
      "loss": 0.0013,
      "step": 3675
    },
    {
      "epoch": 46.977635782747605,
      "grad_norm": 0.19148743152618408,
      "learning_rate": 1.1487179487179487e-06,
      "loss": 0.0022,
      "step": 3676
    },
    {
      "epoch": 46.99041533546326,
      "grad_norm": 1.2269463539123535,
      "learning_rate": 1.1435897435897436e-06,
      "loss": 0.0064,
      "step": 3677
    },
    {
      "epoch": 47.00319488817891,
      "grad_norm": 0.25025907158851624,
      "learning_rate": 1.1384615384615385e-06,
      "loss": 0.0024,
      "step": 3678
    },
    {
      "epoch": 47.01597444089457,
      "grad_norm": 0.17812982201576233,
      "learning_rate": 1.1333333333333334e-06,
      "loss": 0.0016,
      "step": 3679
    },
    {
      "epoch": 47.02875399361022,
      "grad_norm": 0.23735368251800537,
      "learning_rate": 1.1282051282051283e-06,
      "loss": 0.002,
      "step": 3680
    },
    {
      "epoch": 47.04153354632588,
      "grad_norm": 2.0460994243621826,
      "learning_rate": 1.1230769230769232e-06,
      "loss": 0.0078,
      "step": 3681
    },
    {
      "epoch": 47.054313099041536,
      "grad_norm": 0.13436773419380188,
      "learning_rate": 1.1179487179487181e-06,
      "loss": 0.0014,
      "step": 3682
    },
    {
      "epoch": 47.067092651757186,
      "grad_norm": 0.24940666556358337,
      "learning_rate": 1.1128205128205128e-06,
      "loss": 0.0023,
      "step": 3683
    },
    {
      "epoch": 47.07987220447284,
      "grad_norm": 0.16844338178634644,
      "learning_rate": 1.107692307692308e-06,
      "loss": 0.0016,
      "step": 3684
    },
    {
      "epoch": 47.0926517571885,
      "grad_norm": 0.18543598055839539,
      "learning_rate": 1.1025641025641026e-06,
      "loss": 0.0019,
      "step": 3685
    },
    {
      "epoch": 47.105431309904155,
      "grad_norm": 1.0382249355316162,
      "learning_rate": 1.0974358974358975e-06,
      "loss": 0.0056,
      "step": 3686
    },
    {
      "epoch": 47.11821086261981,
      "grad_norm": 0.4113228917121887,
      "learning_rate": 1.0923076923076924e-06,
      "loss": 0.0054,
      "step": 3687
    },
    {
      "epoch": 47.13099041533546,
      "grad_norm": 1.7715373039245605,
      "learning_rate": 1.0871794871794873e-06,
      "loss": 0.0084,
      "step": 3688
    },
    {
      "epoch": 47.14376996805112,
      "grad_norm": 0.19282501935958862,
      "learning_rate": 1.0820512820512822e-06,
      "loss": 0.0016,
      "step": 3689
    },
    {
      "epoch": 47.156549520766774,
      "grad_norm": 0.2740253210067749,
      "learning_rate": 1.076923076923077e-06,
      "loss": 0.0026,
      "step": 3690
    },
    {
      "epoch": 47.16932907348243,
      "grad_norm": 1.1587003469467163,
      "learning_rate": 1.0717948717948718e-06,
      "loss": 0.0037,
      "step": 3691
    },
    {
      "epoch": 47.18210862619808,
      "grad_norm": 0.19705770909786224,
      "learning_rate": 1.066666666666667e-06,
      "loss": 0.0021,
      "step": 3692
    },
    {
      "epoch": 47.194888178913736,
      "grad_norm": 0.14728839695453644,
      "learning_rate": 1.0615384615384616e-06,
      "loss": 0.002,
      "step": 3693
    },
    {
      "epoch": 47.20766773162939,
      "grad_norm": 1.7193306684494019,
      "learning_rate": 1.0564102564102565e-06,
      "loss": 0.0053,
      "step": 3694
    },
    {
      "epoch": 47.22044728434505,
      "grad_norm": 0.907518744468689,
      "learning_rate": 1.0512820512820514e-06,
      "loss": 0.0031,
      "step": 3695
    },
    {
      "epoch": 47.233226837060705,
      "grad_norm": 0.2023201584815979,
      "learning_rate": 1.0461538461538463e-06,
      "loss": 0.0019,
      "step": 3696
    },
    {
      "epoch": 47.246006389776355,
      "grad_norm": 0.13571235537528992,
      "learning_rate": 1.0410256410256412e-06,
      "loss": 0.0016,
      "step": 3697
    },
    {
      "epoch": 47.25878594249201,
      "grad_norm": 2.384904146194458,
      "learning_rate": 1.0358974358974359e-06,
      "loss": 0.0108,
      "step": 3698
    },
    {
      "epoch": 47.27156549520767,
      "grad_norm": 0.15866927802562714,
      "learning_rate": 1.030769230769231e-06,
      "loss": 0.0017,
      "step": 3699
    },
    {
      "epoch": 47.284345047923324,
      "grad_norm": 1.8669986724853516,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 0.0062,
      "step": 3700
    },
    {
      "epoch": 47.284345047923324,
      "eval_loss": 1.00686514377594,
      "eval_runtime": 184.0665,
      "eval_samples_per_second": 0.853,
      "eval_steps_per_second": 0.109,
      "step": 3700
    },
    {
      "epoch": 47.29712460063898,
      "grad_norm": 0.30873823165893555,
      "learning_rate": 1.0205128205128206e-06,
      "loss": 0.0023,
      "step": 3701
    },
    {
      "epoch": 47.30990415335463,
      "grad_norm": 0.1505894660949707,
      "learning_rate": 1.0153846153846155e-06,
      "loss": 0.0018,
      "step": 3702
    },
    {
      "epoch": 47.322683706070286,
      "grad_norm": 2.2037155628204346,
      "learning_rate": 1.0102564102564104e-06,
      "loss": 0.0098,
      "step": 3703
    },
    {
      "epoch": 47.33546325878594,
      "grad_norm": 1.8789855241775513,
      "learning_rate": 1.0051282051282053e-06,
      "loss": 0.0066,
      "step": 3704
    },
    {
      "epoch": 47.3482428115016,
      "grad_norm": 0.1800302118062973,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0022,
      "step": 3705
    },
    {
      "epoch": 47.361022364217256,
      "grad_norm": 0.16504280269145966,
      "learning_rate": 9.948717948717949e-07,
      "loss": 0.0019,
      "step": 3706
    },
    {
      "epoch": 47.373801916932905,
      "grad_norm": 0.41811108589172363,
      "learning_rate": 9.8974358974359e-07,
      "loss": 0.0035,
      "step": 3707
    },
    {
      "epoch": 47.38658146964856,
      "grad_norm": 0.09904661029577255,
      "learning_rate": 9.846153846153847e-07,
      "loss": 0.0011,
      "step": 3708
    },
    {
      "epoch": 47.39936102236422,
      "grad_norm": 0.21644745767116547,
      "learning_rate": 9.794871794871796e-07,
      "loss": 0.0023,
      "step": 3709
    },
    {
      "epoch": 47.412140575079874,
      "grad_norm": 0.1227216124534607,
      "learning_rate": 9.743589743589745e-07,
      "loss": 0.0012,
      "step": 3710
    },
    {
      "epoch": 47.424920127795524,
      "grad_norm": 0.15440890192985535,
      "learning_rate": 9.692307692307693e-07,
      "loss": 0.0017,
      "step": 3711
    },
    {
      "epoch": 47.43769968051118,
      "grad_norm": 0.19403807818889618,
      "learning_rate": 9.641025641025642e-07,
      "loss": 0.0021,
      "step": 3712
    },
    {
      "epoch": 47.45047923322684,
      "grad_norm": 1.5223453044891357,
      "learning_rate": 9.58974358974359e-07,
      "loss": 0.0081,
      "step": 3713
    },
    {
      "epoch": 47.46325878594249,
      "grad_norm": 0.12501774728298187,
      "learning_rate": 9.53846153846154e-07,
      "loss": 0.0012,
      "step": 3714
    },
    {
      "epoch": 47.47603833865815,
      "grad_norm": 0.9358645677566528,
      "learning_rate": 9.487179487179487e-07,
      "loss": 0.0045,
      "step": 3715
    },
    {
      "epoch": 47.4888178913738,
      "grad_norm": 0.18395741283893585,
      "learning_rate": 9.435897435897437e-07,
      "loss": 0.0019,
      "step": 3716
    },
    {
      "epoch": 47.501597444089455,
      "grad_norm": 0.6766879558563232,
      "learning_rate": 9.384615384615385e-07,
      "loss": 0.0036,
      "step": 3717
    },
    {
      "epoch": 47.51437699680511,
      "grad_norm": 0.3552172780036926,
      "learning_rate": 9.333333333333334e-07,
      "loss": 0.0019,
      "step": 3718
    },
    {
      "epoch": 47.52715654952077,
      "grad_norm": 1.697737693786621,
      "learning_rate": 9.282051282051282e-07,
      "loss": 0.0047,
      "step": 3719
    },
    {
      "epoch": 47.539936102236425,
      "grad_norm": 3.0718917846679688,
      "learning_rate": 9.230769230769232e-07,
      "loss": 0.0088,
      "step": 3720
    },
    {
      "epoch": 47.552715654952074,
      "grad_norm": 0.14367461204528809,
      "learning_rate": 9.17948717948718e-07,
      "loss": 0.0016,
      "step": 3721
    },
    {
      "epoch": 47.56549520766773,
      "grad_norm": 0.5913472175598145,
      "learning_rate": 9.128205128205129e-07,
      "loss": 0.003,
      "step": 3722
    },
    {
      "epoch": 47.57827476038339,
      "grad_norm": 0.17336972057819366,
      "learning_rate": 9.076923076923077e-07,
      "loss": 0.002,
      "step": 3723
    },
    {
      "epoch": 47.59105431309904,
      "grad_norm": 0.21685920655727386,
      "learning_rate": 9.025641025641027e-07,
      "loss": 0.0021,
      "step": 3724
    },
    {
      "epoch": 47.6038338658147,
      "grad_norm": 1.3096964359283447,
      "learning_rate": 8.974358974358975e-07,
      "loss": 0.0048,
      "step": 3725
    },
    {
      "epoch": 47.61661341853035,
      "grad_norm": 0.20485009253025055,
      "learning_rate": 8.923076923076923e-07,
      "loss": 0.0018,
      "step": 3726
    },
    {
      "epoch": 47.629392971246006,
      "grad_norm": 0.14636144042015076,
      "learning_rate": 8.871794871794873e-07,
      "loss": 0.0017,
      "step": 3727
    },
    {
      "epoch": 47.64217252396166,
      "grad_norm": 0.14755165576934814,
      "learning_rate": 8.820512820512821e-07,
      "loss": 0.0015,
      "step": 3728
    },
    {
      "epoch": 47.65495207667732,
      "grad_norm": 0.23167769610881805,
      "learning_rate": 8.76923076923077e-07,
      "loss": 0.002,
      "step": 3729
    },
    {
      "epoch": 47.66773162939297,
      "grad_norm": 0.1636737734079361,
      "learning_rate": 8.717948717948718e-07,
      "loss": 0.0019,
      "step": 3730
    },
    {
      "epoch": 47.680511182108624,
      "grad_norm": 0.1251421570777893,
      "learning_rate": 8.666666666666668e-07,
      "loss": 0.0012,
      "step": 3731
    },
    {
      "epoch": 47.69329073482428,
      "grad_norm": 2.23331618309021,
      "learning_rate": 8.615384615384616e-07,
      "loss": 0.007,
      "step": 3732
    },
    {
      "epoch": 47.70607028753994,
      "grad_norm": 0.22965694963932037,
      "learning_rate": 8.564102564102565e-07,
      "loss": 0.0019,
      "step": 3733
    },
    {
      "epoch": 47.718849840255594,
      "grad_norm": 1.785829782485962,
      "learning_rate": 8.512820512820513e-07,
      "loss": 0.0062,
      "step": 3734
    },
    {
      "epoch": 47.73162939297124,
      "grad_norm": 0.25463688373565674,
      "learning_rate": 8.461538461538463e-07,
      "loss": 0.0031,
      "step": 3735
    },
    {
      "epoch": 47.7444089456869,
      "grad_norm": 0.28074225783348083,
      "learning_rate": 8.410256410256411e-07,
      "loss": 0.0016,
      "step": 3736
    },
    {
      "epoch": 47.757188498402556,
      "grad_norm": 0.26788008213043213,
      "learning_rate": 8.35897435897436e-07,
      "loss": 0.0021,
      "step": 3737
    },
    {
      "epoch": 47.76996805111821,
      "grad_norm": 0.20389509201049805,
      "learning_rate": 8.307692307692308e-07,
      "loss": 0.002,
      "step": 3738
    },
    {
      "epoch": 47.78274760383387,
      "grad_norm": 0.21783320605754852,
      "learning_rate": 8.256410256410258e-07,
      "loss": 0.0021,
      "step": 3739
    },
    {
      "epoch": 47.79552715654952,
      "grad_norm": 0.23883669078350067,
      "learning_rate": 8.205128205128206e-07,
      "loss": 0.0017,
      "step": 3740
    },
    {
      "epoch": 47.808306709265175,
      "grad_norm": 1.630712628364563,
      "learning_rate": 8.153846153846154e-07,
      "loss": 0.0082,
      "step": 3741
    },
    {
      "epoch": 47.82108626198083,
      "grad_norm": 2.074826955795288,
      "learning_rate": 8.102564102564104e-07,
      "loss": 0.0098,
      "step": 3742
    },
    {
      "epoch": 47.83386581469649,
      "grad_norm": 0.18601283431053162,
      "learning_rate": 8.051282051282052e-07,
      "loss": 0.0016,
      "step": 3743
    },
    {
      "epoch": 47.846645367412144,
      "grad_norm": 0.4480161666870117,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.0028,
      "step": 3744
    },
    {
      "epoch": 47.85942492012779,
      "grad_norm": 1.5206952095031738,
      "learning_rate": 7.948717948717949e-07,
      "loss": 0.0081,
      "step": 3745
    },
    {
      "epoch": 47.87220447284345,
      "grad_norm": 0.16170930862426758,
      "learning_rate": 7.897435897435899e-07,
      "loss": 0.0014,
      "step": 3746
    },
    {
      "epoch": 47.884984025559106,
      "grad_norm": 1.4309064149856567,
      "learning_rate": 7.846153846153847e-07,
      "loss": 0.0044,
      "step": 3747
    },
    {
      "epoch": 47.89776357827476,
      "grad_norm": 0.2414461225271225,
      "learning_rate": 7.794871794871796e-07,
      "loss": 0.0014,
      "step": 3748
    },
    {
      "epoch": 47.91054313099041,
      "grad_norm": 2.7287521362304688,
      "learning_rate": 7.743589743589744e-07,
      "loss": 0.0102,
      "step": 3749
    },
    {
      "epoch": 47.92332268370607,
      "grad_norm": 1.3179041147232056,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.0056,
      "step": 3750
    },
    {
      "epoch": 47.936102236421725,
      "grad_norm": 1.9829533100128174,
      "learning_rate": 7.641025641025642e-07,
      "loss": 0.0047,
      "step": 3751
    },
    {
      "epoch": 47.94888178913738,
      "grad_norm": 3.517019271850586,
      "learning_rate": 7.58974358974359e-07,
      "loss": 0.0112,
      "step": 3752
    },
    {
      "epoch": 47.96166134185304,
      "grad_norm": 0.16058802604675293,
      "learning_rate": 7.538461538461538e-07,
      "loss": 0.0013,
      "step": 3753
    },
    {
      "epoch": 47.97444089456869,
      "grad_norm": 0.1638914942741394,
      "learning_rate": 7.487179487179488e-07,
      "loss": 0.0017,
      "step": 3754
    },
    {
      "epoch": 47.98722044728434,
      "grad_norm": 0.17627263069152832,
      "learning_rate": 7.435897435897436e-07,
      "loss": 0.0017,
      "step": 3755
    },
    {
      "epoch": 48.0,
      "grad_norm": 2.2881815433502197,
      "learning_rate": 7.384615384615385e-07,
      "loss": 0.01,
      "step": 3756
    },
    {
      "epoch": 48.01277955271566,
      "grad_norm": 0.9438730478286743,
      "learning_rate": 7.333333333333334e-07,
      "loss": 0.0043,
      "step": 3757
    },
    {
      "epoch": 48.02555910543131,
      "grad_norm": 0.1496344655752182,
      "learning_rate": 7.282051282051282e-07,
      "loss": 0.0016,
      "step": 3758
    },
    {
      "epoch": 48.03833865814696,
      "grad_norm": 0.17239366471767426,
      "learning_rate": 7.230769230769231e-07,
      "loss": 0.0014,
      "step": 3759
    },
    {
      "epoch": 48.05111821086262,
      "grad_norm": 0.15948452055454254,
      "learning_rate": 7.179487179487179e-07,
      "loss": 0.0017,
      "step": 3760
    },
    {
      "epoch": 48.063897763578275,
      "grad_norm": 1.7323073148727417,
      "learning_rate": 7.128205128205129e-07,
      "loss": 0.0074,
      "step": 3761
    },
    {
      "epoch": 48.07667731629393,
      "grad_norm": 0.22783014178276062,
      "learning_rate": 7.076923076923077e-07,
      "loss": 0.0019,
      "step": 3762
    },
    {
      "epoch": 48.08945686900959,
      "grad_norm": 0.209009051322937,
      "learning_rate": 7.025641025641026e-07,
      "loss": 0.0021,
      "step": 3763
    },
    {
      "epoch": 48.10223642172524,
      "grad_norm": 0.13151539862155914,
      "learning_rate": 6.974358974358974e-07,
      "loss": 0.0016,
      "step": 3764
    },
    {
      "epoch": 48.115015974440894,
      "grad_norm": 0.1741136908531189,
      "learning_rate": 6.923076923076924e-07,
      "loss": 0.0019,
      "step": 3765
    },
    {
      "epoch": 48.12779552715655,
      "grad_norm": 2.1559276580810547,
      "learning_rate": 6.871794871794872e-07,
      "loss": 0.0051,
      "step": 3766
    },
    {
      "epoch": 48.14057507987221,
      "grad_norm": 0.145247682929039,
      "learning_rate": 6.820512820512821e-07,
      "loss": 0.0018,
      "step": 3767
    },
    {
      "epoch": 48.153354632587856,
      "grad_norm": 2.345256805419922,
      "learning_rate": 6.769230769230769e-07,
      "loss": 0.008,
      "step": 3768
    },
    {
      "epoch": 48.16613418530351,
      "grad_norm": 0.1540064513683319,
      "learning_rate": 6.717948717948719e-07,
      "loss": 0.0019,
      "step": 3769
    },
    {
      "epoch": 48.17891373801917,
      "grad_norm": 1.0705827474594116,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0024,
      "step": 3770
    },
    {
      "epoch": 48.191693290734825,
      "grad_norm": 0.13630788028240204,
      "learning_rate": 6.615384615384616e-07,
      "loss": 0.0012,
      "step": 3771
    },
    {
      "epoch": 48.20447284345048,
      "grad_norm": 0.15815885365009308,
      "learning_rate": 6.564102564102565e-07,
      "loss": 0.002,
      "step": 3772
    },
    {
      "epoch": 48.21725239616613,
      "grad_norm": 0.14926832914352417,
      "learning_rate": 6.512820512820513e-07,
      "loss": 0.0015,
      "step": 3773
    },
    {
      "epoch": 48.23003194888179,
      "grad_norm": 0.2645859122276306,
      "learning_rate": 6.461538461538462e-07,
      "loss": 0.0023,
      "step": 3774
    },
    {
      "epoch": 48.242811501597444,
      "grad_norm": 0.1782495379447937,
      "learning_rate": 6.41025641025641e-07,
      "loss": 0.0015,
      "step": 3775
    },
    {
      "epoch": 48.2555910543131,
      "grad_norm": 0.15166476368904114,
      "learning_rate": 6.35897435897436e-07,
      "loss": 0.0014,
      "step": 3776
    },
    {
      "epoch": 48.26837060702876,
      "grad_norm": 0.2873496115207672,
      "learning_rate": 6.307692307692308e-07,
      "loss": 0.0018,
      "step": 3777
    },
    {
      "epoch": 48.281150159744406,
      "grad_norm": 2.4240097999572754,
      "learning_rate": 6.256410256410257e-07,
      "loss": 0.0062,
      "step": 3778
    },
    {
      "epoch": 48.29392971246006,
      "grad_norm": 0.18980348110198975,
      "learning_rate": 6.205128205128206e-07,
      "loss": 0.0016,
      "step": 3779
    },
    {
      "epoch": 48.30670926517572,
      "grad_norm": 0.17332720756530762,
      "learning_rate": 6.153846153846155e-07,
      "loss": 0.0013,
      "step": 3780
    },
    {
      "epoch": 48.319488817891376,
      "grad_norm": 0.14826764166355133,
      "learning_rate": 6.102564102564103e-07,
      "loss": 0.0015,
      "step": 3781
    },
    {
      "epoch": 48.33226837060703,
      "grad_norm": 0.1659523993730545,
      "learning_rate": 6.051282051282052e-07,
      "loss": 0.0017,
      "step": 3782
    },
    {
      "epoch": 48.34504792332268,
      "grad_norm": 0.9668567180633545,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.0032,
      "step": 3783
    },
    {
      "epoch": 48.35782747603834,
      "grad_norm": 0.24613407254219055,
      "learning_rate": 5.948717948717949e-07,
      "loss": 0.0022,
      "step": 3784
    },
    {
      "epoch": 48.370607028753994,
      "grad_norm": 1.009360671043396,
      "learning_rate": 5.897435897435898e-07,
      "loss": 0.0041,
      "step": 3785
    },
    {
      "epoch": 48.38338658146965,
      "grad_norm": 0.20935726165771484,
      "learning_rate": 5.846153846153847e-07,
      "loss": 0.0021,
      "step": 3786
    },
    {
      "epoch": 48.3961661341853,
      "grad_norm": 2.7185006141662598,
      "learning_rate": 5.794871794871796e-07,
      "loss": 0.0061,
      "step": 3787
    },
    {
      "epoch": 48.40894568690096,
      "grad_norm": 0.20508651435375214,
      "learning_rate": 5.743589743589744e-07,
      "loss": 0.0023,
      "step": 3788
    },
    {
      "epoch": 48.42172523961661,
      "grad_norm": 1.8751914501190186,
      "learning_rate": 5.692307692307693e-07,
      "loss": 0.0052,
      "step": 3789
    },
    {
      "epoch": 48.43450479233227,
      "grad_norm": 0.1863090544939041,
      "learning_rate": 5.641025641025642e-07,
      "loss": 0.002,
      "step": 3790
    },
    {
      "epoch": 48.447284345047926,
      "grad_norm": 0.16236571967601776,
      "learning_rate": 5.589743589743591e-07,
      "loss": 0.0017,
      "step": 3791
    },
    {
      "epoch": 48.460063897763575,
      "grad_norm": 0.15038584172725677,
      "learning_rate": 5.53846153846154e-07,
      "loss": 0.0016,
      "step": 3792
    },
    {
      "epoch": 48.47284345047923,
      "grad_norm": 0.32027992606163025,
      "learning_rate": 5.487179487179488e-07,
      "loss": 0.0034,
      "step": 3793
    },
    {
      "epoch": 48.48562300319489,
      "grad_norm": 0.18535518646240234,
      "learning_rate": 5.435897435897437e-07,
      "loss": 0.0018,
      "step": 3794
    },
    {
      "epoch": 48.498402555910545,
      "grad_norm": 2.3728909492492676,
      "learning_rate": 5.384615384615386e-07,
      "loss": 0.0102,
      "step": 3795
    },
    {
      "epoch": 48.5111821086262,
      "grad_norm": 0.17245027422904968,
      "learning_rate": 5.333333333333335e-07,
      "loss": 0.0017,
      "step": 3796
    },
    {
      "epoch": 48.52396166134185,
      "grad_norm": 0.15697172284126282,
      "learning_rate": 5.282051282051282e-07,
      "loss": 0.0013,
      "step": 3797
    },
    {
      "epoch": 48.53674121405751,
      "grad_norm": 1.6945269107818604,
      "learning_rate": 5.230769230769231e-07,
      "loss": 0.007,
      "step": 3798
    },
    {
      "epoch": 48.54952076677316,
      "grad_norm": 1.6684554815292358,
      "learning_rate": 5.179487179487179e-07,
      "loss": 0.0046,
      "step": 3799
    },
    {
      "epoch": 48.56230031948882,
      "grad_norm": 0.24569877982139587,
      "learning_rate": 5.128205128205128e-07,
      "loss": 0.0019,
      "step": 3800
    },
    {
      "epoch": 48.56230031948882,
      "eval_loss": 1.0110944509506226,
      "eval_runtime": 183.2434,
      "eval_samples_per_second": 0.857,
      "eval_steps_per_second": 0.109,
      "step": 3800
    },
    {
      "epoch": 48.575079872204476,
      "grad_norm": 0.16229873895645142,
      "learning_rate": 5.076923076923077e-07,
      "loss": 0.002,
      "step": 3801
    },
    {
      "epoch": 48.587859424920126,
      "grad_norm": 0.19286809861660004,
      "learning_rate": 5.025641025641026e-07,
      "loss": 0.0018,
      "step": 3802
    },
    {
      "epoch": 48.60063897763578,
      "grad_norm": 0.14536385238170624,
      "learning_rate": 4.974358974358974e-07,
      "loss": 0.0016,
      "step": 3803
    },
    {
      "epoch": 48.61341853035144,
      "grad_norm": 0.17883838713169098,
      "learning_rate": 4.923076923076923e-07,
      "loss": 0.0015,
      "step": 3804
    },
    {
      "epoch": 48.626198083067095,
      "grad_norm": 0.1891726702451706,
      "learning_rate": 4.871794871794872e-07,
      "loss": 0.002,
      "step": 3805
    },
    {
      "epoch": 48.638977635782744,
      "grad_norm": 0.13854210078716278,
      "learning_rate": 4.820512820512821e-07,
      "loss": 0.0014,
      "step": 3806
    },
    {
      "epoch": 48.6517571884984,
      "grad_norm": 0.21373547613620758,
      "learning_rate": 4.76923076923077e-07,
      "loss": 0.0015,
      "step": 3807
    },
    {
      "epoch": 48.66453674121406,
      "grad_norm": 26.711471557617188,
      "learning_rate": 4.7179487179487187e-07,
      "loss": 0.0052,
      "step": 3808
    },
    {
      "epoch": 48.677316293929714,
      "grad_norm": 0.8255504965782166,
      "learning_rate": 4.666666666666667e-07,
      "loss": 0.0032,
      "step": 3809
    },
    {
      "epoch": 48.69009584664537,
      "grad_norm": 0.15420027077198029,
      "learning_rate": 4.615384615384616e-07,
      "loss": 0.0015,
      "step": 3810
    },
    {
      "epoch": 48.70287539936102,
      "grad_norm": 0.19408243894577026,
      "learning_rate": 4.5641025641025646e-07,
      "loss": 0.0019,
      "step": 3811
    },
    {
      "epoch": 48.715654952076676,
      "grad_norm": 0.4737688899040222,
      "learning_rate": 4.5128205128205136e-07,
      "loss": 0.006,
      "step": 3812
    },
    {
      "epoch": 48.72843450479233,
      "grad_norm": 2.072347640991211,
      "learning_rate": 4.4615384615384615e-07,
      "loss": 0.01,
      "step": 3813
    },
    {
      "epoch": 48.74121405750799,
      "grad_norm": 0.18484090268611908,
      "learning_rate": 4.4102564102564105e-07,
      "loss": 0.0016,
      "step": 3814
    },
    {
      "epoch": 48.753993610223645,
      "grad_norm": 0.1918613314628601,
      "learning_rate": 4.358974358974359e-07,
      "loss": 0.0021,
      "step": 3815
    },
    {
      "epoch": 48.766773162939295,
      "grad_norm": 0.17631064355373383,
      "learning_rate": 4.307692307692308e-07,
      "loss": 0.0012,
      "step": 3816
    },
    {
      "epoch": 48.77955271565495,
      "grad_norm": 0.45219704508781433,
      "learning_rate": 4.2564102564102564e-07,
      "loss": 0.0015,
      "step": 3817
    },
    {
      "epoch": 48.79233226837061,
      "grad_norm": 0.22593191266059875,
      "learning_rate": 4.2051282051282054e-07,
      "loss": 0.0029,
      "step": 3818
    },
    {
      "epoch": 48.805111821086264,
      "grad_norm": 0.9036332964897156,
      "learning_rate": 4.153846153846154e-07,
      "loss": 0.0053,
      "step": 3819
    },
    {
      "epoch": 48.81789137380191,
      "grad_norm": 2.239448070526123,
      "learning_rate": 4.102564102564103e-07,
      "loss": 0.0122,
      "step": 3820
    },
    {
      "epoch": 48.83067092651757,
      "grad_norm": 0.9289953708648682,
      "learning_rate": 4.051282051282052e-07,
      "loss": 0.0038,
      "step": 3821
    },
    {
      "epoch": 48.843450479233226,
      "grad_norm": 0.18843135237693787,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.0018,
      "step": 3822
    },
    {
      "epoch": 48.85623003194888,
      "grad_norm": 0.12309607118368149,
      "learning_rate": 3.9487179487179493e-07,
      "loss": 0.0014,
      "step": 3823
    },
    {
      "epoch": 48.86900958466454,
      "grad_norm": 0.1520330011844635,
      "learning_rate": 3.897435897435898e-07,
      "loss": 0.0017,
      "step": 3824
    },
    {
      "epoch": 48.88178913738019,
      "grad_norm": 2.11013126373291,
      "learning_rate": 3.846153846153847e-07,
      "loss": 0.0063,
      "step": 3825
    },
    {
      "epoch": 48.894568690095845,
      "grad_norm": 1.5195311307907104,
      "learning_rate": 3.794871794871795e-07,
      "loss": 0.0054,
      "step": 3826
    },
    {
      "epoch": 48.9073482428115,
      "grad_norm": 3.264005184173584,
      "learning_rate": 3.743589743589744e-07,
      "loss": 0.0156,
      "step": 3827
    },
    {
      "epoch": 48.92012779552716,
      "grad_norm": 2.1601943969726562,
      "learning_rate": 3.6923076923076927e-07,
      "loss": 0.0087,
      "step": 3828
    },
    {
      "epoch": 48.932907348242814,
      "grad_norm": 0.15875303745269775,
      "learning_rate": 3.641025641025641e-07,
      "loss": 0.0019,
      "step": 3829
    },
    {
      "epoch": 48.945686900958464,
      "grad_norm": 0.11997360736131668,
      "learning_rate": 3.5897435897435896e-07,
      "loss": 0.0012,
      "step": 3830
    },
    {
      "epoch": 48.95846645367412,
      "grad_norm": 0.13904717564582825,
      "learning_rate": 3.5384615384615386e-07,
      "loss": 0.0016,
      "step": 3831
    },
    {
      "epoch": 48.97124600638978,
      "grad_norm": 1.2342338562011719,
      "learning_rate": 3.487179487179487e-07,
      "loss": 0.0046,
      "step": 3832
    },
    {
      "epoch": 48.98402555910543,
      "grad_norm": 0.20744307339191437,
      "learning_rate": 3.435897435897436e-07,
      "loss": 0.0021,
      "step": 3833
    },
    {
      "epoch": 48.99680511182109,
      "grad_norm": 0.9647104740142822,
      "learning_rate": 3.3846153846153845e-07,
      "loss": 0.0045,
      "step": 3834
    },
    {
      "epoch": 49.00958466453674,
      "grad_norm": 0.1561460793018341,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.0015,
      "step": 3835
    },
    {
      "epoch": 49.022364217252395,
      "grad_norm": 0.16960713267326355,
      "learning_rate": 3.2820512820512825e-07,
      "loss": 0.0015,
      "step": 3836
    },
    {
      "epoch": 49.03514376996805,
      "grad_norm": 0.20731723308563232,
      "learning_rate": 3.230769230769231e-07,
      "loss": 0.0022,
      "step": 3837
    },
    {
      "epoch": 49.04792332268371,
      "grad_norm": 2.431955575942993,
      "learning_rate": 3.17948717948718e-07,
      "loss": 0.0061,
      "step": 3838
    },
    {
      "epoch": 49.06070287539936,
      "grad_norm": 1.6707125902175903,
      "learning_rate": 3.1282051282051284e-07,
      "loss": 0.007,
      "step": 3839
    },
    {
      "epoch": 49.073482428115014,
      "grad_norm": 0.16068719327449799,
      "learning_rate": 3.0769230769230774e-07,
      "loss": 0.0015,
      "step": 3840
    },
    {
      "epoch": 49.08626198083067,
      "grad_norm": 0.22747410833835602,
      "learning_rate": 3.025641025641026e-07,
      "loss": 0.0023,
      "step": 3841
    },
    {
      "epoch": 49.09904153354633,
      "grad_norm": 0.1663256734609604,
      "learning_rate": 2.9743589743589744e-07,
      "loss": 0.0017,
      "step": 3842
    },
    {
      "epoch": 49.11182108626198,
      "grad_norm": 0.513690173625946,
      "learning_rate": 2.9230769230769234e-07,
      "loss": 0.0063,
      "step": 3843
    },
    {
      "epoch": 49.12460063897763,
      "grad_norm": 0.7397521138191223,
      "learning_rate": 2.871794871794872e-07,
      "loss": 0.0032,
      "step": 3844
    },
    {
      "epoch": 49.13738019169329,
      "grad_norm": 0.14274731278419495,
      "learning_rate": 2.820512820512821e-07,
      "loss": 0.0014,
      "step": 3845
    },
    {
      "epoch": 49.150159744408946,
      "grad_norm": 0.16468511521816254,
      "learning_rate": 2.76923076923077e-07,
      "loss": 0.0017,
      "step": 3846
    },
    {
      "epoch": 49.1629392971246,
      "grad_norm": 0.18096625804901123,
      "learning_rate": 2.717948717948718e-07,
      "loss": 0.0021,
      "step": 3847
    },
    {
      "epoch": 49.17571884984026,
      "grad_norm": 0.15279999375343323,
      "learning_rate": 2.666666666666667e-07,
      "loss": 0.0015,
      "step": 3848
    },
    {
      "epoch": 49.18849840255591,
      "grad_norm": 1.5307972431182861,
      "learning_rate": 2.6153846153846157e-07,
      "loss": 0.0077,
      "step": 3849
    },
    {
      "epoch": 49.201277955271564,
      "grad_norm": 0.17980337142944336,
      "learning_rate": 2.564102564102564e-07,
      "loss": 0.0026,
      "step": 3850
    },
    {
      "epoch": 49.21405750798722,
      "grad_norm": 1.1224316358566284,
      "learning_rate": 2.512820512820513e-07,
      "loss": 0.0045,
      "step": 3851
    },
    {
      "epoch": 49.22683706070288,
      "grad_norm": 0.23723852634429932,
      "learning_rate": 2.4615384615384616e-07,
      "loss": 0.0027,
      "step": 3852
    },
    {
      "epoch": 49.239616613418534,
      "grad_norm": 0.976337730884552,
      "learning_rate": 2.4102564102564106e-07,
      "loss": 0.0027,
      "step": 3853
    },
    {
      "epoch": 49.25239616613418,
      "grad_norm": 0.16196438670158386,
      "learning_rate": 2.3589743589743593e-07,
      "loss": 0.0015,
      "step": 3854
    },
    {
      "epoch": 49.26517571884984,
      "grad_norm": 0.18370366096496582,
      "learning_rate": 2.307692307692308e-07,
      "loss": 0.002,
      "step": 3855
    },
    {
      "epoch": 49.277955271565496,
      "grad_norm": 0.1293938308954239,
      "learning_rate": 2.2564102564102568e-07,
      "loss": 0.0013,
      "step": 3856
    },
    {
      "epoch": 49.29073482428115,
      "grad_norm": 0.2622494101524353,
      "learning_rate": 2.2051282051282053e-07,
      "loss": 0.0022,
      "step": 3857
    },
    {
      "epoch": 49.3035143769968,
      "grad_norm": 0.14362357556819916,
      "learning_rate": 2.153846153846154e-07,
      "loss": 0.0013,
      "step": 3858
    },
    {
      "epoch": 49.31629392971246,
      "grad_norm": 0.14474250376224518,
      "learning_rate": 2.1025641025641027e-07,
      "loss": 0.0017,
      "step": 3859
    },
    {
      "epoch": 49.329073482428115,
      "grad_norm": 4.0463643074035645,
      "learning_rate": 2.0512820512820514e-07,
      "loss": 0.0145,
      "step": 3860
    },
    {
      "epoch": 49.34185303514377,
      "grad_norm": 0.1382463425397873,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.0013,
      "step": 3861
    },
    {
      "epoch": 49.35463258785943,
      "grad_norm": 0.1417696177959442,
      "learning_rate": 1.948717948717949e-07,
      "loss": 0.0014,
      "step": 3862
    },
    {
      "epoch": 49.36741214057508,
      "grad_norm": 0.3333016633987427,
      "learning_rate": 1.8974358974358976e-07,
      "loss": 0.002,
      "step": 3863
    },
    {
      "epoch": 49.38019169329073,
      "grad_norm": 0.17449496686458588,
      "learning_rate": 1.8461538461538464e-07,
      "loss": 0.0019,
      "step": 3864
    },
    {
      "epoch": 49.39297124600639,
      "grad_norm": 5.625931262969971,
      "learning_rate": 1.7948717948717948e-07,
      "loss": 0.0035,
      "step": 3865
    },
    {
      "epoch": 49.405750798722046,
      "grad_norm": 0.2483331263065338,
      "learning_rate": 1.7435897435897435e-07,
      "loss": 0.0018,
      "step": 3866
    },
    {
      "epoch": 49.4185303514377,
      "grad_norm": 0.13882601261138916,
      "learning_rate": 1.6923076923076923e-07,
      "loss": 0.0016,
      "step": 3867
    },
    {
      "epoch": 49.43130990415335,
      "grad_norm": 0.2559881806373596,
      "learning_rate": 1.6410256410256413e-07,
      "loss": 0.0022,
      "step": 3868
    },
    {
      "epoch": 49.44408945686901,
      "grad_norm": 1.5500948429107666,
      "learning_rate": 1.58974358974359e-07,
      "loss": 0.0065,
      "step": 3869
    },
    {
      "epoch": 49.456869009584665,
      "grad_norm": 0.15375666320323944,
      "learning_rate": 1.5384615384615387e-07,
      "loss": 0.0019,
      "step": 3870
    },
    {
      "epoch": 49.46964856230032,
      "grad_norm": 0.1722114086151123,
      "learning_rate": 1.4871794871794872e-07,
      "loss": 0.0018,
      "step": 3871
    },
    {
      "epoch": 49.48242811501598,
      "grad_norm": 0.13881996273994446,
      "learning_rate": 1.435897435897436e-07,
      "loss": 0.0012,
      "step": 3872
    },
    {
      "epoch": 49.49520766773163,
      "grad_norm": 0.24192899465560913,
      "learning_rate": 1.384615384615385e-07,
      "loss": 0.0021,
      "step": 3873
    },
    {
      "epoch": 49.50798722044728,
      "grad_norm": 1.7498934268951416,
      "learning_rate": 1.3333333333333336e-07,
      "loss": 0.0024,
      "step": 3874
    },
    {
      "epoch": 49.52076677316294,
      "grad_norm": 0.18380694091320038,
      "learning_rate": 1.282051282051282e-07,
      "loss": 0.0024,
      "step": 3875
    },
    {
      "epoch": 49.533546325878596,
      "grad_norm": 0.17073246836662292,
      "learning_rate": 1.2307692307692308e-07,
      "loss": 0.0019,
      "step": 3876
    },
    {
      "epoch": 49.546325878594246,
      "grad_norm": 0.1260090172290802,
      "learning_rate": 1.1794871794871797e-07,
      "loss": 0.0011,
      "step": 3877
    },
    {
      "epoch": 49.5591054313099,
      "grad_norm": 0.2924799621105194,
      "learning_rate": 1.1282051282051284e-07,
      "loss": 0.0023,
      "step": 3878
    },
    {
      "epoch": 49.57188498402556,
      "grad_norm": 1.8029056787490845,
      "learning_rate": 1.076923076923077e-07,
      "loss": 0.0056,
      "step": 3879
    },
    {
      "epoch": 49.584664536741215,
      "grad_norm": 0.16758079826831818,
      "learning_rate": 1.0256410256410257e-07,
      "loss": 0.0016,
      "step": 3880
    },
    {
      "epoch": 49.59744408945687,
      "grad_norm": 0.1508578509092331,
      "learning_rate": 9.743589743589745e-08,
      "loss": 0.0014,
      "step": 3881
    },
    {
      "epoch": 49.61022364217252,
      "grad_norm": 0.1619696170091629,
      "learning_rate": 9.230769230769232e-08,
      "loss": 0.0019,
      "step": 3882
    },
    {
      "epoch": 49.62300319488818,
      "grad_norm": 0.24179552495479584,
      "learning_rate": 8.717948717948718e-08,
      "loss": 0.0023,
      "step": 3883
    },
    {
      "epoch": 49.635782747603834,
      "grad_norm": 0.1684618890285492,
      "learning_rate": 8.205128205128206e-08,
      "loss": 0.0016,
      "step": 3884
    },
    {
      "epoch": 49.64856230031949,
      "grad_norm": 2.7814879417419434,
      "learning_rate": 7.692307692307694e-08,
      "loss": 0.0107,
      "step": 3885
    },
    {
      "epoch": 49.66134185303515,
      "grad_norm": 0.23524853587150574,
      "learning_rate": 7.17948717948718e-08,
      "loss": 0.0017,
      "step": 3886
    },
    {
      "epoch": 49.674121405750796,
      "grad_norm": 0.1791078746318817,
      "learning_rate": 6.666666666666668e-08,
      "loss": 0.0017,
      "step": 3887
    },
    {
      "epoch": 49.68690095846645,
      "grad_norm": 1.7943719625473022,
      "learning_rate": 6.153846153846154e-08,
      "loss": 0.0111,
      "step": 3888
    },
    {
      "epoch": 49.69968051118211,
      "grad_norm": 0.17512063682079315,
      "learning_rate": 5.641025641025642e-08,
      "loss": 0.0015,
      "step": 3889
    },
    {
      "epoch": 49.712460063897765,
      "grad_norm": 0.12796635925769806,
      "learning_rate": 5.1282051282051286e-08,
      "loss": 0.0014,
      "step": 3890
    },
    {
      "epoch": 49.72523961661342,
      "grad_norm": 1.830198884010315,
      "learning_rate": 4.615384615384616e-08,
      "loss": 0.005,
      "step": 3891
    },
    {
      "epoch": 49.73801916932907,
      "grad_norm": 0.974043607711792,
      "learning_rate": 4.102564102564103e-08,
      "loss": 0.0049,
      "step": 3892
    },
    {
      "epoch": 49.75079872204473,
      "grad_norm": 0.9838169813156128,
      "learning_rate": 3.58974358974359e-08,
      "loss": 0.004,
      "step": 3893
    },
    {
      "epoch": 49.763578274760384,
      "grad_norm": 0.9195809960365295,
      "learning_rate": 3.076923076923077e-08,
      "loss": 0.003,
      "step": 3894
    },
    {
      "epoch": 49.77635782747604,
      "grad_norm": 1.8277593851089478,
      "learning_rate": 2.5641025641025643e-08,
      "loss": 0.0087,
      "step": 3895
    },
    {
      "epoch": 49.78913738019169,
      "grad_norm": 0.1428929716348648,
      "learning_rate": 2.0512820512820516e-08,
      "loss": 0.0011,
      "step": 3896
    },
    {
      "epoch": 49.801916932907346,
      "grad_norm": 1.8820996284484863,
      "learning_rate": 1.5384615384615385e-08,
      "loss": 0.0072,
      "step": 3897
    },
    {
      "epoch": 49.814696485623,
      "grad_norm": 1.2029500007629395,
      "learning_rate": 1.0256410256410258e-08,
      "loss": 0.0034,
      "step": 3898
    },
    {
      "epoch": 49.82747603833866,
      "grad_norm": 1.5105140209197998,
      "learning_rate": 5.128205128205129e-09,
      "loss": 0.005,
      "step": 3899
    },
    {
      "epoch": 49.840255591054316,
      "grad_norm": 0.34791117906570435,
      "learning_rate": 0.0,
      "loss": 0.0019,
      "step": 3900
    },
    {
      "epoch": 49.840255591054316,
      "eval_loss": 1.0049241781234741,
      "eval_runtime": 183.656,
      "eval_samples_per_second": 0.855,
      "eval_steps_per_second": 0.109,
      "step": 3900
    }
  ],
  "logging_steps": 1,
  "max_steps": 3900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 100,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 50,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.915877583275712e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}